An Introduction to
New Testament Textual Criticism

Adam scriveyn, if ever it thee bifalle
Boece or Troylus for to wryten newe,
Under thy long lokkes thou most have the scalle,
But after my makyng thow wryte more trewe;
So ofte adaye I mot thy werk renewe,
It to correcte and eke to rubbe and scrape,
And al is thorough thy negligence and rape.

          -- Chaucer

Contents: Introduction * Types of Manuscripts * Printed Versions of the New Testament * The Practice of Textual Criticism * List of New Testament Manuscripts * Final Examples * Final Notes: Critical Editions, Science, and Faith * Appendix: The Text of Chaucer's Address to his Scribe *


Chances are that you've played the game "Telephone" some timein your life. "Telephone" is the game in which a group of peoplegather around in a circle. One person thinks up a message, and whispersit to the next person, who whispers it to the next person, and so on aroundthe circle, until you reach the end and the final person repeats the messagealoud. The first person then states the original message.

The two sentences often cannot be recognized as related.

Even if you haven't played "Telephone," you must have read abook or a magazine which was filled with typographical errors. And that'sin a case where the typesetter has the author's original manuscript beforehim, and professional proofreaders were engaged to correct errors.

Now imagine what happens when a document is copied, by hand, tens ofthousands of times, long after the original manuscript has been destroyed.Imagine it being copied by barely literate scribes standing (not sitting,standing) at cold desks in bad light for hours on end, trying toread some other scribe's barely legible handwriting.

Imagine trying to do that when the words are written in all upper-caseletters, with no spaces between words, and you're writing on poor-qualitypaper with a scratchy reed pen using ink you made yourself.

Because that's what happened with all ancient books, and with the NewTestament in particular. Not all scribes were as bad as the secretaryChaucer poked such fun at in the quote above, but none were perfect -- andfew had the New Testament authors looking over their shoulders to makecorrections.

After a few centuries of that, it's easy to imagine that the text of theNew Testament would no longer bear any relationship to the original. Humanbeings just aren't equipped to be exact copyists. And the more human beingsinvolved in the process, the worse the situation becomes.

Fortunately, the situation is not as grim as the above picture wouldsuggest. Despite all those incompetent scribes making all those incompetentcopies, the text of the New Testament is in relatively good shape. The factthat copies were being made constantly, by intent scribes under the supervisionof careful proofreaders, meant that the text stayed fairly fixed. It is estimatedthat seven-eighths of the New Testament text is certain -- all the majormanuscripts agree, and scholars are satisfied that their agreement is correct.Most of the rest is tolerably certain -- we probably know the original reading,and even if we aren't sure, the variation does not significantly affect thesense of the passage. For a work so old, and existing in so many copies,this fact is at once amazing and comforting.

Still, there are variations in the manuscripts of the New Testament,and some of them are important. It is rare for such variants to affect afundamental Christian doctrine, but they certainly can affect the course ofour theological arguments. And in any case, we would like the most accuratetext of the New Testament possible.

That is the purpose of textual criticism: Working with the materials available,to reconstruct the original text of an ancient document with as much accuracy aspossible. It's not always an easy job, and scholars do sometimes disagree. Butwe will try to outline some of the methods of New Testament textual criticismin this article, so that you too can understand the differences between Bibles,and all those odd little footnotes that read something like "Other ancientauthorities read...."

 (Footnote: The description at left is my definitionof textual criticism: Determining,as best we can, the original text of the document. In recent years, with thispost-modern tendency to think that methods matter more than results, there hasbeen a certain tendency to argue that the phases in the history of the documentare the point of textual criticism. I'll say flat-out that, as far as I'mconcerned, this is pure bunk. Such historical criticism is useful andinteresting -- but it's not textual criticism, which should always have itseyes fixed firmly and solely on the original text. Only that and nothingmore.)

Types of Manuscripts

If the task of reconstructing the text of the New Testament may becompared to a detective story, then our "witnesses" are theancient manuscripts. Manuscripts fall into three basic categories:Greek manuscripts, ancient translations (generally called"versions"), and quotations in ancient authors.

The analogy to witnesses in court is apt. Some of our witnessesare fragmentary; they preserve only small parts of the story (thoughoften important parts). Others are fairly complete, but are notvery reliable. Each witness has its own peculiarities, which must betaken into account as we decide between readings. As one scholar putit, to be a successful textual critic, you must "know thepersonality of your witnesses."

Of the three classes of witnesses mentioned -- Greek manuscripts,versions, and quotations -- the most important are the manuscripts, since they preservethe wording in the original language and in the original order. (Exception:the lectionary manuscripts, of course, do not preserve the order.) Theoldest Greek manuscripts date from the second century; from that timeon, the number of manuscripts grows ever greater until the thirteenthcentury, then comes to a fairly abrupt halt at the end of the fifteenthcentury (when first the fall of Constantinople in 1453 and then theprinting of the first Greek Testament in 1516 reduced the need formanuscript copies). Unlike modern books, which are relatively standardized,manuscripts take various forms, and the formof the manuscript (arrangement of columns and lines, style of script used,etc.) can sometimes influence the sorts of readings wefind in it.

The books of the New Testament were almost certainly originally writtenon scrolls. We see evidence of this in the texts of Matthew and Luke, both ofwhich drastically compressed the material in Mark in order to maketheir books fit on the largest possible scroll. These scrolls wereprobably of papyrus, which was the cheapest and most important writingmaterial in the ancient world.

