The Evidence Room · The Copy Room
Summary
We do not have the originals, and the copies do differ. Bruce Metzger and Bart Ehrman (The Text of the New Testament, 4th edition, 2005) class the great majority of the differences as spelling, word order and obvious slips, and the readings that change a meaning are printed with their evidence in the Nestle-Aland and United Bible Societies editions, which is the reason we can say anything about them at all.
THE CLAIM
The copies differ in spelling and slips, not in the message.
WHERE IT STANDS
Nothing hidden: the disputed passages are printed in the open.
high confidenceNobody has the page Paul wrote on. Nobody has the scroll Isaiah dictated. What survives are copies, and copies of copies, made by hand for well over a thousand years before anyone had a printing press.
For most ancient books that is bad news. What makes the Bible unusual is how many copies survived, how early some of them are, and how well we can now compare them.
This page is about counting carefully, which means it will disappoint anyone hoping for the big round numbers. Several of the figures you have heard from a pulpit are too big, and we will say which ones and why.
That is not a concession. It is the whole argument. A site that publishes the smaller, defensible number is a site you can believe about the rest.
5,100–5,300
Greek New Testament manuscripts actually surviving today: a survival count, not the registry total that includes lost, destroyed, and double-numbered items
2,600+
Differences between the Great Isaiah Scroll and the later standard Hebrew text: overwhelmingly spelling, word order, and single words
Emanuel Tov, Textual Criticism of the Hebrew Bible, 3rd ed. (Fortress, 2012)
c. 125 BC
Date of the Great Isaiah Scroll, which puts a complete Isaiah roughly a thousand years earlier than the manuscripts scholars worked from before 1947
AD 125–175
Palaeographic range for P52, the Rylands fragment of John: the earliest surviving scrap of any New Testament book, on the current dating rather than the older one
c. 125 BC
Shrine of the Book, Israel Museum, Jerusalem · Qumran Cave 1, 1947
The whole of Isaiah, a thousand years older than the oldest Hebrew Bible known before it — and Isaiah 53, the suffering servant, demonstrably in circulation long before Jesus of Nazareth.
The exact reading
It differs from the medieval text in spelling and in hundreds of small readings; none of them changes what chapter 53 says.
2nd century AD
John Rylands Library, Manchester · Egypt; identified 1934
A credit-card-sized scrap of John 18 — Pilate and Jesus, "What is truth?" — copied in Egypt, far from where John was written, within a century or so of the writing.
The exact reading
Roberts dated it c. AD 125 in 1935. Nongbri showed in 2005 that the handwriting samples used for comparison actually span the whole second century; the honest range is wider than the famous date.
c. AD 330–360
British Library, London (with Leipzig, St Catherine's, St Petersburg) · St Catherine's Monastery, Sinai, 1844–59
The oldest complete New Testament in existence, written on parchment in four columns a page, with the Old Testament in Greek alongside it.
The exact reading
It also contains the Epistle of Barnabas and the Shepherd of Hermas, and it ends Mark at 16:8 — the manuscript shows the canon and the text still being handled, not a finished printed book.
late 7th century BC
Israel Museum, Jerusalem · Ketef Hinnom, Jerusalem, 1979
Two rolled silver amulets, worn around a neck, carrying the priestly blessing of Numbers 6:24–26. They are the oldest known copies of any biblical text — some four centuries older than the Dead Sea Scrolls.
The exact reading
They carry a blessing, not a book: the find dates the wording, not the whole Pentateuch. And the amulets run shorter than Numbers 6:24–26 — the clauses about God being gracious and lifting up his countenance are not on the silver.
The originals are not hidden and not suppressed: they are simply gone. Papyrus rots and scrolls wear out over time. The autographs of every ancient work, biblical or not, are lost.
So for any ancient text, the real question is not do we have the original? It is whether we can work back to what the original said from the copies we do have.
That is a normal historical task with normal tools, and the Bible happens to be the best-equipped case in the ancient world for doing it.
For the New Testament in Greek, the careful figure is roughly 5,100 to 5,300 manuscripts actually surviving today.
That number comes from Jacob Peterson, writing in Myths and Mistakes in New Testament Textual Criticism (IVP Academic, 2019), a book written by evangelical text critics, precisely to clean up the numbers their own side had been repeating.
The larger figure you usually hear is a registry count, not a survival count. The official register assigns a number to items that were later destroyed, items that have been lost, items registered twice under different numbers, and single scraps the size of a postage stamp. Those numbers stay on the list; the manuscripts do not stay in the world.
One more thing honesty requires: most of these manuscripts are medieval, from the tenth century or later. The early ones are a small minority, and a precious one.
Even after all that trimming, the result is remarkable. Set beside other texts from the ancient world, the New Testament has more surviving copies, and earlier ones, than almost anything else from the period.
And the figure above is a survival count (manuscripts you could go and photograph) rather than the registry number that inflates the usual apologetics chart. Every count on this page is of that kind.
A variant is any place where the manuscripts do not read the same. That sounds ominous until you look at what fills the category.
