Critical Editions of the New Testament

Contents: Introduction* Aland: Synopsis Quattuor Evangeliorum* Bover* Hodges & Farstad* Huck* IGNTP (International Greek New Testament Project)* Legg* Merk* The "Nestle" text: Nestle editions 1-25 | Nestle-Aland editions 26, 27 | Nestle-Aland edition 28* Das Neue Testament auf Papyrus* Souter* Swanson* Tasker* Tischendorf* United Bible Societies Edition* Vogels* Westcott & Hort
Summary: A Comparison of the Various Editions
Appendix: The Variorum Bible
Appendix: Latin Editions
Appendix: Editions of the LXX

Introduction

Karl Lachmann (1793-1851) broke with the Textus Receptusin 1831. This, then, was the first "critical edition" of theNew Testament -- an edition compiled using specific rules based on thereadings of a significant selection of important manuscripts. Since then,many others have appeared. Some of these (Lachmann'sown, and that of his younger contemporary Tregelles) are now almost completelyobscure. Others -- notably those of Westcott and Hort and the United BibleSocieties -- have exercised great influence.

Ideally, a critical edition will include an apparatus supplyinginformation about how the readings were decided upon. There are, however,critical editions (e.g. that of Westcott & Hort)which do not include such information. The list below describesmost of the major editions since Tischendorf's vital eighth edition.

Aland: Synopsis Quattuor Evangeliorum

Editor. Text and apparatus edited by Kurt Aland.

Date of Publication. The first edition appeared in 1963. A revised edition,listed as the fourth, appeared in 1967; another revised edition, the ninth, cameout in 1976. The final major revision, the thirteenth, was published in 1985.The first three major editions (officially listed as the first through twelfth) use the samebasic arrangement of the text; the revisions took place primarily in the apparatus.The thirteenth edition entirely recast the work; a new text was adopted and a newapparatus created. The structure of the synopsis was unchanged, but otherwise itwas an entirely new publication.

The Text. The text of the first twelve editions is essentially that ofthe early Nestle-Aland editions. With the thirteenth edition,the text was adjusted to match that of the Nestle-Aland26th edition.

The Aland Synopsis is one of the more substantial now available. Allfour gospels are presented in full, and there is a complete text of the Gospelof Thomas (in Latin, English, and German; neither Coptic nor Greek texts areoffered!). The critical apparatus is also more than usually complete; an apparatusis usually supplied wherever a passage is cited, not just at its "main"appearance. In addition, the apparatus gives a fairly full list of variants --many more than are found in the equivalent editions of the Nestle-Aland text,and not limited simply to harmonization variants. While SQE will not allowthe student to completely reconstruct the cited manuscripts (especially theminuscules), it includes enough data to allow a valid comparison of the varioustext-types. (This cannot be said of NA²⁷!)

The Apparatus For compactness, SQE uses the same set of critical symbols as theNestle text (for details, see the picture in that article).
Unfortunately, the apparatus does have its drawbacks. (We are nowreferring specifically to the recent editions, from the thirteenth on.) For one thing, ithas a high number of errors (most of them seemingly errors of the press; theseare slowly being corrected). The selection of witnessesis also questionable. The Byzantine text of the uncial era, for instance,is represented by four manuscripts, E F G H. All of these, it should be noted,belong to the K^x recension. Thus, although there are moreByzantine witnesses than in the Nestle-Aland edition (which offers only Kand Γ), they offerless diversity (of the witnesses in Nestle-Aland, K is a member of FamilyΠ, while Γ is K^x). The new minusculesare also an odd lot. Why would anyone make 1006 (purely Byzantine) an explicitlycited witness, while omitting 1241 (arguably the most Alexandrian minusculeof Luke)? As a final note, we should observe that while SQE cites manymember of Family 1 (1 and 209, as well as 205, 1582, 2542 not cited explicitlyas members of the family) and Family 13 (13, 69, 346, 543, 788, 983; note thatthe best family witness, 826, is omitted), it cites them in such a way thatthe readings of the individual manuscripts can only be determined when themanuscript is cited explicitly (that is, if -- say -- 346 is not cited explicitlyon either side of a reading, it may agree either with f¹³ or ℳ.
To sum up, SQE is a good synopsis with a useful critical apparatus,but one should take care not to rely upon it too heavily (due both to itsinaccuracies and its slightly biased presentation of the evidence).

Bover

Editor. Text and apparatus edited by José Maria Bover, S.J.

Date of Publication. The first edition, Novi TestamentiBiblia Graeca et Latina appeared in 1943. The first four editions(1943-1959) are essentially identical; the fifth edition of 1977 andfollowing (revised by José O'Callaghan Martínez) isslightly different, but primarily in the area of the parallel texts.

The Text. The Latin text of Bover, until the fifth edition,is simply the Clementine Vulgate (in the fifth edition the Neo-Vulgatewas substituted and a Spanish version added). Thus the Latin text hasno critical value.
The Greek text is somewhat more reputable. It is a fairly typicalTwentieth Century product, compiled eclectically but with a clearpreference for Alexandrian readings (though not as strong a preferenceas is found in the Westcott & Hort andUnited Bible Societies editions). Ithas been esteemed by some for its balanced critical attitudes;others might view it as having no clear guiding principle.

The Apparatus. In its later editions, Bover's Latin text has no apparatus at all (from the critic's standpoint, there is really noreason for it to be there; there was a very brief apparatus in the first edition, but too short to be of much use), and even the Greek apparatus is limited. Bover's manuscriptdata, like that of Merk, comes almostentirely from von Soden. Like Merk, Bover cites a fewmanuscripts discovered since von Soden's time (papyri up toP⁵²,including the Beatty papyri; uncials up to 0207; a few of the minuscules upto 2430, plus a modest handful of lectionaries).
In construction Bover's apparatus strongly resemblesMerk's, using essentially the same manuscriptgroupings and much the same set of symbols. (For an example,see the entry on Merk). Bover does mark places where he is unsure of the original text by placing a * by marginal readings which he thinks as good as, or almost as good as, the reading in the text. The most significantdifference between Bover and Merk in their presentation of the data, however, isthat Bover also lists the readings of the various editions --T=Tischendorf, S=von Soden, V=Vogels, L=Lagrange (Gospels, Romans,Galatians only), M=Merk, H=Westcott & Hort(h=Hort's margin; (H)=Hort's text against the margin); W=Weiss;J=Jacquier (Acts only), C=Clark (Acts only), A=Allo (1 Cor.,Rev. only).
These critical editions also define the apparatus; Boveronly offers manuscript information at points where the critical editionsdisagree. His apparatus is thus much more limited than thatof Merk or even Nestle.It also shares the defects one would expect from a workbased on von Soden: Many of the collations are inaccurateor imperfectly reported (for details, see the entryon Merk), although Bover claims to have tried to check the readings against Tischendorf and other sources (he said he sometimes worked from the manuscripts themselves, but I didn't see much evidence of this). Bover's transcription of von Soden's symbolsis somewhat more careful (and often more explicit)than Merk's, and is therefore perhaps slightly more reliable.It is, however, less full even for the readings it contains --citing, e.g., fewer fathers (the introduction does noteven list the fathers cited!) and fewer versions. And Bover hasrecast Von Soden's groupings a bit -- instead of having fivesets of witnesses (for Gospels, Acts, Paul, Catholics, Apocalypse),he uses the same groupings for Acts, Paul, and Catholics. This isreasonable in one sense -- the groupings for the three are fairlysimilar -- but it makes it harder to use the apparatus, as one isalways having to look up exceptions (e.g. 1739 files with H inPaul, but I in the other two). Also, a warning for those with oldereyes: The typeface (at least in some editions) is rather unsuitablefor the purpose; the symbols | and ] -- keys to understanding theapparatus -- are almost indistinguishable.

Hodges & Farstad

Editors. Zane C. Hodges and Arthur L. Farstad

Date of Publication. The first edition, The GreekNew Testament According to the Majority Text, appeared in 1982.A slightly revised second edition appeared in 1985.

The Text. Unlike most critical editions, that ofHodges and Farstad does not attempt to reconstruct the originaltext on the basis primarily of the earliest manuscripts. Rather,it assumes that the Byzantine Majority text is the original text,and reconstructs this text. For the most part, this is done by"counting noses" -- looking for the reading which hasthe highest number of supporters (which in the gospels oftenbecomes a matter of printing the reading of K^x).In the Apocalypse and the storyof the Adulteress, however, H & F resort in a limited wayto stemmatics, meaning that they print a few readings which,although well-supported, are not the majority reading.
It should be noted that Hodges and Farstad did not assembletheir text based on manuscript collations; rather, for the mostpart they simply followed Von Soden's K text and itssubgroups (which, in their edition, is denoted ℳ whenentirely unified and M when a portion of the type defects).Thus the edition may not always represent the actual majoritytext. Even so, H & F is the only edition of the Byzantinetext-form to have an apparatus of any sort. This makes ituseful to anyone who wishes to examine the strength anddepth of the Byzantine tradition. (The critic does nothave to subscribe to the editors' theories to find the editionuseful.) The edition also serves as auseful demonstration that the Byzantine text-type, althoughmore united than any other known type, is not the monolithicentity its opponents sometimes make it out to be.

The Apparatus. The H & F text has two apparatus.The first, and more important for the editors' purposes, isthe apparatus of variants within the Byzantine tradition.Here the editors list places where the Byzantine traditiondivides, even noting some of the strands identified byVon Soden (e.g. H & F's M^r isvon Soden's K^r; their M^c is vonSoden's K^c, etc.) They also note thevariant readings of the Textus Receptus (demonstrating, incidentally,that the TR is a poor representative of the Byzantine type). Thisfirst apparatus, which contains relatively few readings, has itsvariants marked in the text with numbers and has lemmata in themargin.
The second apparatus lists variants between the H & F textand the United Bible Societies edition. A quicksample indicates that these are roughly three times as common asvariations within the Byzantine tradition. For these variantsthe editors use the same symbols as the recent editions of theNestle-Aland text.
A handful of witnesses -- Vaticanus, Sinaiticus, Alexandrinus,Ephraemi Rescriptus, and certain papyri -- are noted in bothapparatus, but their readings are noted only for variants includedfor other reasons. The H & F apparatus gives far less informationabout these manuscripts than even the Nestle apparatus, and cannotbe used for textual classification of any specific witness.
Although the apparatus of H & F is very limited, it servesa useful purpose even to those who do not believe in Byzantinepriority. It is the only available tool (other than von Soden'scryptic edition) for determining if a reading is theByzantine reading, a Byzantine reading in cases wherethat text divides, or entirely non-Byzantine. This can beimportant when dealing with mixed manuscripts. Also, H & Fincludes some variants not covered in NA²⁷.

Huck

The name "Huck," like the name Nestle,is actually a term for a constellation of editions (in this case, ofa gospel synopsis rather than a critical edition), with various editorsover the years. The two, in fact, are almost of an age. Albert Huckpublished his first synopsis in 1892, but this was designed for aparticular class and synoptic theory; the third edition of 1906 wasthe first for general use. With the ninth edition of 1936, the bookpassed from the hands of Albert Huck to H. Lietzmann and H. G. Opitz.At this time the text was revised (Huck's own editions were based onTischendorf's text; Lietzmann used a text approximating that of Nestle).The 1981 edition was taken over by H. Greeven, and the arrangement ofpericopes significantly altered. Greeven also altered the text, usinghis own reconstruction rathr than any previous edition.

Editors. Albert Huck; later taken over by H. Lietzmann, H. G. Opitz, H. Greeven

Date of Publication. The first edition was published in 1892;a revised third edition came out in 1906, another revision constitutedthe fourth edition of 1910. The revised ninth edition of Lietzmann-Opitzwas published in 1936. Greeven's thirteenth edition appeared in 1981.

The Text. Prior to the appearance of Greeven's edition, Huckcould not really be considered in any way a critical edition. Huckused Tischendorf's text, Lietzmann a modification of Nestle's. Neithereditor provided a full-fledged critical apparatus. (Lietzmann admitted tohaving a "limited" apparatus. Not only was the number ofvariants limited, but fewer than a dozen Greek witnesses were cited,and the data on the versions was much simplified.) The value of Huck,at that time, lay in the arrangement of the parallel gospels (Matthew,Mark, and Luke; John was not included). This, obviously, was sufficientto keep the book in print for nearly a century, but the editions havelittle value to the textual critic. For this reason, the remainderof this discussion will be devoted to Huck-Greeven, which simultaneouslyprovided a new text (edited by Greeven), a much fuller apparatus(also by Greeven), and a modification of the synopsis itself, includingmore parallels as well as some portions of the gospel of John.
The text of the Greeven revision is somewhat problematic. Greevenclaims that it averages about nine variations per chapter from theUBS/Nestle text. This would be about typical for a modern edition --if anything, it's at the low end of the scale. The problem is, Greevengives not a hint of his critical principles. Nor does Greeven give usa list of differences from UBS. Thus it is almost impossible toreconstruct his method. This makes it difficult to know how far torely upon his text. My impression, in compiling its readings for thelist of Most Uncertain Readings,is that, in those readings at least, it inclines very stronglytoward the Byzantine text; the result is probably about like vonSoden in its "feel," though the rate of actual agreementsmay not be excessively high.
The apparatus is as peculiar as the text. In no sense is itcomplete; the focus in upon parallels, almost to the exclusion ofother variants. It is at first glance an easy apparatus to read; eachreading begins with the lemma, followed by its supporters if theyare relatively few, then a square bracket ] followed by the alternatereadings and their support; different variation units are separated bylarge spaces and bold vertical lines. Deciphering the list ofwitnesses is a much different matter. Witnesses are grouped by type(though Greeven denies that his groups have any actual meaning), andcited by group symbols (e.g. λ φare the Lake and Ferrar groups), and are cited in group order. However,Greeven does not list the order of the witnesses outside the fourgroups (Alexandrian, Lake, Ferrar, Soden). Nor are the contents ofthe various fragments listed explicitly. Thus it is almost impossibleto be certain which manuscripts are actually cited within thenotation Rpl (referring to all uncited uncials and thelarge majority of minuscules). It is best to trust the apparatusonly where it cites a witness explicitly. And even there, it appearsthat many of the citations are from von Soden.
The citation of the versions, as opposed to the citing of theGreek witnesses, is excellent. All Old Latin witnesses are citedby name, with lacunae indicated. Where the Harklean Syriacattests to multiple readings, Greeven shows the nature of each variant.Where the manuscripts of the various Coptic versions do not show a consensus,Greeven indicates the number on each side of the reading. Unfortunately,the Armenian and Georgian versions are not handled with anything likethe same precision, but this is no reason to condemn the edition; mostothers treat these versions with equal disdain.
The list of Fathers cited is quite full and unusually detailed, listingboth the language and the date of the author, and including at least ahandful of Syriac, Coptic, and even Arabic sources as well as the Greekand Latin Fathers. A wide variety of Harmonies are also cited (under asymbol which implies they are versions of the Diatessaron, though this isnot stated). The introduction gives a good concise description of theseharmonies.
Great care must be taken to understand Greeven's apparatus, which isstrongly dependent not only on the order of the witnesses, buton the typographic form in which they are presented (e.g. Or does not mean thesame thing as Or, even though both refer to Origen).
To sum up, the apparatus of Greeven is very difficult, though it offersa wide variety of useful information, and does not list all the variantsone would "expect" to find. Students are therefore advisednot to rely solely upon it, but to use at least one other source -- bothto get a full list of variants in a particular gospel and to check one'sinterpretation of the apparatus for the variants it does contain. Greevencan give a sense of the support for a reading. It cannot and does notgive specifics capable of being transferred to another apparatus.