But the urge to collect the writings that eventually made up theNew Testament must have been very strong. It is generally believedthat collections of Paul's writings were in existence by 100 C.E.if not earlier. This posed a problem: A collection containing thewritings of Paul, or the four gospels, was far too long for a singlescroll. A complete New Testament would have been even more impossible.

The solution was the form of book known as thecodex. This is, in fact, what modernsthink of as a "book." Instead of sheets being placed sideto side to produce a immensely long single "page," they werefolded over each other, permitting books of any length --and, not insignificantly, saving expensive writing material (since codicescould be written on both sides). The Christian church seems to haveadopted codices with great enthusiasm; over 99% of known New Testamentmanuscripts are in codex form, and the few minor exceptions were already-writtenscrolls that Christians salvaged and reused.

The earliest manuscripts rarely if ever contained complete New Testaments(for one thing, the canon of the New Testament was not finally settled untilabout the fourth century). Most manuscripts contained only one section --Gospels, Paul, Acts and Catholic Epistles, Revelation. In addition, earlymanuscripts are often incomplete -- pages have been lost, or parts of pageshave become decayed or torn or simply illegible.

Part of the problem is the writing material. Our earliest survivingmanuscripts are written on papyrus, which grows brittle with age and canbe ruined by damp. Only in Egypt has the dry climate allowed a few papyrusmanuscripts to endure, and even these are often damaged. (A few papyri buried, e.g.,in the eruption of Mount Vesuvius are only a minor exception; these were notChristian writings. These papyri were in fact numerous, but badly damaged,very difficult to read -- very many were destroyed in the attempt to openthem -- and not very valuable) With the exceptionof the papyrus nown as P72(which contains the books of 1 and 2 Peter andJude in their entirety, along with non-scriptural writings), not one papyruscontains the complete text of any book.

Papyrus was not the only writing material used in the ancient world,however. Parchment -- the carefully prepared skins of animals -- wasalso available. It was, in fact, a better material, at once stronger,smoother (which made attractive writing easier), and more durable. But itwas also generally much more expensive. Papyrus is a plant, which growsquickly in Egypt. (There are some reports of papyrus being grown in Italy,but these are uncertain. It is likely that some papyrus sheets were finishedin Rome, but this was probably based on imported materials. David Diringer,in The Book Before Printing, notes that the Latin papyri at Herculaneumwere more brittle than those written in Greek, and speculates that this isdue to the use of remanufactured papyri. Another possibility is that Italianpapyrus, if it actually existed, was inferior.)

 (Image of P13)
A papyrus manuscript: a portion of one section ofP13, containing part ofHebrews. Note the uncial (all-uppercase) letters and the lack ofspaces between words, as well as the damage to the 1700-year-old material.

The early church was poor, and needed many books. Parchment was probablygenerally beyond its means. It was not until the church became legalin the reign of Constantine that parchment came to be widely used forchurch writings. Parchment and papyrus continued to be used side by side for manycenturies. The heyday of papyrus manuscripts was the third and fourthcenturies, but we have papyri from as late as the eighth century (by whichtime the Islamic conquest had largely suppressed Greek-speaking Christianityin Egypt). Parchment manuscripts first appear in the third century, andbecome common in the fourth; they remained dominant until the early partof the second millennium, when paper began to be used.

Both the papyri and the early parchments were written in a style ofwriting known as "uncial" (also sometimes called "majuscule").This is, more or less, what we wouldcall "upper-case letters." The letters were large, and thevarious letterforms were not connected. For the most part, theletters fall between two lines. In the earliest manuscripts,there were no accents, no breathings, no punctuation, and no spacesbetween words. (This doubtless led to certain errors, as scribesmisread undivided words and sentences. So, for example, in uncialscript it would be easy to confuse ΑΛΛΑand ΑΜΑ.)

As the centuries passed, uncials grew more elaborate, with the letterssprouting serifs and other slow-to-write forms (the reader isinvited to examine the chart of uncial letterforms).Manuscripts grew easier to read as scribesgradually started to add breathings, punctuation, etc., but these were slowto write and took up a great deal of writing material. What was needed was a cursivehand -- but it was not until the ninth century that an appropriate scriptwas developed (there were earlier Greek cursive hands, but they were notused for Biblical manuscripts, probably because they were not consideredelegant enough). With the developmentof this script began the "age of the minuscule" -- "minuscule"being the name given to both this cursive style and the manuscripts writtenin the style. The first minuscules were written in the ninth century, andby the end of the tenth century they had essentially driven the uncials outof use (uncials continued to be used in lectionaries for a few more years,but from the thirteenth century on we have no examples of the type exceptin a few marginal notes).

(Image of 1739)
One of the best-known minuscule manuscripts: 1739, of the tenthcentury, with the run of the text in minuscule script and a colophonat the bottom in an uncial hand.

(It is interesting to note that other languages followed a similar history.Early Latin manuscripts are written in Latin uncials, but as time passed,minuscules came into use. Unlike Greek minuscules, however, where the unityof the Byzantine Empire meant that the same general style was adopted throughout,different centers seem to have developed different minuscule styles; we seegreat variety in eighth and ninth century Latin manuscripts, until the CarolingianMinuscle became dominant. Though the later history of the types does diverge;modern Greek print and handwriting bear almost no similarity to late Greekminuscules. I've seen it claimed that modern print -- which gave rise to modernRoman-letter handwriting -- is a form of Carolingian minuscule. This isn't exactly true --most moderns would have real trouble reading Carolingian minuscule -- but they arerelated and fairly similar.)

