How the text reached us, and how reliable it is

"But he said to them, 'Do not be alarmed. You are looking for Jesus the Nazarene, who was crucified. He has been raised! He is not here. Look, there is the place where they laid him. But go, tell his disciples, even Peter, that he is going ahead of you into Galilee. You will see him there, just as he told you.' Then they went out and ran from the tomb, for terror and bewilderment had seized them. And they said nothing to anyone, because they were afraid." (Mark 16:6–8, NET)¹

Now look at what your own Bible prints next. There will be a break in the page and a note, usually in square brackets, telling you that some of the earliest manuscripts do not include 16:9 to 20.¹ Nearly every modern translation prints one. Most readers skip past it. But stop and think about what it's saying. The translators are telling you that the next twelve verses, which include the risen Jesus appearing to Mary, the command to preach to all creation, the promise about handling snakes, and the ascension, are in some manuscripts and not in others, and that the ones missing them are among the oldest we have.

That note is a window into the whole question of this lesson. Lesson 8 told you which books are in the Bible. It said nothing about whether the words in those books are the words their authors wrote. Between a letter Paul dictated in the 50s and the Bible on your table lie about fourteen centuries of copying by hand, and then four of printing. Hand-copying makes mistakes. So how do we know what Paul wrote? The answer runs through the Dead Sea Scrolls, through a count of about 400,000 variants that gets quoted in every direction, and through Bart Ehrman, the best-known critical voice on this subject, who turns out to agree with the conservatives about more than either side advertises.

Two links, not fifty

Start with the misconception that shapes how people hear everything else. "The Bible has been translated so many times that it can't be trusted." The picture behind that sentence is a chain: Hebrew to Greek to Latin to German to English to modern English, each link copying the last, errors piling up like a game of telephone.

That is not how a modern Bible is made. The translators of the ESV, the NIV, the CSB, and the rest sat down with a printed Hebrew text and a printed Greek text and translated directly from them. The Hebrew edition is built on a single medieval manuscript we'll meet in a moment, checked against the Dead Sea Scrolls and other witnesses. The Greek edition is built from thousands of manuscripts compared against each other. So the chain has two links: manuscripts to edition, edition to your Bible. Nothing in between.

What is true, and what gives the myth its grip, is that the King James translators in 1611 worked from a Greek text that rested on a handful of late manuscripts, which is why your Bible and a King James differ at 1 John 5:7 and a few other places. We'll come to that. It's a story about the manuscripts, not about translation, and lesson 11 picks it up from the translation side.

How a text gets copied, and how it gets fixed

Before printing, every copy of every book was made by hand. The New Testament was first written on papyrus, a paper made from reeds, and later on parchment, prepared animal skin. Early Jewish Scriptures were kept on scrolls; Christians took very early to the codex, the bound book with pages, which is why our oldest New Testament manuscripts are fragments of books rather than rolls.²

Two things about those copies make the mistakes easier to picture. The oldest Greek manuscripts are written in capitals with no spaces between the words and almost no punctuation, so a scribe was not reading words off a page the way you are reading this; he was cutting a continuous ribbon of letters into words as he went. And for much of the period, copying was done by ordinary literate Christians rather than by trained scribes in a scriptorium.

A scribe copying by hand makes predictable kinds of mistake, and the standard textbook on the subject, by Bruce Metzger and Bart Ehrman together, sorts them.² Errors of the eye: two lines end with the same word, and the scribe's eye jumps from the first to the second, dropping everything between. Errors of the ear, when a text was read aloud to a room of copyists and two Greek vowels sounded the same. Errors of memory, when the scribe carried a phrase in his head and wrote a familiar version from a parallel passage. And errors of judgement, when a scribe thought the text in front of him was wrong and "fixed" it: smoothed the grammar, harmonised one Gospel with another, or, in a small number of cases that scholars argue about, adjusted a phrase toward what he believed.

You will meet all four kinds in this lesson. The two endings of Mark, and the shorter ending that a few manuscripts print alongside the long one, are errors of judgement: a copyist thought the book stopped too abruptly and supplied a finish. So is the woman caught in adultery, a real story slotted into the place where it seemed to fit. Keep the taxonomy in mind as the examples arrive, because naming the kind of mistake is most of explaining it.

Hold onto what comes next. It is the answer to nearly every objection you will hear.

Many independent copies are better than one perfect one. When a scribe in Alexandria makes a mistake, that mistake travels down to the copies made from his copy. It does not appear in a manuscript copied in Rome from a different exemplar, unless it was already in the text both were copying. So when copies diverge, the divergence itself tells you where an error entered, and comparing families of copies lets you work back toward what they had in common. More manuscripts help because there's more to compare. Older manuscripts help because fewer generations of copying separate them from the source. Widely spread manuscripts help because an error made in one place is exposed by copies made somewhere else.

When the manuscripts are evenly matched, one criterion does most of the work, and it is worth learning by name: prefer the harder reading. Copying has a direction. A scribe who meets rough grammar, an awkward join, or a saying that sounds wrong tends to smooth it, and almost nobody roughens a smooth text on purpose. So when one reading is difficult and the other is comfortable, the difficult one is usually the older, and the comfortable one is the repair. You will watch that criterion decide both worked examples below.

This is why the telephone game is the wrong picture. Telephone is one chain, and you only ever hear the last whisper. This is a tree, with branches you can compare, and the comparing is the point.