IGNTP: The International Greek New Testament Project

Editor. The IGNTP is the successor to Legg, but there is no general editor. The project is overseen by a committee, and even the committee chairs have changed over the years. Contributors to the volumes are listed; for the Lucan volume, if I counted correctly, there are 266! (It is an open question whether this extremely large number of contributors actually improved the volume....) I find it noteworthy that, of the five editors of the United Bible Societies Edition, only Bruce M. Metzger chose to participate.

Date of Publication. An ongoing series. The first volume, Luke 1-12, was published in 1984, with Luke 13-24 following three years later. This edition included all the information -- Greek witnesses, versions, fathers, etc. -- in one place. This plan has now been abandoned, with the volume on John being published in parts -- so far, the papyri and uncials have been done. Thus there is no single volume containing all the witnesses for John. Paul is now underway; the expectation is that the whole project will take about twenty years, following the same format.

The Text. Since Luke, and only Luke, will be published in a single place, the rest of this discussion refers specifically to IGNTP Luke. The text of the Luke volume is the 1873 edition of the Textus Receptus, without intentional alteration. This was deliberately chosen as a text that no one would claim had any critical authority. (This was a major change from Legg, which worked from the best critical edition available, Westcott & Hort's.) The reason for using the Textus Receptus was that it was close to the Byzantine text, so the apparatus would be shorter (since there wouldn't have to be endless lists of Byantine manuscripts which disagreed with the critical text) without either having to determine "the" majority text or (I add somewhat sarcastically) being so close to the Majority text as to let MT advocates nod off and just accept what was printed.

The Apparatus The apparatus is of course the whole point of the IGNTP edition, and it is certainly extensive. For the Lucan volume, I count eight papyri, 62 uncials (although several of these are known to come from the same original copy, a point which the IGNTP does not deign to notice), 128 minuscules (if I counted correctly), 41 lectionaries, 19 individual Old Latin manuscripts, and fifteen other versions (of which only the Georgian is broken out by individual manuscripts).

The apparatus has three parts. First is a list of witnesses, then a list of patristic sources, then the real apparatus, showing the readings of the various manuscripts.

I'm tempted to call these the bad apparatus, the OK apparatus, and the good apparatus, in that order.

The first apparatus: witnesses. The problem with the witness list is the sheer bulk of the IGNTP citation base. With some 250 items cited, obviously the reader needs to know what manuscripts do and don't exist for a particular verse. The list of manuscripts at the beginning does not list this; there is no way to know, for instance, that 𝔓⁷ has parts of only three verses but that 𝔓⁴⁵ has parts of all four Gospels and Acts. There is a card listing all the witnesses cited, but that's all it does: list the witnesses. No indication of contents. The only way to learn which manuscripts exist where is to consult the list of defective witnesses, verse by verse! Want to know if, say, minuscule 1079 is complete or fragmentary? Good luck.

Even if you're looking at the witnesses to a particular verse, rather than checking on a particular manuscript, there are difficulties. The problem is that there are two lists of defects for each verse. One, at the start of each chapter, lists witnesses which are defective for the whole chapter. The other, the apparatus associated with each verse, lists the manuscripts which exist for at least part of that chapter but don't exist, or are incomplete for that verse.

Take Luke 1:9 as an example. The list at the beginning of Luke 1 says that the following witnesses lack all of Luke 1: 𝔓³ 𝔓⁷ 𝔓⁴⁵ 𝔓⁶⁹ 𝔓⁷⁵ 𝔓⁸² N Q T 063 070 0102 0108 0113 0115 0117 0124 0139 0147 0171 0181 0182 0190 0191 0196 0202 0239 0250 0253 0265 0266 0267 669 vt^a2 vtⁱ vt ^λ vt^s sy^cur.

Turning to verse 9 in particular, we learn that 𝔓⁴ 𝔓⁴² G H 0130 0135 0177 174 2322 2399 vt^β Eth^{Bodl 41} lack the verse entirely, that F is missing about the first half, and that vt^a is fragmentary.

Now, first, this omits to mention that 070 0124 0190 0191 0202 are all part of the same manuscript (the one the Alands call 070), so the above list makes it seem as if there are more manuscripts of this book floating around than really exist. But the real difficulty is that the full list of witnesses absent for 1:9 is
𝔓³ 𝔓⁴ 𝔓⁷ 𝔓⁴² 𝔓⁴⁵ 𝔓⁶⁹ 𝔓⁷⁵ 𝔓⁸² (F partial) G H N Q T 063 070+0124+0190+0191+0202 0102 0108 0113 0115 0117 0130 0135 0139 0147 0171 0177 0181 0182 0196 0239 0250 0253 0265 0266 0267 174 669 2322 2399 (vt^a fragmentary) vt^a2 vt^β vtⁱ vt ^λ vt^s sy^cur Eth^{Bodl 41}.

But there is no way to learn this without constantly flipping back and forth. And it really wouldn't have taken much more room to list all the manuscripts that were defective for each verse!

The second apparatus: fathers. This apparatus also suffers from an information problem. There is no way to learn, from the IGNTP volume itself, the sigla used! Readers are instead referred to Lampe's Patristic Greek Lexicon, a book that is far too rare and expensive for people to have sitting around on their own bookshelves,and H. J. Frede, Kirchenshriftsteller: Vereichnis und Sigel, plus some lesser works Fortunately, if you can figure out who is being cited and in what book and edition, the readings of the text are fairly clear. If you can't figure it out, well, at least a subset of the information is included in the third apparatus as well:

The third apparatus: Greek manuscripts and versions. This apparatus finally gets it right. It's straightforward: The lemma (or at least the first and last words) is quoted, then a colon, then the alternate readings and their support. Let's take Luke 1:60 as an example, since its apparatus is relatively sort:

TEXT: και αποκριθεισα η μητηρ αυτου ειπεν, Ουχι, αλλα κληθησεται Ιωαννης.

APPARATUS:
και: om. Gg (I)       αποκριθεισα: + [αυτοις] Gg       αυτου: om. 713       ειπεν: ελεγεν l70^c; + [αυτοις] Sp Dta Dtp Gg (II.III)       ουχι αλλα: om. Chrys       κληθηται ιωαννης: 2, 1 l253 Lvt (c) Et Chrys; [ιωαννης εστι το ονομα αυτου] Am Lc 2, 31 HI Lc       κληθησεται + το ονομα αυτου C* D 213 l211 Lvt (d) Sj (3 mss.) Cb (4 mss.) OS       ιωαννης: ιωανης D 0211 579

So that's eight variants in this verse. (By comparison, NA²⁸ has just one). To interpret these:

• The Georgian version [Gg] omits και
• The Georgian version [Gg] adds an object (αυτοις or some equivalent) after αποκριθεισα
• 713 omits αυτου
• for ειπεν 𝓁^70c has ελεγεν; several versions (Peshitta Syriac [Sp], Arabic and Persian Diatessarons [Dta Dtp], and geo^2,3 [Gg (II.III)]) add an object to the verb.
• Chrysostom omits ουχι αλλα
• for κληθηται ιωαννης 𝓁²⁵³, c of the Old Latin, the Ethiopic version, and Chyrsostom have ιωαννης κληθηται; some secondary sources have some form of ιωαννης εστι το ονομα αυτου
• after κληθησεται C* D 213 𝓁²¹¹ plus d of the Old Latin, three manuscripts of the Palestinian Syriac, four manuscripts of the Bohairic Coptic, and the Old Church Slavonic add το ονομα αυτου. (Note that this is the only reading cited in NA²⁸, and that it cites only C* D and the Bohairic Coptic)
• D 0211 579 spell the name ιωαννης as ιωανης.

There are several things worth noting here. First is how much the whole apparatus is dependent on formatting -- you need to observe the boldface, e.g., to know if something is a version. A second is that there is no marker to show where one unit of variation ends and another begins -- a space, yes, but no other symbol. When a unit spreads across multiple lines, or columns, it can be very hard to know just what you're reading, or where one variant ends and the next begins.

The other thing is the number of variants which are cites based solely on the versions or fathers. As a sample, I took the first 25 verses of Luke 4. In all, I counted 378 points of variation cited for those 25 verses (an average of 15 per verse, which shows how comprehensive the IGNTP apparatus is; that number may be slightly off, because I was skimming the hard-to-read apparatus, but it's close). But, of those 378, fully 115 were cited based solely on versions, or the fathers -- or even the Diatessaron. That's 30.4% of the total. And there were many other instances where there was a variation in the Greek witnesses, but one or more readings were cited which had no Greek support. There was only one verse in the sample (Luke 4:15) were all points of variation involved variation in the Greek. Given that most of the fathers, and several of the versions, were cited from non-critical editions (e.g. the Ethiopic), and others were clearly cited without knowing the genuine habits of the version (e.g. the Old Church Slavonic -- and probably the Armenian and Georgian), I am genuinely bothered by the quality of the versional information.

The bottom line, for me, is that IGNTP Luke is very valuable, but that everything in it needs to be carefully verified.

Legg

Editor. Apparatus edited by S. C. E. Legg, A. M.

Date of Publication. The only volumes to appear were those forMatthew and Mark, which were edited in the 1930s (the Markan volume, from 1935,is new enough to include P⁴⁵; the last uncial cited is 0188).

The Text. The text is that of Westcott & Hort, printed without deliberate modification.

The Apparatus. Legg's might well be called the last of the old-stylecritical apparatus, in the style of Wettstein or Tischendorf: it was compiledby one man, and all the text, including the introduction, is in Latin, notEnglish or German (even though Legg was a native speaker of English). Greekmanuscripts are referred to by Gregory numbers. All uncials are cited; inMark, e.g., this includes ℵ A B C D E F G H K L M N P S U V W X YΓ Δ Θ Π Σ Φ Ψ Ω 047 (the Byzantine uncials E F G H K M S U V Y Ω are grouped asל)and some 28 fragments. Minuscules cited include1 13 22 28 33 69 118 124 131 157 346 543 565 579 700 892 1071 1071 1582 (plusoccasional readings from others as collated by other editors such as Scrivener andvon Soden). The only lectionary cited consistently is the fragmentary l1353.The versions are cited in great detail although generally from printed editions (e.g.Horner for the Coptic, Blake for the Georgian).
Many fathers are cited, but many instances are in a separate apparatus at the bottomof the page. This apparatus is surprisingly short, but where it cites at all,it cites the full context, so that the reader can truly judge if the fatheragrees with the reading, rather than trusting the editor who cited the passage(a regular complaint against the Nestle editions, e.g.).
Apart from this peculiarity (which on the whole is more helpful than otherwise),the apparatus is easy to read as long as you can understand the Latin. The first twovariants in Mark, for instance, are as follows:
1. αρχη του ευαγγελιου: = nil nisi euangelium Sy.^hier. | Ιησου Χριστου sine add. ℵ*Θ 28 (om. χριστου*) 255. 1555* Sy.^hier. (=praem. domini) Geo.¹ Arm. ^{(9 MSS)}, item. cf. Iren. Orig. Epiph. Bas.^hr Tit. Serap. Victorin. Hieron., cf. etiam Scholia in Minuscs. 237. 238. 259: + υιου θεου ℵ^aBDLW; +υιου του θεου AΓΔΠΣΦל Minusc. rell., item +filii dei it. vg. (etiam praem. domini nostri vg. 3 MSS) Sy.^pesh.hl Cop.^sa.bo. Geo.² Arm. (ed. cum 3 MSS optim.) Aeth., item xf. Iren. Orig.^int Aug. Amb. Hier. al.; om. etiam Ιησου Χριστου Iren. semel Epiph. uid. infra.
In other words, from Legg's text Αρχη του ευαγγελιου Ιησου Χριστου, the Palestinian Syriac omits Αρχη and perhaps του. Ιησου Χριστου is the reading of ℵ*Θ 28^c 255 155*, plus the Palestinian Syriac, geo¹, and nine Armenian manuscripts. 28* omits Χριστου. To this ℵ^a B D L W add υιου θεου; E F H K M S U V Y Γ Δ Π Σ Φ Ω 1 13 22 33 69 118 124 131 157 346 543 565 579 700 892 1071 1071 1582 have υιου του θεου;the Old Latin and the Vulgate plus many other versions support one or the other of the latter two readings.
Legg has often been accused of being error-prone, and there is certainly some truth in this. But it is interesting to compare the second of these variants with the apparatus of SQE¹³, Swanson, and UBS⁴, the three editions which cite the most manuscripts.

In this case, at least, we see no disagreements between the various editors about how the manuscripts read. Of course, this is the first substantial variant, so errors would be less likely. But what is noteworthy here is that Legg's list of witnesses is superior to any of the editors who came after him. Swanson has the best list, without question, but he omits 1241 -- and, even worse, he omits 892, which is surely the most important minuscule of Mark!