All told, there are somewhat more than 3000 continuous-text Greek manuscriptsof the New Testament. Between 85% and 90% of these are in minuscule script;the remaining 10-15% (uncials and papyri) are in uncial script.

It will be evident that some system is needed to keep track of all theseassorted manuscripts. The present system, although somewhat imperfect, wasadopted after centuries of trial and error and, frankly, confusion. In it,continuous-text manuscripts are divided into three classes: Papyri, Uncials,and Minuscules.

Papyri are written on (guess what) papyrus, in uncial script.As noted, the earliest papyri date from the second century, and the lastdate from the eighth. Papyri are designated by the letter P (often ina blackletter script) and a superscript letter. ThusP13,P45,P46,P47, P66, P72,P74,and P75are among the most important papyri. As new papyri continueto be discovered, new numbers are added to the series (thus the lower the number,the earlier a papyrus was probably found). As of this writing, the number ofknown papyri is about one hundred. (Note that some papyri have more than onenumber, as different portions came to light at different times. So theactual number of manuscripts in a class will generally be slightlyless than the nominal number.)

The second class of Greek manuscripts are the uncials. In a way, it isunfortunate that uncials are distinguished from papyri, since they arewritten in the same script and there is no great difference in age -- theoldest uncials date from the third century; they continued to be writtenuntil the tenth/eleventh century. The difference lies only in the writingmaterial: Uncials are written on parchment, papyri on papyrus. (It is truethat most papyri are older than most uncials -- the bulk of survivingpapyri are from the third and fourth centuries, while uncials do notbecome common until the fourth century and the bulk of the survivingcopies date from the sixth through ninth centuries. But it is importantto remember that some of the best uncials are as old as or older thanmany of the papyri.)

Uncials were originally designated by letters, i.e. A, B, C, D. As thenumber of known uncials increased (the nominal number now stands at slightly overthree hundred, but -- as with the papyri -- the same manuscript sometimes has multipledesignations, meaning that the actual number is on the order of 270),it became necessary to use Greek letters, then Hebrew letters. Eventually scholarsgave up and took to using a numbering scheme, with each uncial's number precededby a zero. Thus the manuscript A is now also called 02, B is 03, etc. However,most of the best-known manuscripts are still known by the letter designation theyonce had.

Beyond these are the minuscules, recognized by the script in which theyare written (since they can be on either parchment or paper). The earliestminuscules date from the ninth century (overlapping the last uncials), andcontinued to be written up to, and even after, the appearance of the firstprinted New Testament in 1516. For the most part, minuscules are markednot only by their script but by the presence of accents, breathings, wordspacing, paragraphs, punctuation -- all the things whose absencemade the early uncials so hard to read. Minuscules are given simple numbers,from 1 on up to the current total of about 2850.

There is a fourth class of Greek manuscripts, thelectionaries, whichof course contain the lessons read in the Greek church in the order theyare read. Lectionaries are quite numerous (about 2300 are now known), butmost of them are late and fairly standardized. They may be written onparchment or paper, in uncial or minuscule script. Lectionaries aredesignated by a script letter l followed by a number(e.g. l547is the relatively well-known "Ferrar Lectionary," so-calledbecause its text resembles that found in the group of manuscriptscalled Family 13 or the "Ferrar Group"). To this point, thelectionaries have not been very carefully studied, and they are rarely used intextual criticism. Since this article is intended to be short, we will sayno more about them.

A list of some of the more important New Testamentmanuscripts is found elsewhere in this document.

In addition to the Greek manuscripts, we have the testimony of the"versions" -- the ancient translationsof the Greek New Testament. These are highly valuable in some ways --they are usually early (the oldest Latin, Syriac, and Coptic versionsdate from the second to fourth centuries, and the Armenian probably tothe fifth), and we know what part of the world they come from. But theyalso have drawbacks: No translation, even if precise and literal (andnot all these translations are) can exactly render the wording of theGreek original. Also, the versions have a textual history of their own,which means we have to reconstruct their readings. Finally, it isworth remembering that, although a version may exist in thousands ofcopies, it is usually translated from no more than a handful of Greekoriginals. Thus the versions are very important for determining thehistory of a variant reading, but sometimes less useful for determiningthe original text.

The final class of witnesses normally mentioned is the testimony of quotations in theChurch Fathers. This is an amazingly richresource -- many, many authors quoted the New Testament over thecenturies. And we usually know with fair precision both the date of the quotation andthe place where the author wrote. Unfortunately,the authors often cited loosely, adding, paraphrasing, or omittingas they saw fit; they did not cite in order, they rarely citedlong passages; and in any case, their works, just like the manuscriptsthemselves, have been subject to copying and corruption over theyears. Hence the Fathers, like the versions, are best used toestablish the history of the text.

A fourth class of witnesses, not normally mentioned in New Testamentcriticism because they have so small a role, areimitations.