One more thing about the method, so you know what kind of thing it is. Weighing external evidence (age, spread) together with internal evidence (which reading explains the other) is the majority approach, and it has a name, reasoned eclecticism. It is not the only one. A minority following G. D. Kilpatrick and J. K. Elliott argue that we can never really know a manuscript's ancestry, so internal evidence should decide every case on its own. Another minority, Maurice Robinson and Zane Hodges among them, argue that the text carried by the great majority of manuscripts is the transmissionally normal one, and that the early Egyptian papyri survived because Egypt's climate preserves papyrus rather than because their text was better. Their argument is not that more copies win a vote; it is a claim about how texts actually get copied over fifteen centuries. Most textual critics are not persuaded, this course follows the majority, and lesson 11 comes back to the tradition that reads its Bible from the Received Text.³²

The Old Testament: a thousand-year gap, closed in 1947

Start with the oldest thing we have. In 1979, in a burial cave at Ketef Hinnom on the edge of Jerusalem, excavators found two tiny rolled-up sheets of silver.³ When they were unrolled, each carried a form of the priestly blessing from Numbers 6:24–26:

"The Lord bless you and protect you; the Lord make his face to shine upon you, and be gracious to you; the Lord lift up his countenance upon you and give you peace." (Numbers 6:24 to 26, NET)⁴

They date to the late seventh or early sixth century BC, before the Babylonian exile.³ They are amulets, not Bibles, and they are the oldest surviving biblical text of any kind: a few lines, from the First Temple, that a Jew in Jerusalem could read in a synagogue this week.

That's a fragment. For the text as a whole, the story before 1947 was this. The oldest complete Hebrew Bible in the world is the Leningrad Codex, copied in 1008 or 1009, and the slightly older Aleppo Codex, from about 930, lost roughly 40 percent of its pages at some point after the 1947 riot in Aleppo; when and how is still argued.⁵ Both come from the Masoretes, the Jewish scholars who between roughly the sixth and tenth centuries standardised the Hebrew text, added the vowel points, and counted the letters so that copies could be checked. The text they fixed is called the Masoretic Text, and nearly every modern Old Testament translation is made from a printed edition based on the Leningrad Codex, the ones scholars call BHS and BHQ.⁵ Not quite every one: the Hebrew University Bible Project and the Jerusalem Crown work from the Aleppo Codex instead.⁵

That is the Protestant and Catholic practice, and it rests on a decision worth naming. The Orthodox churches take a different view, and a considered one: for them the Septuagint is the church's Old Testament, because it is what the apostles quoted and what the Greek-speaking church has read for nineteen centuries. So "which text" is not only a technical question. It is partly a question about which community's reception counts, which is the same question lesson 8 met over which books.

So before 1947, the oldest complete Hebrew text of Isaiah was about 1,700 years younger than the prophet. Then a shepherd near the Dead Sea threw a stone into a cave.

Between 1947 and 1956, eleven caves near Qumran yielded the Dead Sea Scrolls. On the Israel Antiquities Authority's count, about 230 of the manuscripts are biblical, and every book of the Hebrew Bible is represented except Esther.⁶ Most are fragments. One is not. The Great Isaiah Scroll, catalogued as 1QIsa^a, is a complete copy of Isaiah, all sixty-six chapters, on seventeen sheets of prepared skin sewn end to end into a roll about seven metres long. By the shape of its letters it is dated to about 125 BC; radiocarbon tests give a range of roughly 350 to 100 BC.⁷

Do the arithmetic. The Leningrad Codex is from 1008. The Isaiah Scroll is from about 125 BC. That's more than a thousand years of copying between them, and until 1947 nobody had seen the older end of it.

That's the first half of what the scrolls show, and it's the half everyone quotes: across a thousand years, the Masoretes' text turns out to be a faithful copy of what Jews were reading before Jesus was born. If you'd asked a scholar in 1940 to bet on it, most would have bet against.

Now the second half, which is just as true and gets quoted much less. The Qumran caves held other biblical manuscripts besides the Isaiah Scroll, and some of them are not Masoretic-type texts at all. The clearest case is Jeremiah. Some Qumran copies support the text behind the Greek Septuagint, which is about an eighth shorter than the Hebrew that came down to us and puts the oracles against the foreign nations in a different place in the book.⁷⁸ That is not a difference of spelling; it is two editions of Jeremiah. Samuel shows something similar on a smaller scale, and you can see its effect in your own Bible: in the Masoretic text of 1 Samuel, Saul asks whose son David is one chapter after David has become his armour-bearer, because the longer Hebrew text carries a version of the David and Goliath story that the shorter Greek one does not.⁸ In other words, around the turn of the era, more than one form of the text of some books was in circulation, and the Masoretic Text is one of them, the one that Judaism preserved. Scholars call this the plurality of the text in that period, and it's why an honest lesson says "remarkably stable" and "also plural" in the same breath. Emanuel Tov's textbook is the standard reference for it, and where to go next.⁸ Anyone quoting only one half is quoting half the finding.

Here's one place where you can watch a modern translation use the scrolls. Isaiah 53:11, in the ESV, reads: "Out of the anguish of his soul he shall see and be satisfied", and the footnote says: "Masoretic Text; Dead Sea Scroll he shall see light".⁹ The NIV takes the other option and prints: "After he has suffered, he will see the light of life and be satisfied", with the note: "Dead Sea Scrolls (see also Septuagint); Masoretic Text does not have the light of life."⁹ So the Hebrew of the Middle Ages has the servant "see" with no object; the Isaiah Scroll and the Greek translation made before Christ both have him "see light". Two translation committees, working from the same evidence, made different calls and told you so in the margin.

The New Testament: fewer years, more copies

The New Testament's situation is the mirror image. The gap between writing and oldest copy is much smaller, and the number of copies is much larger. Both facts need care.

The Institut für Neutestamentliche Textforschung in Münster keeps the register, in four classes: papyri, majuscules (early manuscripts in capital letters), minuscules (later ones in a running hand), and lectionaries.¹⁰ Daniel Wallace, who has spent his career photographing them, wrote in 2012 that "over 5800 have been catalogued".¹¹ Counts published since range from about 5,600 upwards, depending on how the register's deletions and merges are handled, so treat the figure as roughly 5,600 to 5,800 rather than a precise total.¹⁰ Add early translations into Latin, Syriac, and Coptic, and quotations in the church fathers, and you have more evidence for the New Testament's wording than for any other ancient text. The versions and the fathers are worth more than their number suggests, and for a reason worth holding: a father quoting a verse in a sermon fixes a reading to a named city in a datable decade, which no undated manuscript can do, and a translation made into Syriac in the second century is a witness to the Greek that lay in front of the translator. That sentence is true and it is not the argument, so read the next paragraph before you use it on anyone.