SQE's list is peculiar indeed. It has most of the important minuscules, but omits 28(important in Mark if nowhere else) and 1241 (not as good in Mark as in Luke, but certainly not to be ignored). In exchange for these important omissions, we get 205 (which is so close to 209 that its inclusion makes no difference), 1006 (valuable in the Apocalypse but worthless in the gospels), 1506 (valuable in Paul but worthless in the gospels), and 2542 (at best, a weak member of family 1). And SQE has only a handful of Byzantine uncials -- and all of them K^x; there is no representative of Family Π or the M groups. Thus the Byzantine tradition is not adequately covered.

Much the same criticism can be leveled at UBS⁴. It at least has 1241, as well as 1071, but it omits 28, and 157, and instead gives us 1006 and some other useless items, and like SQE, it doesn't have a sufficient range of Byzantine uncials. Thus Legg still supplies genuinely useful information not available from any other single apparatus (or, in the case of 22 and 157, any apparatus at all). It may be inaccurate, but it remains surprisingly useful.

Merk

Editor. Text and apparatus edited by Augustinus Merk, S.J.

Date of Publication. The first edition, NovumTestamentum Graece et Latine, appeared in 1933. The tenthedition, issued nearly four decades after the editor's death,was published in 1984. Overall, however, the changes in theedition, in both text and apparatus, have been minimal.

The Text. Merk's Greek text is a fairly typicalmid-Twentieth-Century production, an eclectic edition whichhowever leans strongly toward the Alexandrian text. TheLatin text, as one would expect of a Jesuit, is theClementine Vulgate.

The Apparatus. The significance of Merk lies notin its text but in its apparatus -- by far the fullest ofthe hand editions, and accompanied by a serviceable criticalapparatus of the Vulgate (a noteworthy improvement, in thisregard, over the otherwise fairly similar edition of Bover).
Merk's apparatus is largely that of von Soden, translatedinto Gregory numbers and slightly updated. Merk includes nearlyall the variants in von Soden's first two apparatus, and asignificant number of those in the third. In addition to themanuscripts cited by von Soden, Merk cites several manuscriptsdiscovered since von Soden's time (papyri up toP⁵²,including the Beatty papyri; uncials up to 0207; minuscules upto 2430, although all but four minuscules and three lectionariesare taken from von Soden). Merk also cites certain versions andfathers, particularly from the east, not cited in von Soden.
But this strength is also a weakness. Merk's apparatusincorporates all the errors of von Soden (inaccurate collationsand unclear citations), and adds errors of its own: histranslation of von Soden's apparatus is occasionally inaccurate,plus the edition suffers from a very high numberof errors of the press and the like. Merk does not even providean accurate list of fathers cited in the edition -- e.g. theBeatus of Liébana is cited under the symbol "Be,"but the list of Fathers implies that he would be cited as"Beatus." The Venerable Bede, although cited relativelyoften (as Beda), is not even included in the list of Fathers! The list ofsuch errors could easily be extended (a somewhat more accurate listof fathers cited in Merk is found in the article on theFathers).
Thus the student is advised to take great care with the Merk.As a list of variants, no portable edition even comes close. Every studentshould have it. But knowing how far to trust it is another question.The following table shows a test of the Merk apparatus, based on thereadings found in the apparatus of UBS⁴in three books (Galatians, Philippians, 1 Thessalonians).The first column lists the manuscript, the second the number of readingsfor which it can be cited, the third the number of places where Merk'sapparatus disagrees with the UBS apparatus, and the fourth the percentageof readings where they disagree.

Manuscript	Readings	Disagreements	Percent Disagreement
P46	45	1	2%
	67	0	0%
A	63	0	0%
B	63	1	2%
C	34	1	3%
D	63	0	0%
Ψ	63	7	11%
6	63	5	8%
33	63	3	5%
81	63	1	2%
104	63	4	6%
256	59	5	8%
263	59	8	14%
330	59	9	15%
436	59	9	15%
462	58	5	9%
1175	51	4	8% (but see below)
1319	59	3	5%
1739	63	1	1%
1912	63	4	6%
2127	59	4	7%

(Note: Data for 330 and 462 taken from the collations by Davies.)

We should add one caveat, however: Merk does not list where manuscripts suchas P⁴⁶, C, and 1175 have lacunae -- in the case of 1175, he citesthe manuscript explicitly for certain readings where it does not exist! In addition,it is often impossible to tell the readings of the manuscripts in the bottomparts of his apparatus, as they are cited as part of al or rel pl.Thus the table cites 256 for 59 readings instead of the 63 citations for the Old Uncialsbecause there are four readings where it is simply impossible to know which readingMerk thinks 256 supports.
Still, we see that overall the Merk apparatus is almost absolutelyaccurate for the Old Uncials (though it sometimes fails to note thedistinction between first and later hands). Minuscules vary in reliability,though there are only three -- 263, 330, and 436 (all members of I^a3,which seems to have been a very problematic group) -- where Merk's apparatusis so bad as to be of no use at all. The conclusion is that students shouldtest the apparatus for any given minuscule before trusting it.
The Merk apparatus, adapted as it is from Von Soden, takes gettingused to. The apparatus always cites the reading of the text as alemma, then cites variant(s) from it. Normally witnesses will becited for only one of the two readings; all uncited witnesses areassumed to support the other reading. To know which witnesses arecited for a particular reading, however, requires constant referenceto Merk's list of groups (given in the introduction), as witnessesare cited by position within the groups, and often in a shorthandnotation -- e.g. 1s means "1 and the witness immediatelyfollowing" -- which in the Gospels is 1582; 1ss wouldmean "1 and the two witnesses immediately following"(1582 and 2193).
Note that "1s" is not the same as "1^s."1^s means "1 and all manuscripts which follow to the end ofthe group." So where 1s means 1 1582, 1^s means 1 1582 2193(keep in mind, however, that if the subgroup is large, not all manuscripts of thegroup may be intended). 1^r has yet another meaning: from 1 tothe end of the major group -- in this case, from 1 to 131.
All this is not as bad as it sounds, but the student is probablywell-advised to practice it a few times!
Other symbols in Merk's apparatus include >, indicating anomission; ^|, indicating a part of a versional tradition (or theGreek side of a diglot where the Latin disagrees); "rel" for"all remaining witnesses," etc. Many of the remaining symbolsare obvious (e.g. ~ for a change in word order), but the student should besure to check Merk's introduction in detail, and never assume a symbolmeans what you think it means!
The example below may make things a little clearer. We begin withthe table of witnesess -- in this case for Paul.

Let us take Romans 2:14 as our example verse. Merk's text of the verse(without accents) reads:
(14)οτανγαρ εθνη τα μηνομον εχονταφυσει τατου νομουποιωσιν, ουτοινομον μη εχοντεςεαυτοιςεισιν νομος
In the apparatus we have
14 γαρ ] δε G^| ar Ωρ^| -- i.e. for γαρ, the reading of Merk's text, the Greek side of G (but not the Latin), the Armenian, and part of Origen read δε. All other witnesses support Merk's text.
ποιωσιν B SA-1908 104-1852 D^s 467 1319-38 436 43 Cl Ωρ ] ποιη rel -- i.e. ποιωσιν is supported by B, S (=ℵ), the witnesses from A to 1908 (=A, 1739, 6, possibly 424**, and 1908), the witnesses from 104 to 1852 (=104, 326, 1175, 81, 1852), by D and all other witnesses to the end of its group (=D G 917 1836 1898 181 88 915 1912, with perhaps one or two omitted), by 467, by the witnesses from 1319 to 38 (=1319 2127 256 263 38), by 436, by 43, by Clement, and by Origen. The alternative reading ποιη is supported by all other witnesses -- i.e. by the uncited witnesses in the H group (in this case, P Ψ), by the entire Ca² group except 467, by the uncited witnesses of Ca³ (=920, 1835, etc.), by all witnesses of the Cb groups except 43, and by all remaining witnesses from 1518 on down to L at the end.
ουτοι] οι τοιουτοι G d t vg Ωρ^| -- i.e. for ουτοι G (and its Latin side g), the old latins d t, the vulgate, and part of Origen read οι τοιουτοι. Again, all other witnesses support Merk's text.

The Nestle Text

The history of the "Nestle" text is complex; the texthas undergone one major and assorted minor revisions, while theapparatus has been upgraded repeatedly. The sections below outlinethe history of the early versions of the edition, then proceedsto describe the modern form (Nestle-Aland 27 and its predecessorNestle-Aland 26).

Nestle Editions 1-25

The first edition of "Nestle" was prepared in 1898by Eberhard Nestle (1851-1913). It was not really a criticaltext; Nestle simply compared the current editions ofWestcott & Hort, Tischendorf, and Weymouth.The reading found in the majority of these editions became thereading of the text (if the three disagreed, Nestle adoptedthe middle reading). The apparatus consisted variant readingsfrom the three texts (plus a few variants from Codex Bezae).

The text was slightly revised with the third edition, whenthe text of Bernhard Weiss was substituted for that of Weymouth.With some further slight revisions, this remained the "Nestle"text through the twenty-fifth edition.

The nature of "Nestle" changed radically withthe thirteenth edition of 1927. This edition, under thesupervision of Eberhard Nestle's son Erwin Nestle(1883-1972), for the first time fully conformed thetext to the majority reading of WH/Tischendorf/Weiss.It also added in the margin the readings of von Soden'stext. But most importantly, it included for the firsttime a true critical apparatus.

Over the following decades the critical apparatuswas gradually increased, and was checked againstactual manuscripts to a greater extent (much of thiswas the work of Kurt Aland, whose contributionsfirst began to appear in the twenty-first edition of1952). More manuscripts were gradually added, and morevariants noted. It should be observed, however, that the"Nestle" apparatus remained limited; oftenno more than five or six manuscripts were noted foreach variant (it was exceedingly rare to find more thantwelve, and those usually comprehended under a groupsymbol); most manuscripts were cited only sporadically;the Byzantine text was represented by theTextus Receptus (ℜ) the Egyptian text (ℌ) wascited under an inadequate group symbol. Also, the apparatusincluded fewer variants than might be hoped -- not onlyfewer variants than von Soden and Tischendorf (which wasto be expected), but also fewer variants than Merk. Eventhe readings of Sinaiticus, Vaticanus, the papyri, andthe Textus Receptus were inadequately noted.

In addition, some regard the form of the apparatusas a difficulty. Instead of noting the text of variantsin the margin, a series of symbols are inserted in the text.The advantages of this system are brevity (the apparatusis smaller) and also, to an extent, clarity; the scopeof variants can be seen in the text. (Though the reason appearsto have been rather different: the Nestle apparatus was as itwas because the editors continued to use the original plates ofthe text, meaning that any apparatus had to fit in a fairlysmall space.)

The illustration below illustrates several of themajor features of the Nestle apparatus, along withsome explanations. The form of the apparatus resemblesthat of the twenty-sixth and twenty-seventh editions,but the same symbols are used in all editions. (Note: Ifyou cannot read the symbols clearly, perhaps because youdon't have the required unicode fonts, clickhere for a low-resbitmapped version).

° means that the following word is to be omitted.
⸋...⸌ means that the words between ⸋ and ⸌ are to be omitted
⸆ means that the word(s) in the margin are to be added
⸀ means that the word(s) in the margin are to be substituted for the wordin the text.
⸂...⸃ means that the word(s) in the margin are to be substituted for thewords in the text
⸉...⸊ means that the order of the words in the text are to be rearrangedas described in the margin.

Where a symbol is followed by a dot or a superscript number, it means that there are multiple instance of that sort of variation in the verse, and one is again referred to the appropriate point in the margin. So, for instance, if there are multiple omissions of single words in a verse, the symbols will be °,°¹, °², etc. If there are multiple insertions in the text, the notation will be ⸆, ⸇, ⸆¹, ⸆², and so forth. Multiple substitutions are marked ⸀, ⸁, ⸀¹, etc.

An artificially constructed sample of how the above might work is given below. The sample is of the beginning of Matthew 1, but the apparatus, with the exception of the variant in verse 3 which is found in the actual text, is entirely fake, beingset up to show how the Nestle apparatus works. The Nestle symbols are shown inred; the uncertain text in blue.

1 Βίβλος γενέσεως⸉Ἰησοῦ Χριστοῦ⸊υἱοῦ Δαυὶδ ⸋υἱοῦἈβραάμ⸌
   2 Ἀβραάμ ἐγέννησεν τὸν Ἰσαάκ,⸂Ἰσαάκδὲ⸃ ἐγέννησεν°τὸν
Ἰακώβ, Ἰακώβ δὲ ἐγέννησεν τὸν Ἰούδαν καὶ τοὺς ἀδελφοὺς
°¹αὐτοῦ.3 Ἰούδας δὲ ἐγέννησεν τὸν Φάρες καὶ τόν⸀Ζάρα

The apparatus would appear as follows:

1 ⸉ B | ⸋ L pc •2 ⸂ και Ισαακ א | ° B D al |
°¹ B L 892 sa pc¦  ℙ¹ א D W ⨏¹⨏¹³ 33 𝔐 latt sy •
3 ⸀  Ζαρε ℙ¹ B mae

Here is how this is to be interpreted:

1 ⸉ B indicates that B (only) rearranges the words inthe order Χριστου Ιησου | ⸋ L pc indicatesthat L and a few other, lesser witnesses omit the words υιου Αβρααμ• 2 ⸂ και Ισαακ אindicates that א (only) reads και Ισαακ forΙσαακ δε |° B D al indicates that B, D, and a selection of otherwitnesses omit τον| °¹ B L 892 sa pc¦  ℙ¹ א D W ⨏¹⨏¹³ 33 𝔐 latt syindicates that B, L, 892, the Sahidic Coptic, anda few lesser witnesses omit αυτου; the word is found in ℙ¹, א,D, W, family 1 (⨏¹), family 13 (⨏¹³), the MajorityText (𝔐) and the witnesses included in it (e.g. K, Γ, Δ,Ψ, 28, 565, 579, 1010, 1424), the entire Latin tradition (latt), andthe Syriac tradition (sy).3 ⸀  Ζαρε ℙ¹ B mae indicatesthat ℙ¹, B, and the Middle Egyptian Coptic, and those threewitnesses only, read Ζαρε for the Ζαρα of the text.