Printed Versions of the New Testament

The first complete New Testament to be published was the edition ofDesiderius Erasmus,now known as the Textus Receptus ("Thetext received [by all]" -- a phrase derived from an advertising blurbin a later edition!). This was published, with great haste and on the basisof only a handful of late manuscripts, in 1516 (the printer wanted to beata rival edition onto the market, and so hurried Erasmus and then pushedthe edition through the press without proper oversight). Yet it formed the basisfor all Greek editions for over three centuries; Luther's German translationand the English King James Version (as well as most of the English editionspreceding the KJV) were translated from editions of the Textus Receptus.

The Textus Receptus had a text that was fairly typical of the manuscriptsof its time, and for the first century or so of its existence no one worriedmuch about its text. But in the early seventeenth century the Codex Alexandrinusarrived in England from the Middle East. This produced a sensation, since itwas a very old (fifth century) manuscript which often disagreed violentlywith the Textus Receptus. Suddenly scholars began to realize that therewere different forms of the New Testament text.

It was not until 1831, however, thatKarl Lachmann (1793-1851)published the first Greek testament not based on the Textus Receptus.Lachmann's edition differed from the Textus Receptus at thousandsof points, some of them significant. His text came under immediateand intense attack. Yet almost every Greek edition since Lachmann'stime has been closer to his text than the Textus Receptus. Thereason was that textual criticism was beginning to come into itsown, and the Textus Receptus no longer appeared adequate.

The Practice of Textual Criticism

But why was the Textus Receptus inadequate? Although it wasbased on late manuscripts, and Lachmann's text on early manuscripts,both are based on actual readings. They simply adopted differentreadings at points of variations. So why is Lachmann rightand Erasmus wrong? How do we decide which reading is original?

Scholars have given many names to their answers, and they applythem in different ways. But fundamentally they use two tools:"Internal Evidence"and "External Evidence."

Internal evidence (sometimes called "TranscriptionalProbability" or the like) is based on logic: "Whichreading best explains the others?" It asks questions like,"Is there an easy way for this reading to have been convertedinto that one?"

External evidence is based on the manuscripts. It looks for thereading based on the "best," earliest, or most manuscripts.

Let's show what we mean by looking not at the Bible but at afamous passage from Shakespeare -- Hamlet, I.ii.129 (approximately;in my Yale Shakespeare, it's I.ii.133). This is one ofthe key soliloquies. You've probably heard the first line as

O that this too too solid flesh would melt

It so happens that there are three early witnesses to this passage,and none of them read it in the above form. The first quarto, theearliest published form of the passage, gives it as

Oh that this too much grieu'd and sallied flesh
Would melt to nothing....

The second quarto, the next form to appear, reads

O that this too too sallied flesh would melt

The First Folio of 1623, the only source to contain (almost) all of Shakespeare'splays, reads

O that this too too sollid flesh would melt

It is believed that the "sallied" of the second quarto isto be understood as "sullied." The folio reading is a divergentspelling of the common reading "solid."

So which is it? Solid flesh? Sullied flesh? Grieved and salliedflesh (which might in this case mean something like "battered")?

The first quarto reading can be ignored; it comes from a "badquarto," imperfectly remembered by one of the actors of the play. Butthe second quarto and the first folio are both fairly good texts. And bothreadings make good sense. If it is "solid flesh," it is naturalto ask that it would melt. But "sullied flesh" has its ownaptness, as Hamlet would have inherited it from his mother, who in herweakness has turned to Claudius. In choosing between them, a critic mustdecide which one best explains the other.

There is no definitive answer to this one. The Yale Shakespeare, which strikesme as rather casually edited, reads "solid."The revised Pelican has "sullied."The Riverside Shakespeare, in both the first and second editions, dodges theissue and prints "sallied." I personally think "sullied"the slightly better reading; it's in the second quarto, now considered thebest witness, and the first quarto reading seems to presuppose it; even thefolio reading uses a similar spelling. But we can't be certain; there isno guaranteed way to choose between the texts. This is the general problemof textual criticism, of which New Testament TC is a (somewhat exceptional,and certainly very important) example.

Rather than dwell on non-Biblical examples, let's take a handful of Biblical examples.By seeing how an actual apparatus criticus (table of informationabout variations) is constructed, we can probably make things a lotclearer.

For our first example, take part of 1 John 2:23. The KingJames version renders its Greek text "Whosoever denieth the Son, the same hathnot the Father." After this, however, they add, in italics (meaningthat it is not a correct part of their text) "[but] he that acknowledgeththe Son hath the Father also." Almost all modern version accept this longerreading as original -- that is, as part of the correct and original text.

In the Greek, this variation involves only eight words. The tablebelow shows the various words used here, along with the manuscriptssupporting them (it is customary in such apparati to leave out accentsand breathings. We list witnesses in the order papyri, uncials,minuscules, versions, church fathers). The name"ℵ" in thethird item refers to an important uncial manuscript known by that symbol.If a manuscript's symbol appears in parenthesis, it means that it generallysupports a particular reading but with some minor variation. If a manuscript'ssymbol is followed by an asterisk (e.g. 1739*), it means that this was the readingwritten by the original scribe of the manuscript, which some later owneraltered. The "corrected" reading (we put "corrected" inquotes because such corrections often replace a good early reading with a bad late one)is noted with a superscript c (e.g. 1739c) or sometimes,in older manuals, with two asterisks (e.g. 1739**). If a manuscript is markedvid, it means that the manuscript is incomplete or damaged, butthe surviving portion seems to support the reading in question. Obviously we cite onlya handful of the three-thousand-plus known manuscripts (many of which have not even beencollated yet, so we couldn't cite them even if we wanted to).A very brief description of most of the manuscripts cited here, including age, contents,and how various scholars have classified them, is found in theDescription of Manuscripts of the Catholic Epistles.