Most of those 5,800 are from the ninth century onward. If they were all we had, we'd have a well-attested medieval text and no way to check it against anything earlier. The argument is the age and spread of the early witnesses, and those are worth knowing by name.

The one usually called the oldest is a scrap the size of a credit card. Rylands Greek Papyrus 457, known as P52, holds a few lines of John 18:31–33 on one side and 37–38 on the other, and measures 8.9 by 6.0 centimetres.¹² Its first editor dated it to the first half of the second century, and for decades it was quoted as "about 125". Then in 2005 Brent Nongbri showed that the handwriting fits anywhere from the early second century to the early third, and the Rylands Library's own catalogue now says "Recent research points to a date nearer to 200 CE."¹²¹³ So say "second century" and leave it there. Even at the later end, that's a copy of John circulating in Egypt within about a century of its writing.

From the same period, two much larger papyri survive. P46, around AD 200, contains most of Paul's letters in one codex, and is split between Dublin and Michigan.¹⁴ P75, from about 200 to 225, holds much of Luke and John and is textually very close to the great fourth-century codex we're about to meet, which showed scholars that the text of that codex was not a fourth-century editorial recension, as some had supposed, but was already being copied a century and a half earlier.¹⁴

A century and a half later come the two manuscripts that matter most. Codex Sinaiticus, found by Constantin von Tischendorf at the monastery at the foot of Mount Sinai in 1859 and now mostly in the British Library, is the oldest complete New Testament in the world, from the fourth century, and you can turn every page of it at codexsinaiticus.org.¹⁵ Codex Vaticanus, in the Vatican Library, is from the same century and is, for most of the New Testament, the single most important manuscript there is.¹⁵ Neither shaped the printed Greek Bible for a long time: Vaticanus was not printed until 1587 and Sinaiticus was not known to Western scholars until Tischendorf, which is why the printed text the King James translators used knew nothing of either.¹⁵ When those two agree against the medieval majority, editors take notice, and the three famous passages below are all cases where they do.

Here's the whole shape drawn out, on one scale, so you can see the two gaps side by side.

Look at the dashed line on the Isaiah row. That's the thousand-year gap that the Isaiah Scroll closed, and the red dot is the scroll landing near the older end of it. Now look at the John and Paul rows. The red dots are a finger's width from the writing. The New Testament's gap from author to oldest copy is a small fraction of the Old Testament's, and that's the fact that makes it the best-attested ancient text there is. It's also why lesson 6's argument about the Gospels' dates matters here: the earlier you date the Gospels, the earlier the papyri sit relative to them, but even on the critical dates the gap is short.

What a variant is, and what 400,000 of them means

A textual variant is any place where one manuscript differs from another. Nothing more than that, and the looseness of the definition is why the number is so often misused.

Wallace puts the number at around 400,000, and gives his reasoning: the older estimate of 300,000 was too low once the big collation projects on Luke, Jude, James and 1 and 2 Peter were done, so "the estimates today are closer to 400,000".²⁹ Ehrman's own phrase is "hundreds of thousands of textual variants", and in Misquoting Jesus he gives a range of 200,000 to 400,000.¹⁶¹⁸ They count the same way: a difference counts once, no matter how many manuscripts carry it. Now notice what that method implies. The more manuscripts you have, the more differences you will find, because every fresh copy brings its own slips. A text with 5,800 manuscripts will have vastly more variants than a text with ten. A large count of variants is what abundant evidence looks like, not what corruption looks like. If we had one manuscript of the New Testament we would have zero variants and no way to know whether that manuscript was any good. A text with no variants is a text with no evidence.

Then sort them. Wallace's categories, in his own words: "Spelling and nonsense readings are the vast majority, accounting for at least 75% of all variants."¹¹ Next come differences that don't change the meaning in translation, such as Greek word order and the presence or absence of the definite article before a name. Then differences that would change the meaning but have no chance of being original, because they sit in one late manuscript against everything else. And last: "The smallest category by far is the last category: meaningful and viable variants. These comprise less than 1% of all textual variants."¹¹

Under 1 percent of 400,000 is still a few thousand places, and you have just read that Mark 16 is one of them. So don't let the small percentage do too much work either. Here is the figure that gives it a size. Wallace, comparing the late Byzantine manuscripts, the family furthest from the modern critical text, finds that even they differ from it "only about 4%–5% of the time".¹⁷ That is the outer edge of the disagreement between whole traditions, not the disputed residue, which is far smaller again. The question is what those places are, and here the two men whose names sit on the standard textbook, who differ publicly about what the variants mean, say something close to the same thing.

Here is each man in his own terms.

Wallace, the evangelical, on the meaningful and viable variants: "Yet, even here, no cardinal belief is at stake. These variants do affect what a particular passage teaches, and thus what the Bible says in that place, but they do not jeopardize essential beliefs."¹¹ On his own website, listing what he and Ehrman agree about after their public debates, Wallace writes that "we both agree that the scribes changed the texts hundreds of thousands of times, that certain passages are not authentic (notably, the longer ending of Mark and the story of the woman caught in adultery), that the proto-orthodox changed the text at times, that only a small fraction of all textual changes are both meaningful and viable."¹⁷

Ehrman, the agnostic, agrees about the doctrines, and then asks why doctrines should be the measure. "Why is it that 'threats to significant doctrines' are taken to be the only standard of what is important to the Christian religion? In my view this idea is completely wrong-headed."¹⁸ His examples are worth stating at their full strength, because they are stronger than the summaries usually allow. At Mark 1:41 the manuscripts have Jesus either moved with compassion or angry as he heals the leper, which decides what kind of figure Mark is drawing.³⁰ At Luke 3:22 the voice at the baptism either says "you are my beloved Son, in you I am well pleased" or quotes Psalm 2, "today I have begotten you", and the second reading is the one an adoptionist would want, so the variant bears on whether Luke's Jesus is God's Son eternally or becomes so at his baptism.¹⁸ At Luke 22:19b to 20 the words about the body "given for you" and the cup of the new covenant are present in some manuscripts and absent in others, and on the shorter reading Luke has no doctrine of atonement in his Last Supper at all.¹⁸ And John either has the woman caught in adultery or does not. None of these overturns a creed. Each changes what one book teaches, and Ehrman's point is that if you care what each author meant, rather than only whether the Trinity survives, then these matter. He puts it this way: "changes in the Bible can be inordinately significant without affecting any cherished doctrines of the evangelicals".¹⁶

I think Ehrman is right about that, and I want you to hold his point and Wallace's together, because they don't conflict. The text is recoverable to a very high degree, no article of the creeds rests on a disputed reading, and there are a few thousand places where it matters to a careful reader which reading is right. All three are true. A Christian who says only the first two has left out something the reader deserves to know, and so has a sceptic who says only the third. Ehrman, to his credit, says all three, and so does Wallace. The argument between them is about which one carries the weight.

