² Farstad, A. L. and Hodges, Z. C., The Greek New Testament according to the Majority Text (2nd ed.; Nashville: Thomas Nelson, 1985).Google Scholar

³ Wallace, D. B., ‘The Majority Text: A New Collating Base?’, NTS 35 (1989) 609–18.CrossRefGoogle Scholar I wish to thank D. B. Wallace for forwarding a draft of his article to me. Similarly see Zuntz, G., ‘The Byzantine Text and New Testament Textual Criticism’, JTS 43 (1942) 25–30.CrossRefGoogle Scholar

⁴ Aland, Kurt and Aland, Barbara, The Text of the New Testament. An Introduction to the Critical Editions and the Theory and Practice of Modern Textual Criticism (trans. Rhodes, E. F.: Grand Rapids: Eerdmans, 1987) 106–8, 111, 116Google Scholar. Terminology used in the following table reflects their general text-critical method. Therefore, ‘Byzantine’ corresponds to either the Textus Receptus (as the usual collating base) or the Majority Text ($$$) in N²⁶, and ‘original’ is the reading attested by the Alexandrian witnesses.

⁵ Fee, Gordon, ‘The Textual Criticism of the New Testament’, Biblical Criticism: Historical, Literary and Textual (Grand Rapids: Zondervan, 1978) 136Google Scholar, has named codex A in the Gospels as the earliest manuscript to attest the Byzantine text (c. 475 CE). However, Sanders, H. A. in his original collation of the manuscript (The New Testament Manuscripts in the Freer Collection, Part 1. The Washington Manuscript of the Four Gospels [New York: MacMillan, 1918] 139)Google Scholar argued that W (032) should be dated to the late fourth and possibly early fifth century. Freerianus/Washingtoniensis (W) is often relegated to an inferior status because it is a ‘mixed’ text with sections reflecting Late Alexandrian (Luke 1.1–8.12), Caesarean (Mark 5.31–16.20), and Byzantine (Luke 8.13–24.53) text-types. Although W is often disqualified from consideration by many modern New Testament text critics, Sanders' early dating makes its Byzantine portions the earliest extant Byzantine text and thus our most significant witness in the study of the Byzantine text. The thesis of Sturtz, H. A., The Byzantine Text-Type in New Testament Textual Criticism (Nashville: Thomas Nelson, 1984)Google Scholar that the Byzantine text is extant among the early papyri is without foundation. While there is evidénce of Byzantine readings in the papyri, no single papyrus has yet demonstrated Byzantine readings in such a consistent manner that it could be said to constitute a Byzantine texttype manuscript. (See Fee, G., JETS 28 [1985] 239–42.Google Scholar)

⁶ The low results for P may be due to its fragmentary nature. Since the Alands offer no procedural notes on their collation, the precise reason for P's low standing is not clear.

⁷ Consider the difficulties in using the two most widely available and respected texts in the field. The apparatus of Tischendorf is incomplete by modern standards, and that of von Soden contains numerous errors and generalities.

⁸ The New Testament in Greek: The Gospel According to St. Luke, Parts 1 and 2. Ed. by the American and British Committees of the International Greek New Testament Project (Oxford: Clarendon, 1984, 1987). This work utilizes the Claremont Profile Method. While a secondary source for such a study, its acclaimed accuracy in the Gospel of Luke suggests its great value as a tool for such preliminary investigations.

⁹ Sanders, New Testament Manuscripts.

¹⁰ Sanders, , New Testament Manuscripts, 145–217.Google Scholar

¹¹ Wisse, F., The Profile Method for the Classification and Evaluation of Manuscript Evidence as Applied to the Continuous Greek Text of the Gospel of Luke (Grand Rapids: Eerdmans, 1982) 37–8Google Scholar. He argues as follows: ‘In view of the special character of the TR, the question arises whether the collation base would influence or invalidate the results of the Profile Method. But the problem concerning the TR in the past was the tendency to attach a positive value judgment to every departure from it. If a collation base is left completely neutral, it will be no more than a handy tool to compare the text of one MS with that of others. Any real or artificial text could fulfil this purpose. The TR happens to have been used for this purpose for more than half a century, and has the distinct advantage of being close enough to the Byzantine groups to lighten the task of collation and lessen the chance of error with respect to the bulk of minuscules … On the other hand, one would not want the collation base to be identical to one of the Byzantine groups, for agreements against the collation base lend themselves more easily for classification than agreements with the base.’