These are by no means all the manuscripts supporting either reading, but they give the general impression. Much the larger share of manuscripts support the short reading, though they are mostly minuscules, while the early uncials without exception have the longer reading (K, L, and 049 are uncials, but of late date -- ninth century or so).

The crucial matter, though, is the form of the reading. Note that both long and short readings end with the same set of letters: τον πατερα εχει. It would be very easy for a scribe's eyes to skip from the first occurence to the second. This is the error known as homoioteleuton ("same ending"), and it is incredibly common. Almost all manuscripts display at least a few instances of it. We don't as often see it affecting whole classes of manuscripts, but that is clearly the case here. The longer reading, despite being absent from the majority of manuscripts, is surely original.

A different sort of problem is illustrated by Matthew 19:20. Jesus istalking to the rich young man, and has just told him to keep the commandments.Does the young man say "I have kept all these" or "I have keptall these from my youth"? The evidence is as follows (f1and f13 are small groups of closely related manuscripts; you canlook up the manuscripts in the Description of Manuscriptsof the Gospels):

It is clear that the bulk of the manuscripts include the longer reading "from myyouth." On theother hand, the text without "from my youth" is supported by the twooldest manuscripts (ℵ* and B),and by several other manuscripts with what weshall learn are good or interesting texts. Most scholars would conclude, simplyon the basis of the manuscripts, that the shorter reading is better.

But we have more evidence. This reading, of course, has parallels in Mark (10:20)and Luke (18:21). Both of the other gospels have the words "from my youth."Now suppose you're a scribe. You've heard the phrase "I have kept all thesefrom my youth" a few zillion times in your life. Unless this is your first copyof the gospels, you've written it a few times in your life. If you encountera copy without the words, wouldn't you be tempted to add them? Certainly, if theywere present already, you would have no tendency to delete them.

This process is known as "assimilation of parallels." Scribes havea tendency to make texts read alike. If a text sounds familiar, the scribe tendedto conform it exactly to the familiar form. (You may have done it yourself. Tryreading this phrase: "To be, or not be, that is the question...."Did you notice the omission of the word "to" after "not"?)

So in all likelihood the original reading here is the one which omits "frommy youth."

You may have noticed that in both cases here we went against the readingsupported by the majority of manuscripts. Does this mean that we are undemocratic?

In a word, yes. One of the great rules of textual criticism is that "manuscriptsare to be weighed and not counted." Some manuscripts are good, some are lessgood. (Though all are at least occasionally questionable; as Michael Holmesputs it, "none are perfect, not even one; all have flaws, and fall short of theglory of the autograph" -- Michael Holmes in "A Case forReasoned Eclecticism," not yet published at the time of this writing.)So how do we decide?

This is a matter that scholars have been working on for centuries. When they began,the number of manuscripts known was much smaller than today, and old manuscripts wereespecially rare. Still, at about the turn of the nineteenth century a scholar namedGriesbach (following the lead ofSemler) discovered that the manuscripts known to himseemed to fall into threedistinct groups. The largest of these groups, by far, he called the "Byzantine,"because most of the manuscripts it included were written in the late Byzantine period.The two smaller groups he labelled "Alexandrian" (because it agreed withthe readings of such Alexandrian fathers as Origen and Cyril) and "Western"(because it was associated with the Latin versions used in the western Roman Empire).Thus arose the concept of "Text-types" --groups of manuscripts related at a stage more recent than the original.

This concept was refined in the second half of the nineteenth century byFenton John Anthony Hort (1828-1892),who did most of his work in collaboration withBrooke Foss Westcott (1825-1901).Westcott and Hort adopted Griesbach'sWestern and Byzantine types as given (although they called the Byzantinetext "Syrian"); the Alexandrian text they split into two groupswhich they called "Neutral" and "Alexandrian." (Thislatter distinction has been rejected by most scholars, who believe thatthe Neutral and Alexandrian text-types are just earlier and later forms of thesame sort of text; they generally call it by the name "Alexandrian.")

But Hort didn't just affirm the identity of these types. The discovery thatmade Hort famous was that the Byzantine text was (in his view) late. Hortbased this argument on a number of points (I have amplified some of these):

All of these points have problems. The first two remain true, but theyare an argument from silence. The fourth point is weakened by the fact thatconflations are not as common as Hort would imply, and occur in all types of manuscripts.The third point is the strongest by far, but has never been so fully tested as tosatisfy everyone.(See the article on Byzantine priority.)Still, the overall thrust of Hort's logichas convinced the majority of scholars. The Byzantinetext-type -- even though it contains nearly90% of the witnesses, and has influenced most of the others -- isregarded as a secondary product, derived from earlier text-types.

This left the field open to the earlier text-types, the Westernand Alexandrian.