Most of them are spelling: the same name written two ways, the way English tolerates both grey and gray. A very large share are word order, which matters far less in Greek than in English because Greek marks a word's job with its ending rather than its position. Many are plain slips: a scribe's eye jumps from one line to a similar line, or a word gets written twice.
A much smaller group are meaningful (they change the sense) and a smaller group still are both meaningful and viable, meaning they have a real chance of being what the author wrote.
And here is the part that gets skipped: those are known. They are catalogued, argued about in print, and listed at the bottom of every scholarly Greek New Testament in a set of notes called the apparatus. Any reader can look them up.
Which is the actual argument. We can describe the problem this precisely only because the evidence is abundant enough to describe it. A text with three surviving copies has fewer variants and far less certainty.
Two well-known passages are almost certainly not in what the authors wrote, and both are usually in your Bible with a footnote or a set of brackets.
The first is the ending of Mark, 16:9–20. The oldest and best manuscripts stop at verse 8, with frightened women and no appearance. Several different endings circulated afterward, which is itself a sign that early readers found the abrupt stop unsatisfying.
The second is the story of the woman caught in adultery, John 7:53–8:11. It is missing from the earliest manuscripts, and in some later ones it appears in a different place entirely (after John 21, or over in Luke), which is how a floating tradition behaves.
Notice that you already knew this, or could have, from the note in your own Bible. Nothing was covered up. The note that every critical edition carries was printed in the text you were handed at church.
This site takes the same view: neither passage is a secret, and neither carries a doctrine that is not taught elsewhere in Scripture.
The earliest identified fragment of any New Testament book is a piece of John's Gospel about the size of a credit card, held at the John Rylands Library in Manchester and known as P52. It has a few verses of John 18 on both sides.
You have probably heard it dated to AD 125. That single confident year is the most-repeated false precision in Christian apologetics, and we stopped using it.
Here is what is actually on the table. The original editor, C.H. Roberts, proposed 100–150. Later palaeographers Orsini and Clarysse proposed 125–175. The Rylands Library's own current note says recent research points to a date nearer to AD 200.
Brent Nongbri showed why the tight date was never safe: the comparison manuscripts had been chosen in a circle, and he produced securely dated papyri from the later second and early third century with equally close letter forms. His conclusion was that any serious estimate of the range must include those later dates.
So here is where we land. A scrap of John's Gospel, copied in the second century: possibly early, possibly late, and the library that owns it now leans late. It is still the earliest identified fragment of any New Testament book.
Why is the range the more impressive claim? Because a range is what a careful person says. Handwriting dating is an estimate, not a measurement, and a source that gives you a single year for it is telling you something about itself.
In 1947 a scroll of Isaiah came out of a cave near the Dead Sea, copied around 125 BC, about a thousand years older than the Hebrew manuscripts scholars had been working from.
It carries more than 2,600 differences against the later standard Hebrew text. Almost every one is a spelling, a word order, a singular against a plural, or a single word; a few are larger.
That count is Emanuel Tov's, in the standard handbook of Hebrew textual criticism, and it is the figure worth carrying rather than the round percentage that circulates without an author or a method behind it. Two thousand six hundred differences across a thousand years, almost all of them spelling and word order. It is the same book.
Now the half most presentations leave out. The scrolls do not only show stability. They also show that more than one edition of some books was in circulation.
At Qumran there is a short, differently arranged Jeremiah, matching the kind of text behind the old Greek translation rather than the standard Hebrew one. And a Samuel scroll, 4QSamuel-a, preserves readings that modern translations now print, including a paragraph about King Nahash that had dropped out of the standard text and is restored in several English Bibles today.
Both halves are true, and a reader is owed both. The manuscripts show a text transmitted with remarkable care, and they show that in some books the ancient situation was more varied than a simple story allows.
The grass withers and the flowers fall, but the word of our God stands forever.”
Isa 40:8→
“Then Jeremiah took another scroll and gave it to the scribe Baruch son of Neriah, and at Jeremiah’s dictation he wrote on it all the words of the scroll that Jehoiakim king of Judah had burned in the fire. And many similar words were added to them.”
Jer 36:32→
Each objection below is stated the way the people who hold it state it, at the same type size as the answer. If you hold one of these and think it is put unfairly here, tell us.
THE ANSWER
Two parts, and the first is a straight concession. The count is real and Christians should stop being surprised by it. It is what abundance produces: more copies means more variants, and a book preserved in three manuscripts would have a much prettier number and a much worse text.
The second part is where the objection is answerable. Ehrman's own textbook practice (he is a working text critic, not only a popular writer) is to weigh readings by manuscript age, geographical spread, and which reading better explains the origin of the others. Those are the ordinary tools, they work, and applying them is how we know which variants are trivial.
On the deliberate theological changes: he is right that some exist, and this page will not pretend otherwise. What is disputed is their weight. The changes he documents are mostly cases where a later scribe made an existing orthodox reading more explicit, which is a real phenomenon and a very different thing from inventing a doctrine.
And every one of the passages under discussion is printed in the apparatus of every critical edition, argued in public, and available to any reader who wants to check. A cover-up that publishes its own evidence is not much of a cover-up.