This notation has been preserved in all texts of Nestle, despite occasionalcomplaints. Most of the other problems mentioned above were removed in thecompletely redone Twenty-sixth edition:

Nestle-Aland Editions 26-27

The twenty-sixth edition of Nestle-Aland, published in 1979,was the first to be produced entirely under the supervision ofKurt Aland. The result was very nearly a new book.

The Text. The text of NA²⁶ is, inall major respects, the same asthat of the United Bible Societies Edition,of which Aland was an editor. The only differences lie in mattersnot directly associated with textual criticism, such as accents, punctuation,and arrangement of paragraphs. The characteristics of the text aredescribed under the section on the UBS edition.

The Apparatus. The apparatus of NA²⁶ isequally radically revised. Instead of the haphazard citationof witnesses found in the earlier editions, a select listof witnesses is cited for all readings. The witnesses citedinclude all papyri, all early uncials, and a selection oflate uncials and minuscules -- usually about twenty witnessesfor each reading. The most important of these witnesses, thepapyri and the early uncials, are cited explicitly. (In thetwenty-seventh edition, certain important minuscules -- 33,1739, 1881, 2427 -- are elevated to the ranks of theexplicitly cited witnesses.) The remaining witnesses, mostlyByzantine or mixed, are cited explicitly only when theydiffer from the Byzantine text; otherwise they are containedwithin the Majority Text symbol 𝔐 (that is a Gothic M; your browsermay or may not display it correctly). An example of the use of the Majority Text symbol is shown in the example above.

This apparatus offers distinct advantages. It cites manyimportant manuscripts in a minimum of space, and is quiteconvenient to use once one becomes accustomed to it. In addition,the Nestle-Aland apparatus is probably the most accurate sinceTischendorf. The several appendices offer additional usefulinformation, e.g. about the differences between the majortwentieth century editions. The margin has a much fullerset of cross-references than most comparable editions, andincludes several ancient systems of enumeration.

There are still a few drawbacks. Some witnesses have lacunaewhich are not noted in the appendix. The reader may thereforeassume, falsely, that a witness agrees with the majority textwhen in fact it is defective. (This was a particular problemin the twenty-sixth edition with 33, which is often illegible.This was solved in the twenty-seventh edition by citing 33explicitly. However, the even more problematic 1506 is stillnot cited explicitly. In addition, the Nestle text does notlist lacunae precisely; when it says, e.g., that 81 lacks Acts 4:8-7:17,17:28-23:9, it means that it lacks those verses in their entirety.The verses on the edge of these lacunae -- Acts 4:7, 7:18, 17:27,23:10 -- will almost certainly be fragmentary, so one cannottrust citations from silence in those verses.)

The set of variants in NA²⁶ is stillrelatively limited; with minor exceptions, only those variantsfound in NA²⁵ are cited in NA²⁶. Thethorough critic will therefore need to use a fuller edition --Tischendorf, Von Soden, or Merk -- to examine the full extentof variation in the tradition.

Students are also advised to remember that Nestle-Alandcites only Greek and Latin fathers. The eastern traditionis entirely ignored. Those wishing to know the text ofEphraem, say, will have to turn to another source.

Nestle-Aland Edition 28

The twenty-eighth edition of Nestle-Aland, published in 2012,is clearly an extension of the tradition of the 26th/27th edition,but it was revised by Holger Strutwolf and others; no Nestles orAlands were involved. The text is revised in the Catholic Epistles;the apparatus is revised throughout. There is a new foreword as well,explaining the changes in a very passive-voice sort of way.

In the Catholic Epistles, the text is the entirely new versionfound in "The" Editio Critica Maior, that is, themajor edition released by Institute für Neutestamentliche Textforschung(the fact that they call it "the" major edition perhapstells you something about the editors).

The apparatus has also been updated, substantially. The fundamentalchange is that the cited witnesses are now cited in full. Thereare no constant witnesses of the "second order," that are onlycited as part of 𝔐.

The chart below perhaps illustrates this better than an explanation.It shows the text of 1 Corinthians 6:14-16 and the apparatus as it appearsin NA²⁵, NA²⁶, NA²⁷, and NA²⁸.Note that the text, including even the critical symbols (⸀ etc.), is the samein all the editions, but the apparatus changes. (The critical symbols areshown in color to make them more obvious.)

[14] ὁ δὲ θεὸς καὶ τὸν κύριον ἤγειρεν καὶ ἡμᾶς⸀ἐξεγερεῖ διὰ τῆς δυνάμεως αὐτοῦ.[15] ⸆ οὐκ οἴδατε ὅτι τὰ σώματα⸀ὑμῶν μέλη Χριστοῦ ἐστιν; ⸁ἄρας οὖν τὰ μέλη τοῦ Χριστοῦ ποιήσωμέλη; μὴ γένοιτο. [16] °ἢ οὐκ οἴδατεὅτι ὁ κολλώμενος τῆ πόρνῃ ἓν σῶμά ἐστιν; ἔσονται γάρ, φησίν, οἱ δύο εἰςσάρκα μίαν.

NOTE: In NA^26-28, the ἢ at the start of verse 16 is in [brackets] as dubious; it is not bracketed in NA²⁵.

The NA²⁵ Apparatus

14 ⸀ εξηγειρεν P^46c2 B 1739 r t Or; h : εξεγειρει P^11.46* A D 69 pc : txt P^46c1 א C 𝔎 pm vg     15 ⸆ η G pc  |  ⸀ ημ- א* A Ir^arm  |  ⸁ αρα P al : ἢ αρα G     16 ° P⁴⁶ 𝔎 Dal Mcion

The NA²⁶ Apparatus

14 ⸀ εξηγειρεν 𝔓^46c2 B 6 1739 it vg^mss ¦ εξεγειρει 𝔓^11.46* A D P 1241^s pc ¦ txt 𝔓^46c1 א C Ψ 𝔐 vg sy^h co; Ir^lat Tert Epiph Ambst • 15 ⸆ η F G a; Epiph | ⸀ ημ- א* A; Epiph | ⸁ αρα P Ψ 81. 104. 630. 1175. 1241^s. 1739^v.l. 2495 pm ¦ ἢ αρα F (G) ¦ txt 𝔓⁴⁶ א A B C D K L 33. 365. 1739*. 1881. 2464 pm lat sy; Epiph • 16 ° 𝔓⁴⁶ D K L Ψ 6 pm r sy^h; Tert Spec ¦ txt א A B C F G P 33. 81. 104. 365. 630. 1175. 1241^s. 1739. 1881 2464. 2495 pm lat sy^p; Cl Cyp Lcf Epiph

The NA²⁷ Apparatus

14 ⸀ εξηγειρεν 𝔓^46c2 B 6 1739 pc it vg^mss; Ir^lat v.l. Or^1739mg ¦ εξεγειρει 𝔓^11.46* A D* P 1241^s pc ¦ txt 𝔓^46c1 א C D² Ψ 33. 1881 𝔐 vg sy^h co; Ir^lat Tert Meth Ambst • 15 ⸆ η F G a; Meth | ⸀ ημ- א* A; Ir^arm | ⸁ αρα P Ψ 81. 104. 630. 1175. 1241^s 1739^c 2495 pm ¦ ἢ αρα F (G) ¦ txt 𝔓⁴⁶ א A B C D K L 33. 365. 1505. 1739*. 1881. 2464 pm lat sy; Ir^lat Meth • 16 ° 𝔓⁴⁶ D K L Ψ 6 pm r sy^h; Spec ¦ txt א A B C F G P 33. 81. 104. 365. 630. 1175. 1241^s. 1505. 1739. 1881 2464 pm lat sy^p; Cl Cyp Lcf Epiph

The NA²⁸ Apparatus

14 ⸀ εξηγειρεν 𝔓^46c2 B 6 1739 it vg^mss; Ir^lat v.l. Or^1739mg ¦ εξεγειρει 𝔓^11.46* A D* P 1241 ¦ txt 𝔓^46c1 א C D² K L Ψ 33. 81. 104. 365. 630. 1175. 1505. 1881. 2464 𝔐 vg sy^h co; Ir^lat Tert Meth Ambst • 15 ⸆ η F G ar; Meth | ⸀ ημ- א* A; Ir^arm | ⸁ αρα P Ψ 81. 104. 630. 1175. 1241. 1739^c. 2495 pm ¦ η αρα F (G) ¦ txt 𝔓⁴⁶ א A B C D K L 33. 365. 1505. 1739*. 1881. 2464 pm lat sy; Ir^lat Meth • 16 ° 𝔓⁴⁶ D K L Ψ 6 pm r sy^h; Spec ¦ txt א A B C F G P 33. 81. 104. 365. 630. 1175. 1241. 1505. 1739. 1881. 2464 pm lat sy^p; Cl Cyp Lcf Epiph

The gain in information between NA²⁵ and NA²⁶ is obvious; in NA²⁵, the only manuscripts whose readings you can reliably infer from the apparatus are 𝔓⁴⁶ א A B D G, plus C if you can be sure it is legible. In NA²⁶, you can infer the readings of all these plus K L P Ψ 33 81 104 365 630 1175 1241 1739 1881 2464 2495 (the latter replaced by 1505 in NA²⁷). There is also more information about the non-Latin versions and about the Fathers (although it is interesting to see how the readings of the fathers shift over time; note e.g. how Origen is cited in verse 14 in NA²⁵, disappears in the next edition, then reappears; also how Marcion is cited in NA²⁵ for verse 16, but not in the later editions).

The down side is, you have to know where there are lacunae in the witnesses to use NA²⁶ and NA²⁷. P has major lacunae; do you have them all in your head? And some of the witnesses -- notably 33 -- have un-acknowledged lacunae, where they are illegible and NA²⁶ gives you no clue. This was addressed in NA²⁷ by citing 33 explicitly, but that didn't address the fundamental problem. This was finally taken care of in NA²⁸ by citing all witnesses.

Except that this doesn't help when the apparatus cites only the variant and not a txt reading. Also, when it/lat are cited, there is no way to know which Old Latins support which reading. And the other down side is that the NA²⁶ apparatus is two and a half times as big as that in NA²⁵, and by the time we get to NA²⁸, it's three times as bulky. The increase in size from NA²⁵ to NA²⁶ is certainly justified. That from NA²⁶ to NA²⁸ is far harder to justify -- all that is needed is to note exactly where witnesses are illegible! If NA²⁹ were to revert to the format of NA²⁶, but note lacunae, then NA²⁹ could add perhaps half a dozen more useful witnesses (e.g. 049 6 256 424 442 2127) at no additional cost in space. The Stuttgart Vulgate and many volumes of the Göttingen Septuagint note lacunae "on the page"; there is no reason why "Nestle" could not follow suit.

Still, there is no question that the apparatus is getting better. And the new apparatus of the Catholic Epistles is especially noteworthy. A comparison of the "constant witnesses " in each will make this clear. (Incomplete manuscripts are shown in brackets.)

NOTES:
* NA²⁸ claims to have retained K and L as constant witnesses in the Catholic Epistles, but they do not seem to be cited even when the Byzantine text is divided. Thus, for these witnesses, one must consult NA²⁷.

It will be observed that NS²⁸ has dropped several fragmentary uncials -- a sad decision but one that affects the text very little because they are fragmentary. It also drops the Byzantine uncials K L, which is truly unfortunate; no matter how bad one thinks the Byzantine text, one can get a better feeling for it by having some actual manuscript witnesses as well as the "Byz" symbol. We also lose 1505, an important manuscript of Family 2138. It is changes like this that make it almost a rule of the NA editions that for every two steps forward there must be a step back. But in return for this loss we gain a large number of very important minuscules: 5 307 436 442 642 1243 1448 1611 1735 1852 2344 2492. That's twelve manuscripts, of several different types. With these manuscripts added to the apparatus, it appears that we finally have a sufficient set of manuscripts to span the spectrum of text-types -- except the Byzantine text.

Das Neue Testament auf Papyrus

Editor. Volume 1 (Catholic Epistles) edited by K. Junack and W. Grunewald; Volume 2.1 (Romans, Corinthians) edited by K. Junack, E. Güting, U. Nimtz, K. Witte; Volume 2.2 (Galatians-Hebrews) edited by K. Wachtel and K. Witte; additional volumes forthcoming.

Date of Publication. Ongoing. First volume published 1986.

The Text. This is not truly a critical text; in one senseit is not a text at all. A continuous text (that of the United Bible Societies Edition) is printed,but this is followed by continuous texts of the various papyriextant for the particular passage.

The significance of this edition, therefore, is not for its text butfor its apparatus, which is the fullest collection of the texts of thepapyri and uncials now known. It is also esteemed as highly accurate.

The apparatus in general falls into three parts: The text (asfound in UBS and any extant papyri), the commentary on the papyri(describing their readings as well as information on early editions),and the full apparatus, noting readings of all papyri and uncialsextant for this passage.

It should be observed that the edition is not a true collationof the uncials, though it is a full transcription of the papyri.While every significant variant in the uncials is noted, spellingand orthographic variants are not noted, nor peculiar forms used inthe manuscripts (e.g. the text does not note places where D/06confuses the endings -θεand -θαι).

The apparatus of the Auf Papyrus edition is unusually simpleand straightforward. The three basic sections of the apparatus are shownin the sample below (adapted, obviously, from the apparatus for Philippians1:1. This is the actual apparatus, save that it has been reset for on-screenclarity and omits all sections not relevant to Philippians 1:1).

	The Basic Text: The UBS reading, with the readings of P46 below (in smaller type).
	The Commentary, describing the details of what the papyri read, including comments on previous editions. Note that, had other papyri contained this passage, their readings would also have been discussed under separate heads.
	The Apparatus, showing the major readings of both papyri and uncials. The section for Philippians 1:1 is exceptional in that it has a part both for the book title and the text itself. Most pages will show only one part.