The Western text in the Gospels consisted of only one Greek witness (CodexBezae, D/05, a well-known fifth or sixth century uncial), but it is supportedby most of the Old Latin versions, andby quotations from many early writers such as Irenæus and Tertullian.The Old Syriac versions also seem to belonghere, although they are not as pure and may have elements of other types.In the Acts, Bezae and the Old Latins are still the key elements of thetype, although 614, themargin of the Harklean Syriac, andthe other manuscripts of Family 2138 are believed by many to belong here.In Paul, the Old Latin still supports the type, as do the uncialsD (here D is 06, Codex Claromontanus, of the sixth century, not CodexBezae) and the closely-related ninth century pair F G. There are noknown witnesses to the type in the Catholic Epistles or the Apocalypse.

The Alexandrian text, which includes the majority of the non-Byzantinewitnesses, is more amorphous. It consists of both uncials and minuscules, as wellas versions. In Hort's time, the most important and basic witness to the typewas the famous Codex Vaticanus, B/03,which contains the Gospels, Acts, andCatholic Epistles complete as well as most of the Pauline Epistles. It was alsothe earliest representative of the type, dating from the fourth century. Alsofrom the fourth century, and nearly as important, is Codex Sinaiticus,ℵ /01,the only uncial to contain the entire New Testament. They are supported by theCoptic versions. In addition, they are supported inpart by manuscripts such as the uncialsCand L in the gospels and the uncialsA andCin the Acts and Epistles, as well as by minuscules such as335798921241 inthe gospels and33811175 in the Acts and Epistles.

Most of these latter manuscripts, however, display a phenomenon known as"mixture." This means that they contain readings from more thanone text-type. Typically they will have some Alexandrian and some Byzantinereadings, although there may be a few "Western" readings thrown in as well.

The reason for this is not hard to imagine. Unlike today, when books arecheap enough to simply be purchased and referred to only intermittently, old bookswere used. So the users were always writing notes, commentaries, andcorrections in the margin. It was not unusual for a later copyist to assumethese marginal remarks belonged in the text (or at least might belongin the text), and insert them into the manuscript he was writing.

Then, too, manuscripts were copied in a scriptorium, and corrected. A corrector (διορθωτῄς)would carefully read over the new copy, comparing it to some official,reputable copy. Often this reputable copy would not be of the same typeas the manuscript used to make the original copy, meaning that the correctorwould add readings of a second text-type to the once-pure text of themanuscript. We can actually see this happening in some manuscripts;424 has a Byzantine text thathas been corrected toward the readings of1739, while many famous manuscripts(including ℵand both Ds) have been corrected toward the Byzantinetext. When new copies are made from these manuscripts, of course, thecorrections go straight into the text of the copy, producing mixedmanuscripts.

Mixture makes the task of textual criticism much harder. Since mostmanuscripts have more than one "parent," it means that wecannot trace a simple genealogy. Although P75, B, and L areall related, L is not a child of B, which isnot a child of P75. This means that we cannot simply goback up the generations to find the original reading of a text-type,let alone of the original text.

Still, by careful use of both internaland external evidence, it is usuallypossible to determine the readings of text-types. Hort, for instance,found that B preserved the readings of the Alexandrian/Neutral text in thelarge majority of cases.

But at this point Hort faced a problem. Both the Alexandrian and "Western"types were early, and went directly back to the original. How, then, didone decide between the two in cases where they disagreed?

Here Hort turned to internal evidence. The "Western" text, he found,was marked by paraphrase, expansion, and stylistic "improvements"of all sorts. The Alexandrian text, by contrast, was concise -- evenabrupt -- and had more than its share of infelicitous readings.

On this basis, Hort concluded, the Neutral (Alexandrian) text was best and mostreliable. Unlike the Western text, it was not rewritten; unlike the Byzantinetext, it was not a mixture of older elements. The text printed byWestcott & Hort was largely thatof the Alexandrian text, and of B in particular. And it was widely feltthat the Westcott & Hort text was the best New Testament edition ofthe nineteenth century. Even today, our printed texts are strongly"Hortian."

But the twentieth century has seen changes. New manuscripts -- includingall the papyri and many early uncial fragments -- have been discovered. Ourknowledge of the versions is much greater.

This has had many consequences. A new text-type -- the"Cæsarean" -- has beenproposed (though its existence is not so widely accepted today as in theearly part of the twentieth century). The various substantial papyri --particularly P46and P75 -- have altered ourunderstanding of the early history of the text. Discoveries of newand better manuscripts of the Fathers have helped us understand allstages of that history. And new tools, some computer-aided, haveallowed us to assess many manuscripts (especially minuscules) that hadnever previously been studied. We know of many manuscript groupingswe had not previously been aware of. We have also learned that eventhe Byzantine text is not one great monolith; although it is themost coherent of the text-types, even it has phases and has undergonea certain amount of evolution.

List of New Testament Manuscripts

In the light of the complexity we now see in the relationshipsbetween manuscripts, we cannot do as Hort did and generallyjust follow the text of B. We need to be aware of all the non-Byzantinemanuscripts, and keep their peculiarities in mind. We also must knowand understand the Byzantine text. If we believe, with Sturz andothers, that it is early, we must take its readings into account. Evenif we accept the opinion of Hort in its entirety, and consider theByzantine text late, we still must know its readings so thatwe can see how they have influenced other manuscripts.

The following list briefly describes some of the more important (generally non-Byzantine) NewTestament manuscripts and their characteristics.

