Textual criticism — do we have what was originally written?

Textual criticism is the discipline of working out what a document originally said, given that we have copies rather than originals.

It applies to every work from the ancient world. Nobody has Caesar's own manuscript of the Gallic Wars or Plato's autograph. What survives is copies of copies.

For the New Testament, this discipline is unusually well supplied.

The scale of the evidence

Over 5,800 Greek manuscripts of all or part of the New Testament. Around 10,000 Latin manuscripts. Roughly 9,300 in Syriac, Coptic, Armenian, Georgian, Ethiopic and other languages. Over one million quotations in the writings of the early church fathers — so many that the New Testament could be almost entirely reconstructed from those citations alone if every manuscript vanished.

Set that against other ancient works:

Nobody disputes what Tacitus wrote. He survives in two manuscripts.

The earliest copies

P52 — a credit-card-sized fragment of John 18, in the John Rylands Library in Manchester, dated to roughly AD 125–150. Within perhaps forty years of composition.

P46 — most of Paul's letters, around AD 200.

P66 and P75 — substantial portions of John and Luke, around AD 200.

Codex Sinaiticus and Codex Vaticanus — near-complete Bibles on parchment, fourth century.

Sinaiticus has a story. Constantin von Tischendorf found leaves of it at St Catherine's Monastery on Sinai in 1844. The account that monks were burning it for fuel appears to be his own dramatisation; the monastery disputes it. What is not disputed is that the manuscript is fourth-century and a thousand years older than anything Erasmus worked from.

The 400,000 variants

This is the number sceptics cite, and it needs explaining rather than denying.

A variant is any difference between manuscripts at a given point — a different spelling, a changed word order, an omitted word, an added phrase.

The estimate is roughly 400,000 across the whole tradition. That is more than the number of words in the New Testament.

Here is why the number is large: variants are counted per manuscript. If one scribe in the third century spelled a name differently and his copy was reproduced 3,000 times, that is 3,000 variants — from one slip of the pen.

The number is a direct function of how many manuscripts survive.

What the variants actually are

Daniel Wallace, who has spent his career on this, breaks them down roughly as follows:

About 70–75% are spelling differences. Chiefly the nu movable — the Greek equivalent of writing "a" versus "an" — which cannot even be represented in translation.

Another large portion are word order. Greek is an inflected language, so word order does not determine meaning as it does in English. "Jesus loves Paul" and "Paul loves Jesus" would be distinguished by case endings, not sequence. Reordering changes emphasis, not sense.

Synonyms and articles. "Jesus" versus "the Lord" versus "he".

Obvious errors. Skipped lines, duplicated words, misheard dictation.

Meaningful but not viable. Variants that would change meaning but appear in a single late manuscript with no chance of being original.

Meaningful and viable: fewer than 1%. And of those, none affects a cardinal doctrine.

The passages you have heard about

Mark 16:9-20 — the longer ending. Absent from Sinaiticus and Vaticanus; the vocabulary differs from the rest of Mark. Most scholars judge it a later addition. Modern Bibles include it with a note. Note that it is also the source of the snake-handling practice in some churches, which is a good illustration of why this matters.

John 7:53–8:11 — the woman caught in adultery. Absent from the earliest manuscripts and appearing in different locations in others. Almost certainly not original to John — and quite possibly a genuine incident preserved in early oral tradition, which is why translators are reluctant to remove it and why the church has treasured it.

1 John 5:7-8 — the Comma Johanneum, "the Father, the Word, and the Holy Ghost: and these three are one." This one is not a difficult judgement: it appears in no Greek manuscript before the fourteenth century. Erasmus initially excluded it and, under pressure, included it in a later edition after a manuscript containing it was produced — possibly made to order. It is not in modern translations, and the doctrine of the Trinity does not need it.

Christians should mention this passage first in a conversation about textual criticism. Conceding it immediately demonstrates that the case does not depend on disputed texts, which makes everything else more credible.

How scholars decide

The criteria are not mysterious.

External evidence: How old are the manuscripts supporting each reading? How geographically widespread? How reliable are the manuscript families involved?

Internal evidence — two counter-intuitive principles:

Prefer the shorter reading. Scribes were more likely to add explanatory material than to delete it.

Prefer the more difficult reading. A scribe encountering something confusing was more likely to smooth it than to roughen it. So the awkward reading is generally the earlier one.

Which reading best explains the others? If one reading accounts for how the alternatives arose, it is probably original.

This work is public. The critical apparatus in a Nestle-Aland Greek New Testament lists the variants and the manuscripts supporting each. Anyone can check the reasoning. Nothing is hidden.

And your English Bible's footnotes — "some manuscripts add..." — are that apparatus, simplified.

The Old Testament

The Hebrew text was transmitted by scribes whose care is difficult to overstate. The Masoretes, working between roughly AD 600 and 1000, counted the letters of each book, identified the middle letter, and recorded the frequency of each word — checksums, centuries before the concept existed.

The result was that before 1947, the oldest complete Hebrew manuscripts dated to about AD 1000.

Then a Bedouin shepherd threw a stone into a cave at Qumran.

The Dead Sea Scrolls pushed the evidence back a thousand years, to around 100 BC. The great Isaiah scroll — the whole book, complete — could be compared directly with the medieval text.

The result: substantially identical. Differences of spelling, the occasional word, nothing that changes meaning. A thousand years of hand-copying, and the text held.

What this establishes and what it does not

It does not prove the Bible is true. Establishing that we have accurately what was written says nothing about whether what was written is correct. Do not overclaim; a sceptic will notice.

It removes a specific objection. The claim that the New Testament has been so altered that its original content is unrecoverable is not sustainable on the evidence, and the people best placed to know say so.

Bruce Metzger, the leading textual scholar of the twentieth century, and Bart Ehrman, his student and the most prominent modern sceptic, agree on the data. They differ on what it means.

Ehrman's own appendix to Misquoting Jesus concedes that essential Christian beliefs are not affected by textual variants and that most of the changes are careless or insignificant.

And the thing itself

"Heaven and earth will pass away, but my words will not pass away" (Matthew 24:35).

There is an irony in how that preservation happened. Not one flawless manuscript kept under guard, which could be lost, destroyed or quietly altered.

Instead: thousands of imperfect copies, in dozens of languages, scattered across three continents, in the hands of communities that frequently disagreed with each other and had no means of coordinating.

Which is precisely why nobody could ever change it.