The questions to ask at any note

Before the worked examples, here is the procedure, so that you are learning a method and not three stories. When your Bible flags a passage, ask these in order.

1. Which witnesses lack it, and which have it?
2. How old are the witnesses on each side?
3. How many separate regions and languages do they cover?
4. Does the wording sound like the rest of the book, in vocabulary and style?
5. Does the passage move around in the manuscripts that have it?
6. Which reading explains how the other one arose, and which is the harder one?

Then the vocabulary of the notes themselves, because the wording is not standardised and the differences matter. "Some manuscripts" usually means the reading is disputed but the editors are not telling you how strong the evidence is. "The earliest manuscripts do not include" is stronger and is telling you where the weight sits. Single brackets in the text mean the editors doubt the passage and have kept it. The double brackets in the printed Greek edition mean something more precise: the editors judge the passage not original and print it anyway, because it is early and the church has always read it. A footnote alone is the weakest signal of the four.

And a tool, because you will meet passages this lesson does not work. The NET Bible carries about sixty thousand translators' notes, and at a disputed passage the note names the manuscripts and explains the decision in English. That is the fastest way from "my Bible has a bracket here" to "here is the evidence".

Worked example 1: the ending of Mark, and the woman caught in adultery

Back to the bracket we started with. Here's the evidence behind it, from Bruce Metzger's Textual Commentary, the book that explains the decisions behind the standard Greek edition. His entry on Mark 16 is online in full if you want to read the whole thing.¹⁹

The twelve verses are "absent from the two oldest Greek manuscripts (א and B)", that is, Sinaiticus and Vaticanus, "from the Old Latin codex Bobiensis, the Sinaitic Syriac manuscript, about one hundred Armenian manuscripts, and the two oldest Georgian manuscripts."¹⁹ That's the external evidence: the oldest witnesses, and witnesses from different regions and languages, stop at 16:8. Remember the mechanism. An omission that shows up in Greek, Latin, Syriac, Armenian, and Georgian copies is not one scribe's slip.

Then the internal evidence, which is where you can join in. Metzger: "The vocabulary and style of verses 9-20 are non-Markan", and "the connection between ver. 8 and verses 9-20 is so awkward that it is difficult to believe that the evangelist intended the section to be a continuation of the Gospel."¹⁹ Read 16:8 and 16:9 back to back and you'll feel the join: Mary Magdalene, who has been on stage since 15:40, is reintroduced in verse 9 as if for the first time, "from whom he had driven out seven demons".

There's a further clue. A few manuscripts have a different, shorter ending after verse 8, two sentences long, and some have both endings one after the other.¹⁹ That's what it looks like when a book ends abruptly and later readers try to finish it: more than one attempt survives.

The committee's conclusion: "on the basis of good external evidence and strong internal considerations it appears that the earliest ascertainable form of the Gospel of Mark ended with 16.8."¹⁹

Now the question that matters for a Christian reader. What changes if verses 9–20 are not Mark's? Nothing in the creeds. The resurrection is announced at 16:6, "He has been raised! He is not here", in the part nobody disputes, and every other Gospel narrates the appearances. What does change: the verse about believers handling snakes and drinking poison (16:18) loses its claim to be Mark's, which is worth knowing if you ever meet a church built on it, and the phrase "The one who believes and is baptized will be saved" (16:16) can't be quoted as Mark's either. Whether Mark meant to end at "because they were afraid", or whether his real ending was lost, is the part scholars argue about, and I'll flag it again below.

The second passage has the same shape with one twist. John 7:53–8:11, the woman caught in adultery, sits in your Bible inside brackets or under a note saying that the earliest manuscripts do not include it.²⁰ Metzger, whose entry on this passage is also online, calls the evidence for its non-Johannine origin "overwhelming": it's absent from P66 and P75, from Sinaiticus and Vaticanus, and from a long list of other witnesses, and its style and vocabulary "differ noticeably from the rest of the Fourth Gospel".²¹ The twist is that in the manuscripts that do have it, it floats. "Most copyists apparently thought that it would interrupt John's narrative least if it were inserted after 7.52", Metzger writes, but "others placed it after 7.36, or after 21.25, or after Luke 21.38."²¹ A passage that moves around is a passage that was slotted in.

And then Metzger says something you should quote to anyone who thinks "not original" means "made up": "the account has all the earmarks of historical veracity. It is obviously a piece of oral tradition which circulated in certain parts of the Western church."²¹ So the story of Jesus saying "Whoever among you is guiltless may be the first to throw a stone at her" and "I do not condemn you either. Go, and from now on do not sin any more"²⁰ is, in the view of the scholar who wrote the book, a real early memory of Jesus that was not part of John's Gospel as John wrote it. The current standard edition, NA28, prints it in double brackets for exactly that reason: kept, and marked.²²

Worked example 2: the Johannine Comma

Now a case with a wrinkle, and I'll leave a step for you.

Open 1 John 5:7–8 in the King James Version:

"For there are three that bear record in heaven, the Father, the Word, and the Holy Ghost: and these three are one. And there are three that bear witness in earth, the Spirit, and the water, and the blood: and these three agree in one." (KJV)²³

Now the NET:

"For there are three that testify, the Spirit and the water and the blood, and these three are in agreement." (1 John 5:7–8, NET)²³

A King James reader has just found the clearest statement of the Trinity in the whole Bible: Father, Word, and Holy Ghost, "and these three are one". A modern reader finds it gone. The NIV's footnote explains, in one line, that the longer reading comes from "Late manuscripts of the Vulgate" and is "not found in any Greek manuscript before the fourteenth century".²⁴ Scholars call the extra words the Johannine Comma ("comma" here meaning a short clause).