The above list shows that we know quite a bit about certainmanuscripts. Even so, the matter of manuscript classification remains highlyuncertain. The reader interested in a discussion of contemporaryissues is referred to the article onText-Types and Textual Kinship.

Perhaps as a result of this uncertainty, textual criticism in thetwentieth century has placed increased emphasis on internal evidence.All textual critics balance internal and external evidence to somedegree, but the twentieth century has seen a new class of critics.Often called "Radical" or "Thoroughgoing Eclectics,"they decide readings almost entirely on the basis of internal evidence;manuscripts are simply the sources of the readings to be examined.Foremost among these scholars are G. D. Kilpatrick and J. Keith Elliot.

The "documentary" methods of Hort, meanwhile, have beenalmost completely abandoned. The most common method today is"Reasoned Eclecticism," which attempts to give bothinternal and external evidence full voice. The interested reader istherefore advised to study the list of Canonsof Criticism, examining both the rules for internal and externalevidence.

Final Examples

Let us conclude this far-too-brief survey with a handful of additionexamples that demonstrate both internal and external rules. A handful ofadditional Examples are available in theEncyclopedia, but many of these stress the use of text-types andexternal evidence, and so are perhaps not ideal for beginning students.

In the examples below, where the "lemma" (the Greek text tobe examined) contains the notation [add], it means that some manuscriptsadd words, to be specified in the list of variants which follows the maintext.

James 5:7

ο γεωργος... λαβη[add]προιμον καιοψιμον: the farmer... receives... early and late [add]

This reading can be resolved using either internal or external evidence.Internally, it is clear that the original reading is the short one. If thetext originally said "the farmer waits to receive early and late,"this could easily have confused scribes, who would feel that the verb needsan object. A forerunner of the Byzantine text added "rain," whilea few scribes added "fruit" instead. Thus the reading without either nouneasily explains the others. Whereas if either "rain" or "fruit"were original, there would be no reason to omit it, and even less reason to changethe one to the other.

The manuscript evidence is also clear. "Fruit" is simplyinadequately supported. The support for "rain" is somewhatbetter, consisting of the Byzantine text, Family 2138, and an assortmentof late Alexandrian manuscripts. The omission, however, has the supportof Family 1739, of the earliest Alexandrian witness (B, supported byP74 and the Sahidic), and a wide variety of versions. While thisis not as decisive as the internal evidence, it is strong. Combined, theinternal and external evidence make it all but certain that the shortreading is original.

Matthew 13:9

ωτα [add]ακουετω:with ears [add] let that one hear

This reading will usually be decided based on internal evidence, since the externalevidence is somewhat spilt. The earliest Alexandrians omit "to hear," as doseveral of the best Old Latins. On the other hand, the majority of both Alexandrianand "Western" witnesses, along with the entire "Cæsarean" andByzantine families, add the infinitive. On the basis of the external evidence, mostscholars would probably prefer the short reading, but would be open to counter-suggestion.

The internal evidence is quite decisive, however. In Mark we find the phrase"ears to hear" three times (4:9, 23, 7:16), supported in two instancesby Luke. In Matthew, however, all three instances of the phrase are marked byvariation. In each case, the Byzantine text reads "ears to hear," andat least some early witnesses omit "to hear." Now we know that Matthewabbreviated Mark wherever possible, and we know that scribes were always harmonizingone gospel to another (that is, making both gospels sound alike -- usually by graftingthe longer reading of one gospel onto the shorter reading of another). Therefore thereis every likelihood that the reading without "to hear" is original (here andin 11:15, 13:43), and the longer readings are assimilations to Mark.

Several Final notes....

First, critical editions use many different formats to present data.The system above is by no means typical. A good critical edition will explainhow it is to be read, but you can also find information in the article onCritical Editions -- which also briefly describesthe nature and history of several of the major editions.

Second, it should be stressed that textual criticism, unlike any other Biblicaldiscipline, should not be faith-based. The goal must always be the highestpossible degree of scientific objectivity. This is simply a logical necessity.The Bible is one of the basic pillars of Christian theology (most Protestantsects would say the basic pillar). Therefore it follows that we want toreconstruct it as accurately as possible. But as soon as one allows personalpreference (whether it be called that or "the voice of the Holy Spirit"or the like) to determine the text, where does one stop? I will offer myself asan example. I personally find the doctrine of predestination to be simply abhorrent.It's boring for God and utterly unfair for humans. If I were to allow myown opinions (which feel just as much like the voice of the Holy Spirit asthe next person's opinion) to control me, I would always be tempted to delete or softenpro-predestination references. We will all have such prejudices. The only possiblesolution is to follow objective rules. Your rules may differ from mine, andso may produce different results -- but at least the result will not sufferfrom theological bias. Treat textual criticism as a science (using logicin the application of internal evidence and text-types and mathematicaldata in the evaluation of the external), and you should do well.

Some textual manuals, such as Ellis R. Brotzman's Old Testament Textual Criticism:A Practical Introduction (p. 129) suggest that every time one makes a textual decision, the textual critic should explain its importance for exegesis.I would strongly urge textual critics not to do this, as it muddies thethinking. Readings must be chosen solely on the basis of the evidence, not thecritic's faith. If a textual critic can't perform his or her task objectively, heor she shouldn't be doing textual criticism; if an exegete can't figure out what thevariant readings mean, the exegete should go out and get a real job. It is one thingto mark which readings are most uncertain, as several editions do; it is another forthe textual critic to do the exegete's job.