THE ANSWER
Two things are true here and they point in different directions. It is true that there were other writings, that there were real disputes, and that some books took a long time to be universally accepted: Hebrews, James, 2 Peter, Jude, and Revelation were all argued over. Anyone who tells you the canon was obvious from the start is not being straight.
What is not true is the suppression story as usually told. The other gospels are, with few exceptions, demonstrably later (second century and after) and they can be dated on the same grounds as anything else. They are also available: the Nag Hammadi library has been published and translated, and you will find it in any good library, we will point you to it.
Read the Gospel of Thomas and then read Mark, and the question of why one circulated and the other did not becomes less mysterious. The lists we have from the second, third, and fourth centuries show recognition happening gradually and in many places at once, not a decision handed down at a council. Two specific myths are worth retiring here: the canon was not settled at Nicaea, and it was not instantly universal.
THE ANSWER
The analogy has one flaw that undoes it. In telephone, there is a single chain, each message is whispered once, and the earlier versions are gone. Manuscript transmission is nothing like that.
It is a branching tree, in many countries and languages at once, and (this is the crucial part) the earlier versions did not disappear. When a fourth-century Egyptian manuscript, a fifth-century Syriac translation, and a second-century quotation in a church father all agree against a twelfth-century Byzantine copy, the error is visible and correctable. You cannot compare in telephone; you can compare here.
That said, the objection is right that drift is real. It is why some passages carry footnotes, and it is why the discipline of textual criticism exists at all.
THE ANSWER
Most of what a reader notices comes from translation philosophy rather than from the manuscripts. One team renders word by word, another renders thought by thought, and a third aims for reading rhythm; the same Greek sentence lands three different ways. That is a choice about English, not a doubt about Greek.
A smaller share of the differences are textual, and those are exactly the cases discussed on this page, and they are footnoted, which is why you noticed them. The missing-verse experience is almost always this: an older English tradition followed later Greek manuscripts, newer translations follow earlier ones, and the newer edition tells you in a note what it did and why. Compare any two translations across a whole chapter and the striking thing is not the divergence.
It is how much they say the same thing.
THE ANSWER
This objection is largely right, and we print that fuller picture above in the walk-through instead of burying it in a footnote. Any claim of pure stability that leaves that part out should not be trusted. Where the disagreement remains is what follows from it.
That some books circulated in more than one edition is a fact about the ancient situation; it is not evidence that the text was casually altered or that its contents are unknowable, and it is not distributed evenly. Isaiah, and the consonantal frame of the Torah, are far more uniform than Jeremiah or Samuel, while the Psalms run the other way, since the great Psalms scroll from Cave 11 reorders the last third of the book and mixes in compositions the Hebrew Bible does not carry. The fair summary is that the scrolls show both careful transmission and genuine variety, that scholars can now see which books show which, and that a reader is owed the whole picture rather than the flattering half of it.
WHERE IT STANDS
Nothing hidden: the disputed passages are printed in the open.
The originals are lost, as they are for every ancient book. What survives is roughly 5,100 to 5,300 Greek manuscripts of the New Testament (the careful survival count, not the larger registry total) plus early translations and quotations, and most of them are medieval.
The differences among them are real and numerous, and they are overwhelmingly spelling, word order, and copying slips. The small number that change the sense are known, named, and printed in every scholarly edition, which is why anyone can talk about them at all.
The two famous long passages (the ending of Mark and the woman caught in adultery) are shown here rather than hidden, and your own Bible almost certainly footnotes them already.
On the Hebrew side, the Great Isaiah Scroll from around 125 BC differs from the later standard text in more than 2,600 places, almost all of them spelling, word order, and single words. It is the same book across a thousand years. And the same caves show that some books circulated in more than one edition, which is the half of the story most presentations leave out.
Every number above carries the name of the scholar who counted it and the edition it was counted in. That is the standard the numbers table at the head of this page keeps, and it is the standard to hold any manuscript claim to, including the ones made in this argument's favour.
What would overturn this A manuscript tradition showing a teaching entering the text late, or a variant that changes what a book teaches and is not already printed in the critical editions.
The limits
Textual work reaches the earliest wording the copies can support. It does not reach the page the author wrote on.
What would settle it An autograph, or a copy from the gap between composition and the earliest surviving manuscripts of that book.
Some Old Testament books circulated at Qumran in more than one edition, and how far that ran is open.
What would settle it Further scrolls from the same caves in the same books, published with their variants.
10 works, each named in full. Follow one into the registry to see everything else on the site that leans on it.
The careful survival count of Greek manuscripts, published by evangelical text critics.
The re-dating of the Rylands fragment.
The original edition and dating of P52.
The standard treatment of textual plurality at Qumran.
4QSamuel-a and the Nahash paragraph.
The variant-count objection in its author's own framing.
The openly published record of the variants themselves.
On transmission and the reader's own ability to check.
Images linked, never rehosted; go and look at the manuscripts yourself.
Last reviewed
Found an error? Write to us and name the line.