The first section, at the top of the page, shows the readings ofP⁴⁶ in detail, setting them off against the UBS text. Notethat the apparatus shows even the page layout (e.g. the lineΠΡΟΣΦΙΛΙΠΠΗΣΙΟΥΣis page 168, line 21. This is noted with the notation "|^168,21").Where the text of the papyrus agrees exactly with the UBS text for a given word,this is noted with the ditto mark (,,). If there is any difference, or ifsome of the letters in the papyrus are uncertain or illegible, the wordis spelled out, with (as is normal) dots below letters indicating uncertaintyand letters in brackets [ ] indicating lacunae. Observe that P⁴⁶is totally defective for the final words of verse 1, and so there is notext cited below the UBS text for that line.

Below the actual text is the discussion, describing the actual readingsand the differences between editions. Notice, first, the discussion of order,followed by the discussion of individual lines. So, e.g, we learn that theKenyon edition (Ed. pr.²) omitted the terminal sigma ofΦΙΛΙΠΠΗΣΙΟΥΣ in thetitle, as well as the two uncertain vowels ofδουλοι in line 22 andall letters in line 23.

Below the discussion of the papyri we see the actual apparatus. This isexceptionally clear and easy to understand. To begin with, it lists allpapyri and uncials which contain the passage (though lacunae in theuncials are not noted with the fullness of the papyri). The apparatusis straightforward: Every variant starts with a lemma (the UBS text ofthe variant in question), along with a list of supporters if appropriate.This is followed by the variant reading(s) with their supporters.

Again, we should note what this edition is not. It is not,despite the very full apparatus (which genuinely invites comparisonto Tischendorf, save that it is restricted to readings found in papyriand uncials), a collation. Since the orthographic variants of theuncials are not noted, you cannot use it to reconstruct the actualtext of an uncial. And if you wish a collation of a papyrus, youwill have to do it yourself. Finally, if you wish to know whichcorrector of an uncial gave rise to a correction, you may have torefer to another edition.

Despite these drawbacks, Das Neue Testament auf Papyrus isone of the most useful tools available -- the first real step in manyyears toward a full critical apparatus of the Epistles. It's mostunfortunate that it is priced so high; this volume should be on everytextual critic's desk, not confined to seminary libraries.

Souter

Editor. Critical apparatus by Alexander Souter; the text itself is considered to be that underlying the English Revised Version of 1881.

Date of Publication. The first edition, NovvmTestamentvm Graece, appeared in 1910. A revised edition(offering, e.g., the evidence of the Beatty papyri) was releasedin 1947.

The Text. The text of Souter is that of ArchdeaconEdwin Palmer, and is considered to be the Greek text underlyingthe English Revised Version. This produced a rather curiousedition. To begin with, the scholars responsible for the RVwere mandated to make the fewest possible changes in the textof the King James Version. It was decided that changes in thetext could only be made by a two-thirds majority of the committee.
What is more, the committee had a rather haphazard method fordetermining the original text, allowing Hort(who generally favoured the Alexandrian text) andScrivener (who preferred a more Byzantinetext) to state their cases, then choosing between the two. The resultis a text which frequently follows Hort, but sporadically adoptsByzantine readings as well.
Palmer's method exacerbated this problem. Since he wished tokeep the text as close as possible to the KJV and the TextusReceptus, he made only the minimal number of revisions to theGreek text. Thus the text of Souter always follows the TR atpoints of variation which cannot be rendered in English, whilemore often than not following the text of Westcott & Hortat points where the variation affects the sense of the passage.
At least, this is what commentaries on the edition say.Interestingly, Souter's introduction does not mention Palmer. Evenmore interesting, a check reveals that the text of the Apocalypsewas not prepared by this method; it regularly goes againstthe TR in variants which have no significance in English. I donot know the source of Souter's text of that book. Mark's textalso has many agreements with Westcott and Hort where a TR readingwould be expected, though here it is less consistent. One suspectsthat Palmer was not very careful in this book.
Still, that leaves perhaps 25 books largely based on the TextusReceptus. For this reason, critical editors rarely pay much attentionto the text of Souter. The apparatus is another matter.

The Apparatus. Souter's apparatus lists only a limitednumber of variants (perhaps a third the number found in Nestle-Aland).The apparatus is, however, exceptionally clear and easy to use (whichis fortunate, since the introduction consists of a mere two and ahalf pages, in Latin). The reading of the text is given, usually followedby its support (in the order papyri, uncials, minuscules, version,fathers; Souter does not classify witnesses into types). The variant readingsand their support follow (in some readings where the variant isthinly supported, the evidence for the text is not listed).
A noteworthy feature of Souter's apparatus is the degree ofdetail it gives about the Fathers. These are cited in carefuland specific detail. This is one of the best features ofSouter's edition.
The revised edition of Souter cites papyri through P⁴⁸,uncials through 0170, minuscules through 2322, a full list of versions(including Armenian, Gothic, Georgian, and Ethiopic), and nearly twohundred fathers of all eras. The Byzantine text is cited under thesymbol ω.

Swanson

Note that Swanson edited two somewhat different editions, the Horizontal Line Synopsis of the Gospels, intended to show parallel passages, and New Testament Greek Manuscripts, which is a presentation of Greek manuscripts. Both undertake to show the complete readings of all manuscripts cited.

Editor. Critical apparatus and parallels compiled by Reuben J. Swanson.

Date of Publication. The first part of the original Swanson volume, The Horizontal Line Synopsis of the Gospels, Greek Edition; Volume I.The Gospel of Matthew, was published in 1982

The editions of Greek New Testament Manuscripts began with Matthew in 1995, followed by Mark (1995), Luke (1995), John (1995), Acts (? -- I don't have this one), Galatians (1999), Romans (2001), 1 Corinthians (2003). Dr. Swanson died in 2009, so presumably the series will not be completed.

The Text. The Greek text of the Horizontal Line Synopsis, as noted, is that of the UBS edition; Greek New Testament Manuscripts doesn't really have a base text). The value of Swanson lies not in it text but in its bulky butextremely clear apparatus -- in Paul, even more than the Das Neue Testament auf Papyrus volumes, Swanson probably deserves the term "the new Tischendorf."

The Apparatus. Swanson's apparatus in New Testament Greek Manuscripts is clear enough -- just the texts of all the cited manuscripts in parallel lines. The more complicated Horizontal Synopsis apparatus consists of three parts:Texts with parallels, critical apparatus, and list of Old Testament allusions (thelater simply a list of the Gospel verses and the Old Testament passages they cite).

The apparatus of parallels is perhaps the simplest of any now available. Thefirst line of the text is that of the Gospel under consideration. (This text canreadily be recognized by the typeface; in Matthew, e.g., it is underlined.) Below itare the texts of the other gospels. This arrangement in parallel lines has theadvantage of allowing much easier comparison with the other gospels. Theparallels are pointed up by the type, since places where the other gospels matchthe chosen edition are printed in the same style. The example below illustratesthe point for the opening words of Matthew 9:1 and its parallels in Mark 5:18,Luke 8:37b.

The apparatus is equally straightforward (and equally bulky). The apparatus for theabove line of text, for instance, appears as follows, showing the full text of all thewitnesses Swanson cites, including variations in spelling:

As mentioned, in New Testament Greek Manuscripts, Swanson simply prints the texts of the various manuscripts. So here, for instance, is a part of Galatians 5:12.

(Where u represents the UBS reading, w that of Westcott and Hort, and τ the Textus Receptus; all other sources cited are Greek manuscripts.)

Thus Swanson lists seven different readings here, although one of them -- οἱ αναστατοῦν of P -- is not really a separate reading; it's just that P is defective at the end of the reading and cannot testify to the presence or absence of ὑμᾶς. Still, we have two real variants: the particular verb and its form, and the presence or absence of ὑμᾶς. Only one witness -- 1 -- omits ὑμᾶς. As for the verb, the vast majority of witnesses have ἀναστατοῦντες; the only exceptions are C, 131, 330, and 910. Hence the use of the notation [↑, which means that all the witnesses on this line following this symbol go with the reading above them. So ἀναστατοῦντες is read not only by B P⁴⁶ ℵ A D F G K L Ψ 056 075 0122 6 33 69 88 104 205 209 226 323 440 460 489 517 but in fact by all the following witnesses:P⁴⁶ ℵ A B D F G K L Ψ 056 075 0122 1 6 33 69 88 104 205 209 226 323 440 460 489 517 547 614 618 796 927 945 999 1175 1241^s 1242 1243 1245 1270 1315 1319 1352 1424 1448 1505 1573 1611 1646 1734 1735 1738 1739 1827 1836 1837 1854 1874 1891 1982 2125 2147 2344 2400 2412 2464 2495 2815.

ἀναστατοῦντες is also the reading of the editions cited: u w τ Er¹, that is, the United Bible Societies fourth edition/Nestle-Aland 27th edition (u), Westcott and Hort (w; the particular edition cited is the 1935 Macmillan printing), the Textus Receptus (τ; the particular version cited is the 1873 Oxford edition), and Erasmus's first edition (Er1). Observe that, because Swanson prints only continuous texts of manuscripts, and the editions do not always agree with any particular manuscript, the editions may be cited for several lines (especially in the case of w, since Westcott and Hort had both the reading of text and margin).

It will be evident that Swanson cites a lot of witnesses -- more than any other recent edition -- and cites them in their entirety, not just for specific readings. It is an amazing publication.

This strength of Swanson is also a weakness, as it results in extremely massive volumes.The Synopsis of Matthew, for instance, requires 362 pages of text and apparatus. Takingpage size into account, this is 15.4 square metres of paper surface. By comparison, the Alandsynopsis of all four gospels takes only 29.1 square metres, and manages to include more material(more manuscripts in the apparatus, if perhaps a poorer selection; citations from non-canonicalgospels and other sources; a fuller set of cross-references, etc.)

Notes at the bottom of the page cite various divisions of the text (κεφαλια, τιτλοι, lectionary incipits, Eusebian numbers), limited cross-references, and (in some volumes) errata in other editions.

The list of witnesses cited in Swanson is, in many ways, superior to thevarious Aland editions. It is a relatively short list, omitting fragmentarymanuscripts and (for obvious reasons, given the nature of the apparatus)versions and fathers, but the witnesses are generally balanced (as opposed to theAland apparatus, which is biased toward the Alexandrian text and heavily biasedagainst the Byzantine). Again takingMatthew as an example, Swanson includes the earliest Alexandrian witnesses(ℵ B C L),the one and only "Western" witness (D), several leading "Cæsarean"witnesses (Θ 1 13 28 565 1582),two important mixed witnesses (P⁴⁵ W), and (most unusually) an adequateset of Byzantine witnesses (A E F G K Y Π).While the apparatus contains some errors (inevitable in a project of such scope), itis generally accurate, and contains details not found in any other critical edition.It is also interesting to examine a passage such as Matthew 15:22, where the Nestletext seems to indicate a fairly stable tradition (no variant with more than fourreadings), but Swanson reveals no fewer than thirteen variants in this passage, despiteonly fifteen of his witnesses being extant.

Tasker

Editors. Text and apparatus compiled by R. V. G. Tasker based on the version translated in the New English Bible.

Date of Publication. The New English Bibleitself appeared in 1961; Tasker's retroversion into Greek, TheGreek New Testament, Being the Text Translated in The New EnglishBible, appeared in 1964. (As noted, Tasker's text is aretroversion; for the most part the NEB committee did not actuallyprepare a text.)

The Text. As has often been the case when a text iscompiled by a translation committee, Tasker's text is ratheruneven. It has been admitted that the reading adopted is oftensimply that preferred by the person who first attempted atranslation. The result is a text largely Alexandrian (normallyfollowing the pre-UBS Nestle text on which it is largely based), butwith odd mixtures of "Western" and Byzantine readingsdepending on the opinions of the translators. This text, sinceit does not adhere to any textual theory or display much coherence,has not met with widespread approval.

The Apparatus. Tasker's apparatus is very limited;it discusses only the few hundred variants noted in the NEBmargin. Only a handful of manuscripts (including 11 papyri upto P⁵¹, 27 uncials up to 0171, and 44 minusculesup to 2059) are cited, and those sporadically. It is a rarenote that cites more than ten manuscripts. On the other hand,the notes do describe why the committee adopted thereading it did -- a useful practice since adopted by the UBScommittee in its supplementary volume.

Tischendorf

Editors. Text and apparatus edited by Constantin vonTischendorf.

Date of Publication. Tischendorf published no fewerthan eight major editions in his life, as well as abridgededitions and various collations and facsimiles. His magnumopus, however, was the Editio octava critica maior (1869-1872),which remains unsurpassed as a complete edition of the NewTestament text.

The Text. Tischendorf's text is eclectic, taking readingsfrom many sources; Tischendorf did not have a detailed textual theory.In practice he had a strong preference for the readings of hisdiscovery ℵ,especially where it agreed with D. His text thus has somethingof a "Western" tinge, although it is generallyAlexandrian (insofar as that text was known in themid-mineteenth century, before B was made widely known).The resulting text, therefore, is not held in particularlyhigh regard; the value of Tischendorf lies in...