Third, I've had people come to me saying, in effect, "Help! Thistextual criticism stuff is undermining my faith." I would stress thatthis is no concern of the textual critic, who has a job to perform. (Yet anotheradvantage of textual critics with no religious axe to grind.) But I suppose weshould speak to this point.

For starters, it should be noted that every ancient writing extant inmultiple copies shows variations -- often much more significant variationsthan we find in the New Testament text. If 6,000 New Testament manuscriptsshowed no variation at all, it would be clear and direct evidence ofsupernatural influence (note that such influence need not have been God's;it could theoretically be the work of a being opposed to God). But Godpresents no other such explicit evidence; why offer it only in a strangeand obscure form that no one could appreciate until recently when we haveat last been able to study enough manuscripts to prove the point? Evenif you have some sort of inerrantist belief, it makes no sense. And thereis a faith issue the other way, too: What sort of God would keep the Bibleinviolate but allow wars and rape and murder and child abuse? A God whosimply takes a "hands off" attitude is one thing, a capriciousGod is another.

As to how the textual critic can answer the doubts of laypeople confrontedwith the alleged issue of textual criticism, I would suggest simply havingthe doubter consult one of the modern English translations. The New RevisedStandard Version, for instance, records textual variations with the words"other ancient authorities read...." Have the person read some ofthese footnotes. Do any of them really affect the person's beliefs? Does itreally matter if the Greek transliteration of the name of the Hebrew King Amonwas "Amon" or "Amos"? Does it matter if people in Alexandriaspelled their verbs in a way modern writers consider uncouth?Variation in the text is real andis widespread. Few if any scholars believe that we have recovered the originaltext with absolute certainty -- but I know of none who regard the differenceas so substantial as to be actually capable of producing heresy. Scholars suchas Burgon and Pickering have been intemperate (and, in the latter case atleast, demonstrably inaccurate) in their attacks on other scholars' methods. But eventhey have not shown any instance of modern (as opposed to ancient) editorsproducing any readings which affect Christian doctrine; doctrine is a unityand does not rest on a particular passage.

Though I would strongly argue, personally, that if such a reading doesexist, it is still the textual critic's duty to adopt that reading if theevidence supports it."καιγνωσεσθε τηναληθειαν και ηαληθειαελευθερωσειυμας" (John 8:32, a verse with nosignificant variants).

Good textual critics will not go too far. As P. Kyle McCarter once commentedas regards the canon of criticism regarding the harder reading, "The moredifficult reading is not to be preferred when it is garbage."

There is an interesting analogy in Karen H. Jobes and Moisés Silva'sIntroduction to the Septuagint (page 124): Consider purifying our watersupplies (or anything else involving sanitation, e.g. washing hands orpasteurizing milk): No matter how hard you try, none of these activities willeliminate all contamination. Does that mean that it's not worth purifying water --that we should drink dirty water and assume it's clean? Only if you like typhoidfever. We can't reconstruct the original text perfectly, because we are humanand it is a text copied by humans. But we can produce better and purer text.We can -- but only if we're willing to concede the need. Textual criticism doesnot threaten the Bible. Refusing to engage in TC is the threat.

Good luck!

Appendix: The Text of Chaucer's Address to his Scribe

The poem which opens this article is Chaucer's own comment on theaccuracy of his secretary's work. A non-poetic paraphrase into modernEnglish will show that inaccurate copying was just as much a problem forChaucer as for Biblical copyists:

Adam the scrivener, if ever it befalls you
To write Boethius or Troilus anew,
Under your long hair you must have scales
Unless you copy what I write more truly!
So often I must redo your work:
To correct it, and rub and scrape,
And all because of your negligence and rape.

There are only two extant pre-critical texts of this poem:Cambridge MS. Trinity College R.3.20 and the text of Stowe's 1561edition, seemingly from a lost manuscript. There are no variantsin the text listed by Benson in the third edition of TheRiverside Chaucer (text on p. 250, textual notes on p. 1188;the second edition of this work, by Robinson, notes some variantspellings and several conjectures by other editors),but the titles are different: The Cambridgems. calls the poem "Chaucer's Wordes Unto Adam, His OwneScriveyn," while Stowe's edition uses the title "ChaucersWoordes vnto his own Scriuener." Nonetheless there are differencesbetween modern authorities. The text at the top of this article,reprinted below with line numbers added, is from Benson.There follows a collation showingthe variants (mostly trivial, real variants shown in bold)in James J. Wilhelm's Medieval Song.

1 -- Adam scriveyn, if ever it thee bifalle
2 -- Boece or Troylus for to wryten newe,
3 -- Under thy long lokkes thou most have the scalle,
4 -- But after my makyng thow wryte more trewe;
5 -- So ofte adaye I mot thy werk renewe,
6 -- It to correcte and eke to rubbe and scrape,
7 -- And al is thorough thy negligence and haste.

1: scriveyn ] scrivain
2: Troylus ] Troilus; for to ] to; wryten ] writen
3: lokkes ] lockes; most ] moste
4: makyng ] making; thow ] thou; wryte ] write; trewe ] true
5: adaye ] a-day; mot ] moot
6: eke ] eek; rubbe ] rub
7: al ] all; thorugh ] thourgh