¹² Sanders collated W against the Oxford 1880 TR (New Testament Manuscripts, 143). Hodges and Farstad employed the Oxford 1825 edition citing all differences between the TR and their text in the first apparatus on each page (Majority Text, xiii, xviii). The IGNT committee employed a similar base for their work, the Oxford 1873 reprint of the 1828 edition (‘Luke’ 3.1.vi-viii). While the Oxford editions of 1825 and 1828 (and thus 1873) are considered identical, the 1880 edition is often not. After the revision of all three editions of the TR for the passages under scrutiny in this study, only three variations were discovered: γεννήματος (1825) for γενήματος (1880) in Luke 22.18; βαλαντίον (1825) for βλλαντίον (1880) in Luke 22.35; and, βαλαντίου (1825) for βαλλαντίου (1880) in Luke 22.36. Thus the assumption of no significant effect on the results was confirmed.

¹³ As an example of the effect of this collation base change, consider the result of the transformations of Sanders' data for W. Where W agreed with the Majority Text against the TR, the variant was removed from consideration in Sanders' list. Similarly where the Majority Text disagreed with TR but no variant was indicated for W, it was assumed that W read with the TR and W's reading was added to the list as a variant from the MT. Consequently W's readings were corrected toward the Majority Text (away from the TR) 90 times and corrected away from the Majority Text (toward the TR) 31 times. This indicates that a collation base change could significantly change collation statistics.

¹⁴ A Caesarean representative (Θ) was included in light of Hurtado's, L. work, Text-Critical Methodology and the Pre-Caesarean Text, Codex W in the Gospel of Mark (Grand Rapids: Eerdmans, 1981)Google Scholar which shows that the portions of W previously labelled ‘Caesarean’ (or more properly ‘pre-Caesarean’) may actually represent a ‘Proto-Byzantine’ text. The Western text family was excluded from this study since it is often thought that W's affinities in Luke lie with the Proto-Alexandrian rather than the Western text.

¹⁵ Other ‘good’ representatives of the Late Alexandrian period, such as T (029) or 33, were either defective in passages where the other chosen witnesses were extant or were too fragmentary to be of significant use for this study.

¹⁶ Streeter, B. H., The Four Gospels: A Study of Origins Treating of Manuscript Tradition, Sources, Authorship and Date (rev. ed.; New York: MacMillan, 1927) 147Google Scholar note 1, offers his suggestion for the choice of representative uncials in such a collation: ‘… the Byzantine text should be determined by some objective standard, such as the agreement of two out of the three manuscripts S V W, or perhaps better, E S V’ (where S refers to the uncial 028, Vaticanus). Unfortunately this manuscript (S) is defective in the necessary Lukan passages and so could not be used in this study.

¹⁷ Colwell, E. C., ‘Method in Establishing Quantitative Relationships between Text-Types of New Testament Manuscripts’, Studies in Methodology in Textual Criticism of the New Testament (Grand Rapids: Eerdmans, 1969) 56–62.Google Scholar However, relative statistical affinity was given precedence over Colwell's minimum postulate of 70% affinity for a texttype relationship. The general validity of holding fast to Colwell's 70% agreement rule is dependent upon the number of manuscripts and the number and kind of variants consulted, a consideration which is factored into this study. Firsthand collation demonstrates that the inclusion of singular readings (particularly the inclusion of itacisms) raises the general agreement of all manuscripts. However, as one can note from this study, the issue is not so much the ‘magic number’ of 70% as the obvious numerical gap in the statistical relationships when the cumulative results are compared (compare Table 3). See the studies by Richards, W. L., The Classification of the Greek Manuscripts of the Johannine Epistles (SBLDS 35; Missoula: Scholars, 1977)Google Scholar; ‘A Critique of a New Testament Text-Critical Methodology -The Claremont Profile Method’, JBL 96 (1977) 555–66Google Scholar; ‘Manuscript Groups in Luke 10 by Quantitative Analysis’, JBL 98 (1979) 379–91.Google Scholar

¹⁸ Although an excellent tool, IGNT does not include every variant reading in every manuscript. A sample comparison of IGNT with facsimiles of some of the major uncials suggests that the editors have already eliminated some of these ‘insignificant’ readings.

¹⁹ Colwell, , Studies, 57–9.Google Scholar

²⁰ See footnote 14.