Here's the evidence, from Metzger again. The words are "absent from every known Greek manuscript except eight", and he lists them: the earliest that has the words in its actual text is a manuscript from the fourteenth or fifteenth century; the rest are sixteenth century or later, or are older manuscripts where a later hand added the words in the margin.²⁵ The passage "is quoted by none of the Greek Fathers", who would have loved it in the fourth-century Trinitarian controversies if they had known it.²⁵ Its earliest appearance is in a Latin treatise of the fourth century, attributed to the Spaniard Priscillian or a follower, and from there it made its way into copies of the Latin Vulgate.²⁵

How did a Latin gloss get into a Greek printed text and from there into the King James? When Erasmus printed the first published Greek New Testament in 1516, he left the words out, because his Greek manuscripts didn't have them. Under pressure, and once a Greek manuscript containing the words was produced, he included them in his third edition of 1522, and that edition fed the line of printed Greek texts, later called the Textus Receptus, that the King James translators used in 1611.²⁶ The familiar story that he had rashly promised to print them if a single Greek manuscript could be found is itself disputed among historians, so hold it loosely.²⁶ The English Revised Version of 1881 dropped the words, and every major modern translation since has omitted or footnoted them.²⁶

The wrinkle, and the reason this example matters for lesson 11, is that the King James is not "wrong" here in the way a mistranslation is wrong. Its translators rendered the Greek in front of them faithfully. The Greek in front of them was a late text with a Latin gloss in it. That's the difference between a translation problem and a manuscript problem, and knowing which is which will save you from a great many arguments that talk past each other.

What is established, and what is argued

Now sort the whole subject, in the three-way language this course has used since lesson 1.

History establishes three things. The New Testament is by far the best-attested text from the ancient world, in the number, age, and spread of its witnesses. Its earliest recoverable form, which the editors of the standard edition call the initial text, can be established to a very high degree of confidence. And the Masoretic Text of the Old Testament is a faithful copy of a text already old before Christ, while other forms of some books were in circulation then too.

One claim that usually appears on that list needs care about what kind of claim it is. "No article of the creeds depends on a disputed reading" is not a measurement; it is a judgement about what the readings amount to, and Wallace and Ehrman both make it.¹¹¹⁶ Ehrman adds a caveat you should hear: the creeds were framed by the church that copied and chose these books, so the fact that no variant unsettles them is partly a fact about how the creeds were framed. The variants critics point to are real and you can look them up. At 1 Timothy 3:16 the manuscripts divide between "who was revealed in flesh" and "God was revealed in flesh". At John 1:18 they divide between "the only Son" and "the only God". At Hebrews 2:9 they divide between "by the grace of God" and "apart from God" he tasted death. Wallace's reply is that in each case the doctrine is taught elsewhere in undisputed text, so nothing rests on the reading alone. That reply is sound, and it is a reply, not a measurement.

Scholars argue about several things, and you should know them.

Whether early scribes changed the text on purpose for doctrinal reasons, and how often. Ehrman's 1993 book The Orthodox Corruption of Scripture argued that scribes in the second and third centuries sometimes altered readings to defend orthodox Christology against rival views.²⁷ Wallace and Larry Hurtado, among others, replied that most such changes are better explained as accident or harmonisation, that the examples are few, and that the alterations, where real, never succeeded in hiding the original, since we can see them.¹⁷ Both sides agree the changes exist; they argue about motive and scale.

How Mark ended. That the earliest recoverable Mark stops at 16:8 is the consensus; whether Mark intended to stop there, with the women afraid and silent, or whether a page was lost, is not settled and may never be.¹⁹

Whether "the original text" is even the right goal. Eldon Epp and David Parker have argued that for some books there may never have been a single original wording to recover, only early forms that were already being shaped as they were copied, and that the aim should be to study the living text rather than chase a lost autograph.²⁸ Most textual critics still work toward the earliest recoverable text, and the standard edition's editors call it the "initial text" for that reason, but the debate is real.

What happened in the first hundred years, which is the period we can see least of. Our first substantial copies are from around AD 200. The century before that is both the least attested and the one in which copying was least controlled, done by ordinary literate Christians rather than by trained scribes. Ehrman's argument is that the changes we can measure from the third and fourth centuries onward are therefore a floor and not a ceiling. Wallace and Larry Hurtado reply that the early papyri we do have show careful copying going on in that period too, and that the close agreement of P75 with Vaticanus is not what a text being reshaped freely would look like. Both sides are reasoning from the same thin evidence, which is why the argument continues.

Which family of manuscripts stands closest to the autographs. The great majority of scholars hold that the early Alexandrian witnesses do, which is the position this course follows and the one behind every modern translation. A minority defends the priority of the Byzantine or Majority text with serious arguments about how transmission actually works. Lesson 11 gives that case in its own advocates' words.

We hold by faith that these words, recovered as well as human scholarship can recover them, are God's Word. Textual criticism cannot show that and doesn't try. What it shows is the narrower and still remarkable thing: that what we have is, in nearly every line, what they wrote. Whether what they wrote is true is a question lesson 10 takes up.

What people get wrong

"It's been translated so many times it can't be trusted." Two links, not fifty. Modern translations are made from Hebrew and Greek editions built from the manuscripts. The King James differs from your Bible at 1 John 5:7 because of the manuscripts its translators had, not because of translation.

"The Dead Sea Scrolls prove not a word has changed." Half right. The Isaiah Scroll shows a text overwhelmingly stable across a thousand years, and it also shows small differences (you read one at 53:11), and other scrolls show other forms of some books.⁷⁸⁹ Say both halves.