The Apparatus. Tischendorf's apparatus was, inits time, comprehensive, and it remains the most completeavailable. It cited all major readings of all major manuscripts,offering the evidence of almost all known uncials,plus noteworthy readings of many minuscules, the versions,and the Fathers.
Tischendorf's apparatus is generally easy to read,particularly if one knows Latin. A lemma is citedfor all variants. If each variant has significant support,the evidence for the text is listed following thelemma, followed by the variant reading(s) and their support.If the variant is supported by only a few witnesses, thevariant reading is cited immediately after the lemma.So, for example, in Gal 1:4 the apparatus reads:

περι cumℵ*ADEFGKLPal⁵⁰ fere syr^p Or^1,238 etc ...ς (= Gb Sz)υπερ cumℵ^cB17. 67** al sat mu Ign^intpol314 al

This translates as περι,the reading of Tischendorf's text (read also by the uncited editions,i.e. Lachmann and Tischendorf⁷) is supported by the uncialsℵ*A D E(=D^abs) F G K L P and about fifty other witnessesplus the Harklean Syriac (syr^p) and the cited textof Origen. The variantυπερis supported by the Textus Receptus (ς)and the editions of Griesbach and Scholz; byℵ^c,B, 17 (=33), 67** (=424^c), by many other Greek witnesses, andby the cited text of Ignatius.
The greatest single difficulty with Tischendorf'sapparatus is the nomenclature. Tischendorf died beforehe could finish his introduction, so many of the witnessescited were difficult to identify (this is particularlytrue of the Fathers, cited by a complex system of abbreviations).Another complication is attributions; Tischendorf lived in thenineteenth century, and even he did not have thetime or the resources to verify everything he cited (norcould he always identify the manuscripts cited in prioreditions). So one often encounters a notation such as"6 ap Scri" (i.e. 6 according to Scrivener) or"cop^ms ap Mill et Wtst" (i.e. a manuscriptof the [Bohairic] Coptic according to Mill and Wettstein).An introduction supplying much of the needed background wassupplied by Caspar Rene Gregory in 1894, but it is worthremembering that Tischendorf wrote before Gregoryrevised the manuscript numbering system. Thus almostall minuscules (except in the Gospels), and even someof the uncials, have the wrongnumbers. In Paul, for instance, the minuscules most oftencited include 17, 31, 37, 39, 46, 47, 67, 71, 73, 80,and 115; in modern notation, these are 33, 104, 69, 326,181, 1908, 424, 1912, 441+442, 436, and 103. In addition,the names used for the versions have sometimes changed(e.g. syr^p is the Harklean version, not thePeshitta!). To make matters worse, Tischendorf often didnot even use numbers for manuscripts; the sigla for morerecently-discovered documents often consists of a letterand a superscript indicating a collator, e.g. a^scrmeans the "a" manuscript collated by scr=Scrivener.This is the manuscript we know as 206. Most of the manuscriptscited under these symbols are relatively unimportant, butit is worth noting that lo^ti=p^scr isthe important minuscule 81.
To save space, in the Gospels Tischendorf citesa group of uncials as unc⁹; these representa block of Byzantine uncials.
In addition to manuscripts, Tischendorf cites thereadings of earlier editions: the Stephanus and Elzevireditions of the Textus Receptus,Griesbach, Scholz, Lachmann, and Tischendorf's ownprevious edition). (In fact, Tischendorf's editio minorincludes only those variants where these editionsdisagree.) Tischendorf also gives more explicit Latinevidence than most editions; see the notes onTischendorf under theLatin Editions.

United Bible Societies Edition

Editors. Original edition compiled by Kurt Aland,Matthew Black, Bruce M. Metzger, and Allen Wikgren; Carlo M.Martini joined the committee for the second and third editions;the fourth edition was prepared by Barbara Aland, Kurt Aland,Johannes Karavidopoulos, Martini, and Metzger.

Date of Publication. The first edition, The GreekNew Testament, appeared in 1966. The second edition,slightly revised, appeared in 1968. The third edition (1975) containeda significantly revised text (now generally cited as UBSor GNT) and a slightly revised apparatus. Thefourth edition (1993) has the same text as the third, but asignificantly revised apparatus.

The Text. The UBS³ text, which is also sharedby the 26th and 27th editions of Nestle-Aland, was prepared by acommittee. As a result, it has few of the erratic readings which mightbe found in the text of a single editor (a fact which has been inlarge measure responsible for its widespread adoption). On theother hand, it is a strongly eclectic text, with no clear textualtheory behind it. In general it follows the Alexandrian witnesses,and is closer to the Westcott & Hort text than most of the othermodern editions, but it is not as radically Alexandrian as Westcottand Hort.
The supplementary volume to the edition describes how the committeedecided its text -- but only by example. The volume gives the basis of whythe committee chose many readings -- but makes noattempt to describe the theories followed by the five editors. Nor dowe know how the individual editors voted on the various readings (exceptfor the handful of readings where they have filed signed "minorityopinions"). We have very little real sense how the text came about.Despite its widespread acceptance, it does not really conform to anyparticular theory of the text.

The Apparatus. The apparatus of UBS is extremely limited;it is concerned only with variants "meaningful for translators."In any given chapter of a book, one can expect to find only a half dozenor so variants. Thus the apparatus can in no sense be considered complete.
On the other hand, the apparatus is easy to use and very full. Foreach reading, all papyri, all early uncials, and a handful of lateuncials are cited, as are several dozen minuscules, an assortment oflectionaries, a number of versions, and a wide selection of fathers.All witnesses are explicitly cited for all variants, usually in theorder papyri, uncials, minuscules, lectionaries, versions, fathers.(There are a few minor exceptions to this; lectionaries are generallygrouped under the symbol Lect, and in the fourth editioncertain uncials are listed following the symbol Byz, denotingthe Byzantine text.)
Care must be taken with the list of witnesses, however.UBS¹-UBS³ contain lists of uncials and minusculescited; however, many of the uncials (e.g. E F G H of the gospels) arecited only exceptionally (this even though the list implies they arecited fully), and many of the minuscules are cited foronly part of their content. The correct list of minuscules citedfor each section of UBS³ is as follows:

• Gospels: (family 1) (family 13) 28 33 565 700 892 1009 1010 1071 1079 1195 1216 1230 1241 1242 1253 1344 1365 1546 1646 2148 2174
• Acts: 33 81 88 104 181 326 330 436 451 614 629 630 945 1241 1505 1739 1877 2127 2412 2492 2495
• Paul: 33 81 88 104 181 326 330 436 451 614 629 630 1241 1739 1877 1881 1962 1984 1985 2127 2492 2495
• Catholics: 33 81 88 104 181 326 330 436 451 614 629 630 945 1241 1505 1739 1877 1881 2127 2412 2492 2495
• Revelation: 1 94 1006 1611 1828 1854 1859 2020 2042 2053 2065 2073 2081 2138 2344 2432

This problem has been reversed in UBS⁴, which explicitly listswhich minuscules are cited for which sections -- but no longer lists theactual contents of the manuscripts. This information must now begathered from other sources.

Vogels

Editors. Heinrich Joseph Vogels.

Date of Publication. Original Greek text published 1920;Latin parallel added 1922; final edition published 1955.

The Text. It's hard to imagine a critic who would ratethis text highly. The editing principle, if there is one, seems tohave been "choose the Alexandrian reading unless the Byzantineis easier." This is especially true in the gospels, where theByzantine element is very strong (almost strong enough that we couldcall it a Byzantine edition for those books), but has some truthelsewhere also. The text has many major agreements with the Byzantinetext (e.g. Colossians 2:2, where Vogels chooses the Byzantine readingagainst the united opinions of every modern editor), but also curiousagreements with the Alexandrians. It is thus the most Byzantine ofthe major editions, with some influence from Von Soden, but notByzantine enough to be considered even faintly a Majority Textedition.

The Latin side, as one would expect of a Roman Catholic scholar,is the Clementine Vulgate.

This perhaps explains the nature of the text. This is not reallya Greek edition. It is the Vulgate in Greek dress. Vogels, by andlarge, took the Latin, found the closest Greek reading among themanuscripts, and adopted it. There is very little critical sense to theresult, and even less value in the result. Vogels had no theory of thetext; he was just making a crib.

The Apparatus. The apparatus is as frustrating as thetext. The number of variants cited is at the low end of adequate,the number of witnesses cited is small -- and the minuscules arecited by Tischendorf numbers!

It's not hard to read the apparatus; it uses the fairly standardsystem of citing the lemma, then a bracket ], then the variant readings,then their support. Vertical bars | separate the variants. Thereal question is, why would anyone want to use the apparatus? Ifyou're going to have to deal with Tischendorf numbers anyway, whynot use Tischendorf (since it's now available online)?

The Latin apparatus records a handful of variants, but without indicationof the manuscript tradition behind them (it could be Amiatinus or itcould be most of the tradition); it's even less use than the Greekapparatus.

Westcott & Hort

Editors. Brooke Foss Westcott (1825-1901) and Fenton JohnAnthony Hort (1828-1892)

Date of Publication. The text was published in 1881 (underthe title The New Testament in the Original Greek; anIntroduction [and] Appendix, authored by Hort, appearedin 1882 (revised edition by F. C. Burkitt in 1892).

The Text. The WH text is a very strongly Alexandrian text --so much so that Hort has been accused of constructing his text simply bylooking for the reading ofCodex Vaticanus. The situation is not that simple; a better statementwould be to say that the edition used B as aproof text. Hort (who wasthe chief architect of the textual theory of the book) would followother witnesses if the internal evidence was sufficiently strong.The most noticeable instance of this is the famousWestern Non-Interpolations. Still,it is fair to say that Hort's text falls closer to B than doesany other critical edition -- and that Westcott & Hort is theonly New Testament edition which approaches the method, usedin some forms of non-Biblical criticism, of editing from aproof text.

The Apparatus. The WH edition has no true critical apparatus;not one manuscript is cited in the main body of the edition. Thereare a few variant readings in the margin; these are readings where thetwo editors disagreed on thetext or were very uncertain of the original readings. They alsohave a list of "interesting" variants. In neither apparatusdo they supply a list of witnesses. The only textual evidence they giveis in the discussion of readings in their Introduction [and]Appendix, and even these are difficult to use as manuscripts are(inevitably) cited using Tischendorf numbers.

The lack of an apparatus in WH has been criticised by some. This israther unfair in context. They worked very shortly after Tischendorfpublished his eighth edition; they had nothing to add to it. (As bothmen were caught up in academic and pastoral duties, they did not havethe leisure to go and examine manuscripts in odd places. In any case,all manuscripts known to be valuable, save B itself, had been studiedby Tischendorf.) The problem with the WH edition is not its lack ofan apparatus, but the fact that the coordinated apparatus (Tischendorf's)is now hard to find and hard to read.

The WH edition has another interesting feature: Some dozens of readingsare obelized as "primitive errors" -- i.e. passages where theoriginal reading is no longer preserved in the extant manuscripts. Westcottand Hort did not see fit, in these cases, to printconjectural emendations (they printed whatthey regarded as the oldest surviving reading), but the presentation oftheir data makes it clear that they felt it to be needed in these passages.

Summary: A Comparison of the Various Editions

This section offers various comparisons of the materials in the sundryeditions, to show the qualities of each edition. (Note: Some editions, suchas Swanson, are not included in certain of the comparisons, because theycount variants in different ways.)

For a truly detailed comparison of the major editions for the book ofColossians, see the Sample Apparatusof Colossians.

Statistic 1: Variants Per Chapter

Let's take a few selected chapters, and count how many variants are citedin each chapter by the various editions (note: variants are usually but not quitealways counted based on the way the editor of the edition divides them; the fact that SQE¹³and Huck/Greeven both show 76 variants in Matthew 10, for instance, does notmean that they have the same variants or even include similar classes of variants,just that they have about as many separate citations in the apparatus):

Sample 1: Matthew 10

* For comparison, the equivalent sections in Huck/Lietzmann show 5 variants

Sample 2: Mark 2

* For comparison, the equivalent sections in Huck/Lietzmann show 12 variants

Sample 3: John 18

Sample 4: Acts 6

Sample 5: Acts 18

Sample 6: 1 Corinthians 13

Sample 7: Colossians 2

Sample 8: James 2

Sample 9: 1 John 4

Sample 10: Revelation 8

Sample 11: Revelation 15

Possibly this is worth graphing, to give a comparison. I will offer only twographs, one for the Gospels only (so that we can include the synopses) and onefor the New Testament as a whole.

Variants in the Gospels, total for Matthew 10, Mark 2, John 18:

Here are the totals for the entire New Testament based on the chapters above:

Appendix: The Variorum Bible

This nineteenth century English Bible is never included among the list of critical editions, but arguably it should be. Certainly it would be a wonderful thing for students if such an edition were prepared based on modern materials (the original Variorum Bible dates to 1880).

The basic idea is this: The Variorum Bible is an edition of the English-language King James Version -- a true King James rendering, not one of these modern reprints which cut out the notes to the reader and the prefaces and the like. To this are added a series of reader helps such as an extensive set of cross-references and clarifying glosses. Most important from our standpoint, it has a series of textual footnotes listing variant readings and citing the supporting manuscripts and editions.

A few examples may be the best way to explain how the edition works.

In Matthew 1:10, we have three parts: The list of parallels, the text with annotations, and the list of variant reading(s). So the note ^m means that the statement that (H)ezekia(h) begot Manasse(h), and Manasseh begot Amos/Amon, and Amos/Amon begot Josia(h) is paralleled in 2 Kings 20:21 (which is the account of Hezekiah's death and Manasseh's succession) and in the genealogy in 1 Chronicles 3:13.

Below the text we see the VAR. READ orvariant reading(s). In this case, the variant concerns the text "Amos," which is marked in the text by β. The apparatus
       β So L, Mcl. We.; Amos ℵ B C Δ, Al. La. Ti. Tr. WH
means that "Amon," the reading found in the King James Version text and marked by the symbol β, is read by the manuscript L and the edition of J. B. McClellan (1875) plus the (then-incomplete) text of We(iss). The alternate reading is "Amos," which is found in ℵ B C Δ and the editions of Al(ford), La(chmann), Ti(schendorf), Tr(egelles), and W(estcott-)H(ort). It is also, of course, the reading of the UBS text.

The section on John 21:15-16 is the most complex, because it shows not only cross-references and textual variants but also "variant renderings" -- in effect, linguistic notes. So there are, e.g., no parallels cited to John 21:15, but to the section about feeding the sheep in verse 16, note ⁱ cites parallels to Acts 20:28, Hebrews 13:20, and 1 Peter 2:25, 5:2, 5:4.

The VAR. REND. or Variant Rendering in the first case (note ⁵, Rather breakfast) is simply a clarification that the Greek refers to a morning meal. Similarly, notes ⁷ and ⁹ are simply explanatory glosses, the second perhaps otiose. But notes ⁶ and ⁸ are significant, because they show (as few English versions do) that Peter and Jesus are using different verbs, αγαπαω and φιλεω. How significant this is remains the subject of debate (I personally would maintain that the distinction is deliberate and that Jesus is manipulating Peter to confess that he has gone beyond φιλια to feeling αγαπη toward Jesus), but the Variorum Bible version allows the reader to see that there is a distinction.