²¹ This method is similar to the ‘Comprehensive Profile Method’ (CPM). Here, however, the procedure was more detailed. The analytic principles were applied, not to the group characteristics, but to the individual manuscripts. For a larger discussion of CPM see the excellent essays by Ehrman, B., ‘Methodological Developments in the Analysis and Classification of New Testament Documentary Evidence’, NovT (1987) 22–45Google Scholar; ‘The Use of Group Profiles for the Classification of New Testament Documentary Evidence’, JBL 106 (1987) 465–86.Google Scholar

²² This listing sets out distinctive reading alignments within and across groups in descending frequency of readings to distinguish the more common distinctive alignments, and suggest genealogical relationships in the history of transmission. The percentage consanguinity figure can show how these alignments represent their statistically established relationship, for example: high consanguinity and many distinctive readings support strong genealogical relationship; high consanguinity and low distinctive readings generally support a homogeneous tradition.

²³ The issue is not whether there was a single Alexandrian archetypal manuscript or a family of manuscripts, but whether the data support an expanding tradition which suggests a more defined form of that text at an earlier period. See Martini, C. M., ‘Is There a Late Alexandrian Text of the Gospels?’, NTS 24 (1977–1978) 285–96CrossRefGoogle Scholar, who argues against the notion of a singular archetype in favour of such a trajectory approach to texttypes.

²⁴ Although only one Late Alexandrian manuscript was included, the relatively stable statistical position of L with respect to the Majority Text congruent to the earlier Alexandrian witnesses suggested that it could be examined for deviation against the early witnesses in a way similar to the procedure followed for the two Byzantine subgroups.

²⁵ The manuscript dates on the graph are only approximate (by century) as provided by Aland, , Text, 106–16.Google Scholar

²⁶ Hodges, Z. C., ‘A Defense of the Majority Text’ (unpublished paper, Dallas Theological Seminary, n.d.) 4–9.Google Scholar

²⁷ Fee, G., ‘Modern Textual Criticism and the Revival of the Textus Receptus’, JETS 21 (1978) 26Google Scholar raises the significant, contrary analogy to support this thesis: later Vulgate manuscripts appear to bear a closer relationship to each other than to Jerome's original.

²⁸ Another factor camouflaging the existence of the Caesarean text is that the TR attests Caesarean distinctive readings. Streeter's conclusion (The Four Gospels, 105) affirms this problem in collations against the TR. This also argues that the Majority Text ought to be used as the collating base in studies of the Caesarean text.

²⁹ Eusebius, Life of Constantine 4.36 as cited by Metzger, B., The Text of the New Testament (2nd ed.; Oxford: Clarendon, 1968) 7.Google Scholar

³⁰ Metzger, Text, 14; Aland, K., ‘The Text of the Church’, TrinJ 8 NS (1987) 131–44Google Scholar, argues for a similar evolutionary process on the basis of patristic evidence.

³¹ The early Byzantine group members departed significantly from the Majority Text 137 times. These readings were taken as a baseline and (unlike the more general Quantitative Analysis) the reading entered the final calculation only when a member of this group departed from the Majority Text. The result is a measure of statistical affinity of the early Byzantine group members where each departs from the Majority Text text in a non-accidental way. This reveals the degree to which each member of the group emerged from the Alexandrian stream into the Byzantine trajectory.

³² ‘The nineteenth century was the age of the uncials; the mid-twentieth century was the age of the papyri – this marked a striking advance over the nineteenth century. But now we are entering the age of the minuscules.’ Aland, Kurt, ‘Introduction’, Novum Testamentum Graece (26th ed.; Stuttgart: Deutsche Bibelstiftung, 1979) 47.Google Scholar His statement did not mean that scholarship is finished with the uncials, or that the minuscules ought to be accorded a greater weight in the decisions of the text. Rather, having generally classified the uncials and papyri, it is now time for New Testament critical scholarship to study the great number of unexplored minuscules as a prelude to an understanding of the history and transmission of the New Testament text during the period of the Middle Ages.

³³ Doyle, A. C., ‘The Adventure of the Abbey Grange’, The Strand Magazine 28.165 (09 1904)Google Scholar; reprinted, The Complete Original Illustrated Sherlock Holmes (Secaucus, N.J.: Castle Books, 1976) 607Google Scholar, with my compliments to Art Farstad for his love of ‘things Hol-merian’.