"400,000 variants means the text is hopeless." A variant is any difference in any manuscript, counted once. More manuscripts, more variants. At least three-quarters are spelling and nonsense; under 1 percent are meaningful and viable; and the meaningful ones are known and printed in the margins, not hidden.¹¹

"There are no variants that matter." You've just read three that do, and Ehrman's list is longer. What's true is that none touches a creed. What's also true is that a reader who wants to know what Mark, John, and the author of 1 John wrote needs to know where the text is disputed. The notes in your Bible exist so you can.

"The Bible was written centuries after the events." Lesson 6 handled this for the Gospels: the mainstream critical range puts all four between about AD 65 and 110, within eighty years of the crucifixion, the conservative range is earlier still, and nobody working in the field puts them centuries later. The papyri show them circulating within a century of that. For the Old Testament, be honest that the gap is real and larger. The oldest substantial copies are from the last centuries before Christ, and most of the books describe events centuries earlier than that. How the text was transmitted before the scrolls is a question the scrolls don't answer, and this lesson doesn't pretend they do.

"Sinaiticus and Vaticanus are the Bible." They're the two most important witnesses, not the text. Editors weigh them, and sometimes the papyri or a reading spread across regions outweighs them. You have already read one such case without noticing: at Isaiah 53:11 the NIV followed the Isaiah Scroll and the Septuagint against the Masoretic Text, and in the New Testament the editors regularly print a reading against Sinaiticus where the earlier papyri and the versions agree against it.

Practice

Connections

Lesson 1 told you that chapters and verses were added in the Middle Ages and the sixteenth century. Now you've seen what lies under them: manuscripts with no chapters, no verses, and in the oldest Greek ones no spaces between words. The numbering is a layer; the text is older.

Lesson 6 gave you the dates of the Gospels, both the critical and the conservative ranges. This lesson is where those dates pay off: the shorter the gap between writing and P52, P75, and the great codices, the stronger the case for the wording, and even on the later dates the gap is short.

Lesson 8 gave you the list of books; this lesson gave you the words in them. Lesson 11 explains why the King James and the NIV differ at the three passages, from the translation side: the printed Greek text of 1522 against the editions of today. And lesson 10 asks the question textual criticism cannot answer: granted that this is what they wrote, is it true, and what does it mean to call it God's Word?

Go deeper

• Bruce Metzger and Bart Ehrman, The Text of the New Testament: Its Transmission, Corruption, and Restoration, 4th ed. (Oxford, 2005). The standard textbook, revised by Ehrman from Metzger's classic, by a mainline scholar and the student who became the field's best-known sceptic. This course has not read it directly; it is cited here as the reference the whole field works from.
• Daniel Wallace, Textual Criticism of the New Testament, free lecture course at BiblicalTraining.org. The conservative case at full length, with the counting of variants explained by the man whose numbers this lesson uses.
• Bart Ehrman, The Bart Ehrman Blog, the posts "My Very First Post: Do Textual Variants Matter??" and "Do Textual Variants Really Matter for Anything?". The critical view in its own words, including his agreement that no doctrine is at stake and his argument that doctrine is the wrong test.
• Emanuel Tov, Textual Criticism of the Hebrew Bible, 3rd ed. (Fortress, 2012). The standard reference for the Old Testament side, including the plurality of the text at Qumran and the two editions of Jeremiah. Heavier going than the rest of this list.
• Paul Wegner, The Journey from Texts to Translations (Baker, 1999). The whole story, both testaments, in one readable volume, for anyone who wants it in order.
• Dale Martin, Introduction to the New Testament, lecture 5, "The New Testament as History", Open Yale Courses, free with video and transcript. The critical academic frame for the whole question, from outside this course's standpoint.
• Codex Sinaiticus, the full manuscript online with transcription. Go to the end of Mark and see for yourself where it stops.