We have two variant readings in this section, one of which (β) applies in both verses 15 and 16; γ applied only to verse 16. Variant β concerns the name of Peter's father. "John" is read b A and C² plus the edition of McClellan (which, as you may be gathering, is very close to the Textus Receptus); all the more valuable editions (Lachman, Tregelles, Alford, Tischendorf, and Westcott-Hort) read "John" with ℵ^a B C* D. In variant γ, "sheep" (i.e. προβατα) is the reading of ℵ A D, the editions of Alford and Lachmann, and is the primary reading of Tregelles (hence TR¹); προβατια, "little sheep," is found in B C and read by Tischendorf and Westcott-Hort.

In the Corinthians verse, the note ^c tells us that the verse has a (claimed) parallel in Matthew 6:1-2; the note β tells use that the reading "to be burned" is supported by C D and is read by Lachmann, Tischendorf, Tregelles, Alford, and Scrivener; the alternative, "that I may glory" is that of ℵ A B and is the reading of Westcott and Hort's text. The note also states, somewhat misleadingly, that this is a difference of one letter in the Greek. It thus implies that the only readings are the καυθησωμαι of K Ψ and the King James version and the καυχησωμαι of ℵ A B and the Westcott-Hort and UBS texts, ignoring the fact that C D in fact read καυθησομαι, which differs by two letters from the other variant. Nonetheless the apparatus makes clear the translatable part of the difference.

The Variorum Bible cites only uncials (no minuscules and, for the most part, no versions or Fathers), and only those known as of about 1880, and mostly confines its attention to the most famous uncials, but as far as Greek evidence is concerned, this is almost as much data as Westcott & Hort had in assembling their text.

One curiosity of the Variorum Bible is that it lists the older editions in almost random order -- notice the various places where Alford is cited above. And, of course, it includes Scrivener as an "edition" as if his text were a critical text, and similarly the Gospels text of McClellan, which has little genuine value. Other editions (e.g. Ellicott) are also incomplete, and Lachmann's edition, although the first to break with the TR, was based on too few manuscripts; the only editions cited which have real value are Tregelles, Tischendorf, and Westcott-Hort. There would be great value in a new Variorum Bible based on the New Revised Standard Version, but the demand may not be there to support it.

Appendix: Latin Editions

In addition to a full set of Greek editions, a thorough student ofthe New Testament text should have access to a variety of Latin editions.We will not dwell at length on the various Latin editions, but thefollowing section supplies brief notes.

Observe that only editions with an apparatus are listed.So, for example, the Latin text of Bover, which is the Vulgatewithout apparatus, is ignored. Similarly the widely availableNeo-Vulgate, published by the Catholic Church in 1979, is not listed;although it is a modern edition with a modern text, it is not somethingone can use for text-critical work, having been conformed largely tothe Greek and usually published without manuscript variants.

Merk. (For publication data, see the entry on GreekMerk). This is in many ways the handiest ofthe Latin editions, as it combines Greek and Latin editions sideby side, with a critical apparatus of each. The Latin text isthe Clementine Vulgate, but the apparatus (quite full for a manualedition) makes it easy to ascertain which variants are older. Morethan three dozen Vulgate witnesses are cited in total, with usuallyseveral dozen in each book; in addition, the Old Latin codices arecited heavily.

Unfortunately, the result is not as accurate as might be hoped.Tests against Tischendorf and the smaller WW edition seem toindicate a high rate of errors, at least for am and ful. If exactknowledge of the readings of these manuscripts is for some reasonessential, the student is advised to rely on other sources ifpossible.

Nestle. This exists both as a standalone edition and asa Greek/Latin diglot; I've used the diglot. The scope of the editionis extremely limited: The text is the Clementine Vulgate, and theonly variants noted are those in Amiatinus (A), Fuldensis (F),and editions such as the Sixtine and Wordsworth-White editions.In addition, the presentation is such that it is often nearlyimpossible to determine which just which manuscripts supportwhich readings. As a parallel to Greek Nestle, Latin Nestlehas some slight value (mostly because the parallels line upnicely). It is not, in itself, a particularly useful edition,either in text or apparatus.

Note that this should not be confused with the more recentNestle-Aland Greek-Latin diglot, which uses the Neo Vulgate; thisis of no use for textual critics although it might be a nice cribfor Catholics. For the limitationsof the Neo Vulgate as a critical tool, see the entry on theVulgate.

Tischendorf. Tischendorf published Latin editions (whatdidn't he publish?), but this is a reference to the eighthedition of his Greek New Testament. This, of course,lacks a Latin text, but if you are using the Latin solely forpurposes of examining the Greek, Tischendorf's edition ismore useful than several of the other editions here. Tischendorfcites the Clementine Vulgate (vg^cle) and four manuscripts consistently:am(iatinus), demid(ovianus), fu(ldensis) and tol(etanus), with theirconsensus being noted simply as vg. He alsocites others, such as harl(eianus), occasionally. It's only ahandful of manuscripts, but at least you know exactly what youare getting.

The Vetus Latina. An immensely valuable, immenselycomplicated, immensely immense edition which is still ongoing. To explainit really requires reproducing several pages; it's too complicated tosimply describe. It reproduces the text in a parallel-lines formatsomewhat like Swanson, except that the linesare textual groups, not individual manuscripts. At the top ofeach page is the presumed Greek original, then the various Latintext-types (typically five or so, although not all of these typesrepresent actual manuscripts; some are reconstructed from fatherssuch as Tertullian or Cyprian). Then comes the apparatus, whichtakes an even higher fraction of the page than Tischendorf and makesevery other critical apparatus I've seen seem straightforward. Theamount of patristic information is very high. Sadly, instead of usingstraightforward sigla, it uses sigla of its own devising (Beuronnumbers, an elaborate set of abbreviations for the Fathers). And,because it's so big, it's not very portable. If you need to knoweverything about the Old Latin (at least for the books which havebeen published so far), this is a terrific resource. But you'll haveto use it in a library that can afford to stock it....

Weber or Gryson/Weber(the Stuttgart Vulgate). The vg^stof the Nestle editions. In some ways, the best of the hand editions;it is the only edition other than Wordsworth-White (on which itis significantly dependent) to have a critical text, and the onlyone other than Merk to have a real apparatus with a significantselection of witnesses. Plus, it notes the exact extent of allthe manuscripts cited. And, unlike Merk, the apparatus is generallyregarded as accurate. Sadly, it has two drawbacks: Not enoughvariants, and not enough range of witnesses. To demonstrate thepoint about variants, we look at 1 Thessalonians. The Stuttgartedition has, by my casual count, 88 variants, often of veryslight scope. This is twice the count of the lesserWordsworth-White -- but Merk has 104 variants, often coveringmore text, in this book. Thus, as with the Greek, one reallyshould have two hand editions. For the Greek, it's Nestle foraccuracy and Merk for a full list of variants; on the Latinside, one should have vg^st for accuracy and Merkfor range.
In terms of the text, the Stuttgart and Wordsworth/WhiteVulgates are very similar -- much closer than even the mostsimilar editions of the Greek New Testament; based on the figurescompiled by Bonifatius Fischer, the first edition of the StuttgartVulgate has only 772 differences from Wordsworth-White.

Wordsworth-White Editio Minor. This is probablythe sort of edition that should have been used in the Nestlediglot. It is a critical text (identical in some parts to thelarger Wordsworth-White edition, though distinct in certain bookswhere the larger edition was unfinished at that time). Thecritical apparatus cites enough good manuscripts to be useful,as well as the readings of the Sixtine and Clementine editions.That's the good news. The bad news is, the manuscripts are notcited with any regularity. All variants in the editions are noted,but readings of the manuscripts only rarely. Taking as a randomexample the book of 1 Thessalonians, the edition cites a total of45 variants. Only five of these cite the manuscripts; the restcite only editions. Thus the apparatus, while generally accurate,is quite limited. It differs from the larger edition inCorinthians, Galatians, and Ephesians, since this edition wascompiled before those books were published. For the remainingbooks (later Pauline epistles, Catholics, Revelation), the EditioMaior used the test of the Editio Minor, even though itwas edited in haste; it is said that the text shows more differencesfrom the Stuttgart Vulgate in these books.

Wordsworth-White Editio Maior. Also known as theOxford Vulgate. Although nowout of date (since much of it is a century old), this remainsthe most complete critical edition. The problem is that it'sbig and expensive; even if you can afford the expense, you aren'tgoing to carry it around with you.... For the most part thetext is the sameas the Editio Minor.

Appendix: LXX (Septuagint, Greek Old Testament) Editions

Although a textual critic may not consider the Septuagint (Greek Old Testament, LXX) to be canonical, the critic nonetheless really should have a copy of the LXX at hand, simply because the New Testament quotes the Old, and the reading of the LXX may often have influenced the reading of the New Testament.

This section will not consider the problems of Old Testament Textual Criticism, which are much too large for a subsection of an article, but we can at least take a look at the materials available to a student who needs to look at the LXX text.

Rahlfs

Even today, there is only one complete critical edition of LXX: that of Alfred Rahlfs, Septuaginta, completed in 1935. In 2006, Robert Hanhart published a revised edition. This did not, however, significantly revise the text; the main effort went into the apparatus. And even that is changed very little. How little? The table below compares through the two editions, checking every hundredth page and marking the changes (shown in red).

In all, that's 22 pages tested, and 405 variants. There were only 8 changes, mostlyquite minor. The new revision changed only 2% of variants tested.

The style of the apparatus did not change at all with the new edition. For the most part, the apparatus consists of a lemma, a bracket ], and then the variants. So, for instance, the first variant in chapter 2 of Genesis (2:4) reads:

24 ο θεος M ] pr. κυριος A (in O sub ※)

So in verse four of chapter two, the text printed by Rahlfs (ο θεος) is found in the uncial M. A precedes this with κυριος (i.e. κυριος ο θεος), which is also found in the Origenic text but with κυριος in ※ (i.e. ※κυριος※ ο θεος) to indicate that it is not in the Hebrew.

This is much the most common form of variant (that is, lemma ] variant), but we also find cases where only the variant reading is given (generally when the variant is a different form of the same root as in the text). So, for instance, in Genesis 3:24 the apparatus reads

24 χερουβιν A†

That is, for the reading χερουβιμ of Rahlfs's text, A (and only A) reads χερουβιν (-ν for -μ).

There are a number of special symbols in the text. One is seen in the preceding note: the dagger †. This indicates that the witness shown is the only one known to have the reading cited (so in this case, A is the only manuscript known to read χερουβιν rather than χερουβιμ).

Also very common is the > symbol, which means "omit." So in Genesis 3:20, for instance, the text reads του ονομα της γυναικος αυτου Ζωη, and in the apparatus we find
20 αυτου > A†
That means that A (and only A, given presence of the symbol †) omits the word αυτου, meaning that A's text reads του ονομα της γυναικος Ζωη.

Also common is the symbol + for an addition, after the lemma (for an addition before the lemma, the text uses "pr."), e.g. in Genesis 7:15 the Rahfls text reads δυο δυο απο πασης σαρκος, with the apparatus reading
δυο δυο ] + αρσεν και θηλυ A†
That means that A (alone) reads δυο δυο αρσεν και θηλυ απο πασης σαρκος

As in many other editions, the "leap" symbol ⌒ is used to indicate a haplography caused by repeated letters. However, ~ does not indicate transposition; instead, "tr." is used.

Unique to LXX editions, of course, are the Origenic symbols ※ and ÷ -- the former for words that were lacking in LXX but found in the Hebrew and moved to the Greek text from Theodotian (or somewhere); the latter for words in the LXX not in the Hebrew.

The Origenic text is designated O, that of Lucian L, and the other Greek editions are α' (Aquila), σ' (Symmachus), θ' (Theodotian). There are also a few uses of ε' for the Quinta column of the Hexapla. In Rahlfs, the symbol ℳ does not mean the Majority Text (of either the Old or New Testament) but the Massoretic Text of the Hebrew. In the Prophets, there is also a "Catena" text, C, the nature of which Rahlfs does not really identify, although it contains in one place or another the manuscripts 87 91 97 490.

The Göttingen Edition

Since this is a series of volumes, by multiple editors, the style of the various apparatus varies somewhat. But for the most part the format is similar to that of Rahlfs. For instance, it uses the same lemma ] variant format. And it uses many of the same symbols, e.g. + for add, > for omit, ⌒ for haplography. But there are many additional symbols, e.g. "La" for the Latin -- and La^M for a particular manuscript (in this case, the Speculum).

The fact that many more witnesses are cited also forces a change. Rahlfs rarely cites more than four witnesses, most of them uncials. So the symbol "BA" of Rahlfs is unambiguous. But Göttingen often links manuscripts in groups. So, e.g., the first verse of 1 Maccabees (Volume IX, fascicle 1, edited by Werner Kappler, 1967) begins

Καὶ ἐγένετο μετὰ τὸ πατάξαι ’Αλέξανδρον τὸν Φιλίππου

The first variant reads

1 om. τὸν 1⁰ L-19-93 311

This means that the reading without τὸν is supported by the Lucianic witnesses 64, 236, 381, 534, 728 (the pure Lucian group L) plus the weaker Lucianic witnesses 19 and 93, plus the non-Lucianic witness 311. Most books in the Göttingen series will have similar groups of related witnesses. E.g. the first line of the Theodotian text of Daniel is

’Εν ἔτει τρίτῳ τῆς βασιλείας Ιωακιμ βασιλέως Ιουδα ἦλθε

And the second variant is

(1¹) om. τῆς C 534

So the "Catena" manuscripts, 87 91 490, omit τῆς, as does 534. The Catena group is a complex group, the core manuscripts being 87 91 490, with a weaker group, c, including 49 90 405 764. The combined group, C', is C + c, i.e. 49 87 90 91 405 490 764.

Citations of groups can be quite intricate, because groups can have subgroups, and members can defect. For example, in verse 6 of the story of Susanna there is a variant

κρινόμενοι ] συναγομενοι L' ^-36

This means that the reading συναγομενοι is that of all members of the Lucianic recension L' (22-36-48-51-96-231-763) except 36.

Ezekiel gives an even more complicated example. One of the most interesting aspects of the text of Ezekiel is the rendering of the divine name and title. The Hebrew text is very fond of "the Lord God;" instead of adonai YHWH the Greek texts often prefer a single title, with 967 almost never having having two words, and B and the Coptic versions having two only rarely, but all other texts frequently having two words -- but varying between reading κυριος κυριος and αδωναι κυριος. So here is a variant from Ezekiel 7:9:

κύριος B A' 87-534-710 Co Arab] + κυριος O^-Q C ' ^-87-130-233-86-403' 106 Arm; pr. αδωναι rel. = 𝔐↓

Note in particular the symbol ', which generally refers to a group. (To add to the fun, there are actually two sigla for groups, both the straight mark ' and the curled quote ’; for the Lucianic group, for instance, one may have L, L', L’, and L'’).