Sources

1. Mark 16:6–8 and 16:9, 16, NET. Every scripture quotation in this lesson was re-fetched from the NET at labs.bible.org during the Stage 4 fact-check on 2026-09-08 and matched word for word. Modern translations mark the break before 16:9 with brackets or a note saying that some of the earliest manuscripts do not include 16:9 to 20. Scripture quoted by permission. Quotations designated (NET) are from the NET Bible® copyright ©1996, 2019 by Biblical Studies Press, L.L.C. All rights reserved. See Editorial Standards 4.7 for why this course uses the NET.
2. Bruce M. Metzger and Bart D. Ehrman, The Text of the New Testament: Its Transmission, Corruption, and Restoration, 4th ed. (Oxford University Press, 2005). Materials (papyrus, parchment, scroll, codex; the early Christian preference for the codex); the kinds of scribal error (sight, hearing, memory, judgement; unintentional and intentional changes); the method; the great uncials. This course has not opened the book itself, and the Stage 4 fact-check could not verify a page from it; the claims drawn from it here are the standard account of the field, and the lesson no longer rests its statement about the creeds on Metzger.
3. Ketef Hinnom silver scrolls, found 1979 by Gabriel Barkay in a burial cave at Ketef Hinnom, Jerusalem; two silver amulets carrying a form of the priestly blessing of Numbers 6:24–26; late seventh or early sixth century BC; the oldest surviving biblical text. The scholarly edition is Gabriel Barkay, Marilyn J. Lundberg, Andrew G. Vaughn and Bruce Zuckerman, "The Amulets from Ketef Hinnom: A New Edition and Evaluation", Bulletin of the American Schools of Oriental Research 334 (2004), 41–71.
4. Numbers 6:24–26, NET.
5. Aleppo Codex (c. 930) and Leningrad Codex (1008/1009, the oldest complete Hebrew Bible; the base of the BHS and BHQ printed editions from which nearly every modern Old Testament is translated); the Masoretes. On the Aleppo Codex's losses: roughly 190 of an estimated 487 leaves are missing, and the standard account that they burned in the 1947 riot is not supported by evidence of fire; the leaves are now generally thought to have disappeared at some point between 1947 and 1958, and how remains argued. The Hebrew University Bible Project and the Jerusalem Crown edition work from Aleppo rather than Leningrad. Checked at Stage 4, 2026-09-08.
6. Israel Antiquities Authority, Leon Levy Dead Sea Scrolls Digital Library, "Introduction" (deadseascrolls.org.il), read at draft time: "About 230 manuscripts are referred to as 'biblical Scrolls'"; "partial or complete copies of every book in the Hebrew Bible (except the book of Esther)"; scrolls found "in eleven caves near the site of Qumran"; "Scroll dates range from the third century bce ... to the first century of the Common Era".
7. The Great Isaiah Scroll, 1QIsa^a: the only complete biblical book among the scrolls, all sixty-six chapters on sheets sewn into one roll; c. 125 BC by palaeography, radiocarbon c. 350–100 BC; about a thousand years older than the Masoretic codices and agreeing with them overwhelmingly, with mostly orthographic and minor differences; other Qumran manuscripts (Jeremiah, Samuel) in Septuagint-like and other forms. As recorded in SOURCES.md.
8. Emanuel Tov, Textual Criticism of the Hebrew Bible, 3rd ed. (Fortress, 2012). The textual plurality of the Second Temple period; the Masoretic Text as one of several forms then in circulation.
9. Isaiah 53:11 in the ESV ("he shall see and be satisfied"; footnote: "Masoretic Text; Dead Sea Scroll he shall see light") and the NIV ("he will see the light of life and be satisfied"; footnote: "Dead Sea Scrolls (see also Septuagint); Masoretic Text does not have the light of life."). Both read at BibleGateway at draft time; this verifies the 1QIsa^a and Septuagint reading flagged in OUTLINE.md.
10. Institut für Neutestamentliche Textforschung (Münster), Kurzgefasste Liste, the register itself at https://ntvmr.uni-muenster.de/liste (resolves; checked 2026-09-08). Four classes (papyri, majuscules, minuscules, lectionaries). Published totals differ with how deletions and merges are counted: Wallace gave "over 5800" in 2012 (source 11), the Center for the Study of New Testament Manuscripts gave "over 5,600" in 2022. The Stage 4 fact-check could not reach an INTF page stating 5,800, so this lesson gives the figure as a range and rests the claim on Wallace's quoted words.
11. Daniel B. Wallace, interviewed by Justin Taylor, "An Interview with Daniel B. Wallace on the New Testament Manuscripts", The Gospel Coalition, 22 March 2012, at https://www.thegospelcoalition.org/blogs/justin-taylor/an-interview-with-daniel-b-wallace-on-the-new-testament-manuscripts/ (re-fetched and every quoted string confirmed on the page, 2026-09-08). "As far as Greek manuscripts, over 5800 have been catalogued." "Spelling and nonsense readings are the vast majority, accounting for at least 75% of all variants." "The smallest category by far is the last category: meaningful and viable variants. These comprise less than 1% of all textual variants." "Even here, no cardinal belief is at stake. These variants do affect what a particular passage teaches, and thus what the Bible says in that place, but they do not jeopardize essential beliefs." This verifies the "under 1 percent" figure flagged in SOURCES.md as via secondary.
12. John Rylands Library, Manchester Digital Collections, "Greek Papyrus 457" (P52), read at draft time: John 18:31–33 (recto) and 37–38 (verso); "This small scrap of papyrus only measures 8.9 x 6.0 cm"; "The first editor dated the Fragment to the first half of the second century (between 100-150 CE)"; "Recent research points to a date nearer to 200 CE."
13. Brent Nongbri, "The Use and Abuse of P52: Papyrological Pitfalls in the Dating of the Fourth Gospel", Harvard Theological Review 98 (2005), 23–48. The hand of P52 fits anywhere from the early second to the early third century. As recorded in SOURCES.md.
14. P46 (Chester Beatty Library, Dublin, and University of Michigan; Pauline letters; c. 200) and P75 (Luke and John; c. 200–225), dates as recorded in SOURCES.md from the Chester Beatty, Michigan, and standard descriptions. The close textual relationship between P75 and Vaticanus is the standard finding of the field, reported in Metzger and Ehrman (source 2); the Stage 4 fact-check could not open the primary study and says so rather than passing it.
15. Codex Sinaiticus (fourth century; British Library and other holdings; the oldest complete New Testament; full digital edition at codexsinaiticus.org, where the page showing Mark 15:16–16:1 was opened at draft time) and Codex Vaticanus (fourth century; Vatican Library). As recorded in SOURCES.md.
16. Bart D. Ehrman, "My Very First Post: Do Textual Variants Matter??", The Bart Ehrman Blog, 31 March 2019, at https://ehrmanblog.org/my-very-first-post-do-textual-variants-matter/ (re-fetched 2026-09-08). "none of your cherished doctrines appears to be in real danger because of variations in our surviving manuscripts (at least the variations that we know about)"; the parenthesis is Ehrman's and this lesson quotes it, because the concession is his to qualify. "changes in the Bible can be inordinately significant without affecting any cherished doctrines of the evangelicals". The same post says that "there are numerous passages where scholars continue to debate what the 'original' text of the New Testament said; and there are some places where we will never know", which is why the quiz explanation states his position that way. The often-quoted sentence from the 2005 paperback appendix of Misquoting Jesus ("does not actually stand at odds with Prof. Metzger's position...") could not be checked against the printed page at draft time and is therefore not quoted here; the blog wording is Ehrman's own and says the same thing.
17. Daniel B. Wallace, "The Bart Ehrman Blog and the Reliability of the New Testament Text", danielbwallace.com, 1 May 2012 (read at draft time). The four points of agreement: "we both agree that the scribes changed the texts hundreds of thousands of times, that certain passages are not authentic (notably, the longer ending of Mark and the story of the woman caught in adultery), that the proto-orthodox changed the text at times, and that only a small fraction of all textual changes are both meaningful and viable." Also Wallace's figure that even late Byzantine manuscripts differ from the reconstructed text "only about 4%–5% of the time", and his count of more than 500 Greek New Testament manuscripts written before the tenth century. Both are used in the lesson. The 400,000 total is Wallace's, and is cited to its own source at note 29 rather than to this page.
18. Bart D. Ehrman, "Do Textual Variants Really Matter for Anything?", The Bart Ehrman Blog, 29 November 2014 (read at draft time). "hundreds of thousands of textual variants", most "completely unimportant and insignificant"; "Why is it that 'threats to significant doctrines' are taken to be the only standard of what is important to the Christian religion? In my view this idea is completely wrong-headed."; the examples of Luke's baptismal voice at Luke 3:22 ("today I have begotten you", bearing on whether Luke's Jesus is God's Son eternally or becomes so at his baptism), Luke's Last Supper words at 22:19b–20 (bearing on whether Luke has a doctrine of atonement there at all), the ending of Mark, and the adulteress passage. The Stage 4 fact-check could not find Mark 1:41 in this post; that example is cited to note 30 instead.
19. Bruce M. Metzger, A Textual Commentary on the Greek New Testament, 2nd ed. (United Bible Societies, 1994), on Mark 16:9–20, as reproduced at bible-researcher.com/endmark.html (read at draft time). "absent from the two oldest Greek manuscripts (א and B), from the Old Latin codex Bobiensis (it k), the Sinaitic Syriac manuscript, about one hundred Armenian manuscripts, and the two oldest Georgian manuscripts"; "The vocabulary and style of verses 9-20 are non-Markan"; "The connection between ver. 8 and verses 9-20 is so awkward that it is difficult to believe that the evangelist intended the section to be a continuation of the Gospel"; "on the basis of good external evidence and strong internal considerations it appears that the earliest ascertainable form of the Gospel of Mark ended with 16.8"; the shorter ending, and manuscripts containing both.
20. John 7:53–8:11, NET; John 8:7 and 8:11 quoted. Modern translations bracket or footnote the passage as absent from the earliest manuscripts; the NA28 edition prints it in double brackets.
21. Metzger, Textual Commentary, on John 7:53–8:11, as reproduced at bible-researcher.com/adult.html (read at draft time). "The evidence for the non-Johannine origin of the pericope of the adulteress is overwhelming"; absent from P66, P75, Sinaiticus, Vaticanus, and others; "Most copyists apparently thought that it would interrupt John's narrative least if it were inserted after 7.52... Others placed it after 7.36... or after 21.25... or after Luke 21.38"; "The style and vocabulary of the pericope differ noticeably from the rest of the Fourth Gospel"; "the account has all the earmarks of historical veracity. It is obviously a piece of oral tradition which circulated in certain parts of the Western church."
22. Nestle-Aland Novum Testamentum Graece, 28th ed. (2012) and UBS Greek New Testament, 5th ed.: John 7:53–8:11 and Mark 16:9–20 printed in double brackets. As recorded in SOURCES.md.
23. 1 John 5:7–8 in the King James Version (1611), which is in the public domain, and in the NET.
24. 1 John 5:8, NIV footnote: "Late manuscripts of the Vulgate testify in heaven: the Father, the Word and the Holy Spirit, and these three are one. And there are three that testify on earth: the (not found in any Greek manuscript before the fourteenth century)". Read at BibleGateway.
25. Metzger, Textual Commentary, on 1 John 5:7–8, as reproduced at bible-researcher.com/comma.html (read at draft time). "absent from every known Greek manuscript except eight": 61 (early sixteenth century); 88 (a sixteenth-century hand added to a fourteenth-century codex); 221 (added to a tenth-century manuscript); 429 (added to a sixteenth-century manuscript); 629 (fourteenth or fifteenth century); 636 (added to a sixteenth-century manuscript); 918 (sixteenth century); 2318 (eighteenth century); "quoted by none of the Greek Fathers"; earliest in a fourth-century Latin treatise attributed to Priscillian (died about 385) or his follower Instantius.
26. Erasmus's Greek New Testament: first edition 1516 without the Comma; third edition 1522 with it, from which it entered the Textus Receptus and the King James Version (1611); omitted from the English Revised Version (1881) and every major modern translation since. As recorded in SOURCES.md.
27. Bart D. Ehrman, The Orthodox Corruption of Scripture: The Effect of Early Christological Controversies on the Text of the New Testament (Oxford University Press, 1993). The argument that some early variants were doctrinally motivated. As recorded in SOURCES.md; the replies by Wallace and Hurtado as recorded there under contested question 6.
28. Eldon J. Epp and David C. Parker on whether "the original text" is a recoverable or well-defined goal (Epp, "The Multivalence of the Term 'Original Text' in New Testament Textual Criticism", Harvard Theological Review 92 (1999); Parker, The Living Text of the Gospels (Cambridge, 1997)), and the Nestle-Aland editors' term "initial text" (Ausgangstext). As recorded in SOURCES.md under contested question 6. Both the Epp article (HTR 92, 1999, 245–281) and Parker's book were confirmed at Stage 4, 2026-09-08.
29. Daniel B. Wallace, "The Number of Textual Variants: An Evangelical Miscalculation", danielbwallace.com, 9 September 2013, at https://danielbwallace.com/2013/09/09/the-number-of-textual-variants-an-evangelical-miscalculation/ (fetched 2026-09-08): "with the work done on Luke's Gospel by the International Greek New Testament Project, Tommy Wasserman's work on Jude, and Münster's work on James and 1-2 Peter, the estimates today are closer to 400,000." The 400,000 figure is Wallace's; the lesson no longer attributes it to Ehrman, who gives a range of 200,000 to 400,000.
30. Bart D. Ehrman, Misquoting Jesus (HarperSanFrancisco, 2005), chapter 5, for the Mark 1:41 variant (Jesus "moved with compassion" or "angry" as he heals the leper) and for the 200,000 to 400,000 range. Cited from the field's standard reporting of the book rather than from the printed page, which this course has not opened.
31. Luke 22:43–44 and its notes, used in the quiz and the exercise. ESV footnote at Luke 22:43, "Some manuscripts omit verses 43 and 44"; the NET prints the verses in single brackets. Both read at Stage 4, 2026-09-08.
32. Reasoned eclecticism and its rivals: thoroughgoing eclecticism (G. D. Kilpatrick, J. K. Elliott) and Byzantine priority (Maurice Robinson, Zane Hodges, William Pierpont). Named here so the lesson does not present the majority method as the only one; the Byzantine and Received Text case is given at length in lesson 11 from sources read in full there.