So the apparatus above is to be understood that the reading of the text, τάδε λέγε κύριος, is read by B, by the A' group (A and 26), 87, 534, 710, the Coptic versions (bo and sa), and the Arabic version.

The first alternative reading, τάδε λέγε κύριος κύριος, is found in the Origenic group less Q (O usually consists of Q, 88, and the Harklean Syriac, so O^-Q means 88 sy^hark), the larger Catena group C' less 87 (C is 87, 91, 490, and the expanded group C' includes also 49 90 764, so C ' ^-87 consists of 49 90 91 490 764), the individual manuscripts 130 and 233 (both associated with C’), the manuscript 86, the two manuscripts in group 403 (i.e. 403 613), the manuscript 106, and the Armenian version. So the full list of manuscripts represented by O^-Q C ' ^-87-130-233-86-403' 106 Arm are:
49 86 88 90 91 106 130 233 403 490 613 764 sy^hark arm.

The second alternative, pr. αδωναι rel. = 𝔐↓, means that the reading αδωναι κύριος, which agrees with the Hebrew text (𝔐), is supported by the remaining witnesses (rel.). This includes the one remaining uncited uncial, V, and the sundry other minuscules, 22 36 46 48 51 62 96 147 198 231 239 306 311 380 407 410 449 544 611 770.

Newer volumes of the Göttingen series, and revised editions, have become increasingly complex. Increasingly they contain not one apparatus but two, the second being devoted to Hexaplairic variants. Confining Hexaplairic variants to this second apparatus makes sense in a way, since they generally aren't in the LXX Greek texts, but of course it also means the user is forced to consult two apparatus. So, for example, in Isaiah 53:1 the last two words are τίνι ἀπεκαλύφθη. The main apparatus has a variant here,

53 1 τίνι] pr. (※) επι 88=𝔐↓

while the second apparatus reads

53 1 τίνι] θ' επι (α'θ' ※ επι Q, Syh sub α'σ') τινα 86

Thus both apparatus refer to the same variant, although the former refers more to manuscripts and the latter to translations. To be sure, this seemingly inconvenient arrangement makes little difference to the textual critic, who isn't much concerned with α' θ' σ' -- we know that they almost always agree with the extant Hebrew text.

In addition to the actual volumes of criticism, many of the Göttingen volumes come with associated studies of the actual text of LXX -- e.g. for Jeremiah Joseph Ziegler not only prepared the critical text and apparatus but also published Beiträge zur Ieremias-Septuaginta discussing a number of practical problems in the text. There are two, perhaps three, drawbacks. One is that most of the volumes are in German -- obviously a language that any good textual critic should know, but perhaps English-speaking readers will miss some of the subtleties. Second is that it's not actually a textual commentary; it's a discussion of the tendencies of the translation, e.g. how it handles articles and such. And third, and perhaps worst, there is no index of passages discussed. There is an index of words and names (surprisingly short), but that isn't the same thing.

It will be evident that, although the Göttingen edition is a major work that is incredibly useful for critics of both the Hebrew Bible and the New Testament, there is still a great deal that could be done in editing the LXX. And, of course, until that is done, we can't really edit the Hebrew either....

Brooke and McLean / Brooke, McLean, and Thackeray (The Larger Cambridge Septuagint)

Like the Göttingen edition, this was released in multiple volumes, with multiple fascicles per volume, but was never completed. It is now out of copyright, and the volumes are being scanned and re-released by the sundry so-called publishers which produce such books. A great deal of caution is urged if you are considering buying such editions; even if they are scanned from a good copy (many are not), the scans are usually done very quickly and the printing is sloppy. I often find myself having to turn to Rahlfs to try to understand what my cheap reprints actually say.

There is, however, also a four-volume reprint by Cambridge University Press, and this is properly printed and entirely legible. It is also well worth having. The completed volumes as reprinted include 1: Genesis-Leviticus, 2: Numbers-Deuteronomy-Joshua-Judges-Ruth, 3: Samuel-Kings-Chronicles-1 Esdras-2 Esdras (Ezra/Nehemiah), 4: Esther, Judith, Tobit.

It should perhaps be noted that the Cambridge LXX does not do as Rahlfs does and sometimes print parallel texts of Joshua from A and B. The main text is always the leading uncial.

The Cambridge edition is constructed in a very different manner from Göttingen. Göttingen is a critical edition; Cambridge is a diplomatic edition, with a text essentially that of Vaticanus as far as that manuscript is extant.

This results in a strange multi-part apparatus in which the first part is the detailed variants in the most important texts, then a second, much fuller, apparatus, where all the variants are listed. For example, the first line of the text in ΕΣΔΡΑΣ Β (Ezra/Nehemiah) reads (omitting accents and breathings)

B    ΚΑΙ εν τω πρωτω ετει Κυρου του βασιλεως Περσων, του τελεσθηναι απο στοματος

And the apparatus for this text runs

11 τελεσθηναι]+λογον B^ab

Ab-ehj-npqtwye₂𝔈

I 1 πρωτω ετει ] ετει τω πρωτω be₂ | τοι 1° BAbke₂ ] om rell Eus | τελεσθηναι B* ] τελεσθηναι ρημα κυ be₂ + λογον B^abhk + λογον κυ A rell 𝔈 Eus | απο ] δια be₂

It is a strange apparatus, because the first part offers very little that is not available in the third part as well. The "B" at the beginning of the text says that the running text here is taken from B (Vaticanus); to repeat, it is not a critical text. (B is the main text for most parts of the edition, but where it does not exist, another manuscript is used, e.g. A for most of Genesis and some other manuscript where both A and B are defective.) So we see that the apparatus here shows a correction found in B.

But the first apparatus isn't just B; it also includes the major uncials, A and (where extant) א (cited as S); also occasional others such as F. So the first apparatus of Brooke and McLean is almost the same as the apparatus of Rahlfs (except that Rahlfs published a critical text).

The second line is simply a list of additional witnesses. Thus, for this passage, in addition to B, the witnesses cited are the uncial A and a selection of minuscules, cited as b, c, d, e, h, j, k, l, n, p, q, t, w, y, and e₂, plus the Ethiopic version 𝔈.

The third apparatus is the real one, and it's pretty standard: printed text as a lemma, then a bracket ], then the variants. Note that, in some cases, the same variants occur in the first and last apparatus -- in this case, τελεσθηναι]+λογον B^ab in the first apparatus and τελεσθηναι B* ] τελεσθηναι ρημα κυ be₂ + λογον B^abhk + λογον κυ A rell 𝔈 Eus in the second. This is redundant, but it isn't really a difficulty; you can get all the information you need from the second apparatus.

The real complication lies in the symbols used for the manuscripts. The use of letters makes for a very compact apparatus -- the symbols are mostly one letter, and no space is needed between them. So the "Lucianic" group boc₂e₂ can be written out in 6 characters. This compares to 13 for the Rahlfs numbers 108-19-127-93. But it means that you need a conversion tool between Brooke/McLean letters and the official manuscript list (there is such a tool in the introduction to the various volumes, but not in some of the individual fascicles). The words "a pain" spring to mind. Plus the edition typically cites only about half as many manuscripts as Göttingen. Thus, as Göttingen volumes appear, they will surely supplant Brooke and McLean. Or would, if Göttingen would get reasonable and charge something semi-affordable for the volumes, anyway. In the meantime, perhaps the best thing to do is to make up a translation card to equate Brooke/McLean letters with Rahlfs numbers.

It should be noted that, again unlike Göttingen, Brooke/McLean make no attempt at all to classify their witnesses. Thus, in the above list, be₂ are the Lucianic text. It was my strong feeling that there was also a group containing dpqt (i.e. 106 107 120 134) and perhaps one or two others such as z, which I now know is Hanhart's a in 1 Esdras etc., but which has no name that I know of (it's close to the Complutensian Polyglot). In any case, these witnesses are not cited as a group in Brooke/McLean; all witnesses are cited in alphabetical order. So you have to sort through the apparatus first to seek groups (as I did with dpqt) and then to determine the group reading.

A Comparison of the LXX Editions

In light of what was said above, it is perhaps interesting to compare the three LXX editions in detail. This is particularly true since Brooke and McLean is now once again available in a proper edition -- and the cost for the entire thing is roughly the same as two fascicles of the Göttingen edition. (If you think I'm saying Göttingen is overpriced -- well, I'd love to be buying the thing, but I can't afford it! I have half a dozen volumes, but often they are first editions of volumes for which a second edition is available.) So what extras do you get with the Göttingen edition?

As a reasonably fair standard of comparison, let's take the book of 1 Esdras -- a book where Hebrew influence is remote. The table below shows the witnesses cited by the two major editions. (Rahlfs, of course, cites only B, A, and ℵ consistently; it is a critical text but not a true reference edition.)

Hanhart's list of manuscripts also includes 52 68 122 125 610, but these are not regularly cited.

Thus Hanhart adds as regular witnesses 46, 74, 98, 130, 236, 314, 340, 381, 728, 731, 745, 762, and 841; he drops 52 and 93. This sounds like a major addition, but let's break this down by groups (excluding the fragment 841).
Of the pair 58', B&M have just 58, Hanhart has 58 340
Of the pair 119', B&M have just 119, Hanhart has 119 745
Of the group a, B&M have 106 107 120 121 134 370, to which Hanhart adds 74 130 236 314 762
Of the group b, B&M have 64 243 248, to which Hanhart adds 46 98 381 728 731
Every one of Hanhart's additions, except 841, belongs to 58', 119', a, or b. Hanhart has dramatically expanded the groups, especially b, but he has not found any new groups. The only major additions to his apparatus are 340 (to supplement 58) and 745 (to supplement 119), since they give us two witnesses to these types; in exchange, he has dropped 93 entirely, and 52 is not cited regularly. Thus there is still some slight use for B&M, and although Hanhart has dramatically increased the number of witnesses, he has increased the number of groups not at all.

Let's do an even more detailed comparison of the book of Esther. This was one of the last volumes prepared by Brooke and McLean, and Hanhart's Göttingen edition was published after the series was well established, so both series had had the bugs worked out by then.

Both editions have both texts of Esther -- the primary LXX edition, but also the much different rendering found in 19 93 108 319 392, which Brooke and McLean label the "A" text and Hanhart refers to as L because the witnesses to it are all Lucianic (although there is no reason to think that the A/L text is actually derived from Lucian). In Brooke and McLean, the two texts are printed as separate books; Göttingen conveniently prints them in parallel, facilitating direct comparison.

What is really interesting, to me, is that every witness cited by Brooke and McLean is also cited by Göttingen. And Göttingen adds more witnesses, and classifies them, so you get more information in less space. But is it a lot more information? The table below summarizes the manuscripts cited by each, with Hanhart's classification.

Thus it will be seen that while Göttingen regularly cites eleven witnesses not cited in BM (specifically 71 74 98 236 314 318 370 381 392 728 762), all of these except 318 belong with Hanhart's groups a, b, or L, all of which are adequately represented in the BM apparatus (a by 76 106 107 120 130, b by 46 64 243 248, L by 19 93 108 319). Göttingen is better than BM in terms of witnesses, but only very slightly so.

To put it another way, let's this time list the witnesses cited by Hanhart, in their groups, and list the ones cited by B&M as well. A witness cited by both is shown in bold, with its B&M symbol for the minuscules.

• B
• ℵ/S
• N/V
• 967
• O = 58[k] 93[z] 583[f]
  • A 311[r]
• a = 71 74 76[g] 106[p] 107[d] 120[q] 130[s] 236 314 370 762
• b = 46[e] 64[c] 98 243[j] 248[w] 381 728 731
• 55[h]
• 108[a]
• 249[v] 670
• 318
• 392
• 542[u]

Thus, here again, B&M have representatives of every type identified by Hanhart, and the only mixed manuscripts they lack are 318 and 392. The one major absence from B&M is, of course, 967, the vital and early papyrus copy. For Esther, as for Ezekiel and Daniel, the Göttingen edition is probably mandatory just for 967. But the general picture is clear: unless there is an important discovery since the time of B&M, the Cambridge edition is sufficient if all you need is a critical apparatus. If you need a critical text, of course, Göttingen is your only choice.

Of course, it is not just the list of witnesses that is different; the text differs also, and the organization of the apparatus. BM simply print the text of B; Göttingen has a critical text. The table below compares the two (minus accents, which are too hard in HTML), with the differences between the two texts shown in bold:

In essence, there are three differences between content (as opposed to the form) of the two apparatus:

1. Hanhart cites 967 more completely (BM has some readings of 967, but not a full set)
2. Hanhart cites a number of additional minuscules, but mostly close relatives of manuscripts cited also by BM
3. They differ in their handling of itacisms: BM use the reading of B as given; Hanhart standardizes the orthography

Hanhart also groups manuscripts and gives more explanations for why particular readings arose. These are real advantages. But we see that BM actually offers some data not in Hanhart, and other than 967, its selection of witnesses is not materially inferior to Hanhart's. For books where there have been no major recent discoveries, the Göttingen edition's main advantage is that it has a critical text. Whether this justifies the additional cost is, again, something the reader must decide.

The situation is somewhat different in other books -- the Göttingen apparatus for the Pentateuch is rather fuller than in the later historical books. Here again we can do a comparison. Let's take Leviticus as a sample, since it's in the middle of the Pentateuch (sparing us the defects that we often find at the beginning or end of manuscripts):

So if we take the various groups cited by Wevers, we find that Brooke & McLean cite the following:

Mixed manuscripts cited by Wevers but not by B&M are: 18, 319, 416, 424, 642, 646, 799

Cited by B&M but not Wevers: 61[d₂]

Thus, as in 1 Maccabees, B&M have at least one member of every group identified by Wevers, although they lack the recent papyri and frequently have only one manuscript of each type. The Wevers apparatus has more information, but not tremendously more.