New Testament Manuscripts
The New Testament survives through manuscripts copied long after its events, with no originals and thousands of textual differences.
6 min readUpdated
About 5,800 Greek manuscripts of the New Testament survive, more than for any other work of antiquity. No original document survives, and no copy made from an original survives.
This page sets out what the manuscript evidence consists of, when it dates from, and what follows.
The manuscripts by type
Catalogued by the Institute for New Testament Textual Research in Münster, which maintains the standard register.
| Type | Approximate count | Usual date range |
|---|---|---|
| Papyri | About 140 | 2nd–8th century |
| Majuscules (uncials) | About 320 | 4th–10th century |
| Minuscules | About 2,900 | 9th–16th century |
| Lectionaries | About 2,450 | 9th–16th century |
The great majority date from the 11th century onward — roughly a thousand years after composition. Most early evidence is fragmentary.
The earliest witnesses to the Gospels
| Manuscript | Date | Extent | Content |
|---|---|---|---|
| P52 | c. 125–175 | Fragment, roughly credit-card sized | John 18:31-33, 37-38 |
| P104 | c. 150–200 | Fragment | Matthew 21:34-37, 43, 45 |
| P137 | c. 150–250 | Fragment | Mark 1:7-9, 16-18 |
| P75 | c. 175–225 | Substantial | Portions of Luke 3–18, 22–24, and John |
| P66 | c. 200 | Substantial | Most of John |
| Codex Vaticanus | c. 300–325 | Near-complete Bible | Gospels complete; ends at Hebrews 9:14 |
| Codex Sinaiticus | c. 330–360 | Complete New Testament | All four Gospels, plus Barnabas and Hermas |
P52 is the earliest identified fragment of any New Testament text. It preserves about a hundred Greek letters across both sides — a few verses of John's trial scene — and its date is estimated from handwriting style, a method with a margin of several decades in either direction.
P137 was publicized for several years as a first-century fragment of Mark before publication; when it appeared in 2018 as P.Oxy. 5345, the editors dated it to the late second or early third century.
The earliest witnesses to Paul
| Manuscript | Date | Content |
|---|---|---|
| P46 | c. 175–225 | Romans, Hebrews, 1–2 Corinthians, Ephesians, Galatians, Philippians, Colossians, 1 Thessalonians |
| P32 | c. 200 | Titus 1:11-15, 2:3-8 |
| P87 | c. 250 | Philemon 13-15, 24-25 |
| Codex Sinaiticus | c. 330–360 | All the canonical Pauline letters |
P46 is the most substantial early witness to Paul. It omits the Pastoral Epistles — 1 and 2 Timothy and Titus — which are the three letters most widely regarded as written by someone other than Paul. See Paul's Letters.
The earliest witnesses to the rest
| Manuscript | Date | Content |
|---|---|---|
| P29 | c. 175–225 | Acts 26:7-8, 20 |
| P47 | c. 250–300 | Revelation 9:10–17:2 |
| P72 | c. 300–350 | 1–2 Peter, Jude, with several non-canonical works |
Variants
Estimates of the number of differences among the manuscripts range from about 300,000 to 500,000. Peter Gurry's 2016 study put the figure near 500,000 excluding differences in spelling.
The number is larger than the New Testament's word count, which is about 138,000 words in Greek. This is a consequence of counting across roughly 5,800 manuscripts: a single word differing in a thousand copies contributes a thousand variants.
The overwhelming majority are spelling differences, word-order changes that do not affect meaning in Greek, and obvious copying slips. A smaller number affect the sense of a passage, and a few affect passages of doctrinal significance. Those are collected in Interpolations.
Common objections
"There are more manuscripts than for any other ancient work"
This is true. Caesar's Gallic War survives in perhaps a few hundred manuscripts, the earliest around 900 years after composition; Tacitus and Thucydides are attested more thinly still.
The comparison establishes that the New Testament is unusually well attested by the standards of ancient literature. That is a real and relevant fact.
It does not establish what it is often used to support. Historians do not treat Caesar as free of error because his text is well transmitted — transmission and reliability are separate questions. A perfectly preserved text can be wrong about what happened, and the comparison shows only that the wording is recoverable, not that the contents are accurate.
The count also includes the medieval majority. For any specific disputed passage, what matters is the number of early and independent witnesses to that passage, which is often small.
"Ninety-nine percent of the variants are trivial"
Substantially correct. Most are spelling, word order, or evident slips of the pen, and Greek word order carries less meaning than English, so many reorderings change nothing.
Two qualifications. The figure is a proportion, and one percent of 400,000 is still several thousand meaningful differences. And the variants that matter are not distributed evenly — they cluster in passages of theological interest, which is what would be expected if some were deliberate.
The examples in Interpolations include the ending of Mark, the woman taken in adultery, and 1 John 5:7, the only verse stating the Trinity outright.
"No Christian doctrine depends on a disputed reading"
Broadly defensible, and stated by scholars across the spectrum, including Bruce Metzger and Bart Ehrman.
The claim holds because doctrines are supported by multiple passages, so removing one leaves the others. What it does not show is that the disputed passages are unimportant. 1 John 5:7 is absent from every Greek manuscript before the 14th century and is the only verse stating the Trinity explicitly; the doctrine survives its removal, but the verse was used as proof of it for centuries.
The claim is also compatible with the observation that some changes appear theologically motivated, which is a separate point about how the text was transmitted.
"The variants let scholars reconstruct the original text"
Comparing manuscripts does allow errors to be identified and often corrected, and modern critical editions are the product of that work.
The limit is what the method can reach. Textual criticism reconstructs the earliest recoverable form of the text, which is not necessarily the author's original. For most books, the reconstruction rests on witnesses no earlier than the third or fourth century, and the copying that happened before that cannot be inspected.
D. C. Parker and others have argued that the goal of a single original text may be misconceived for works that circulated and were revised in their earliest period.
What the evidence shows
The New Testament is the best-attested work of antiquity by manuscript count. No autograph survives, the earliest witnesses are small fragments from the second century, and the great majority of manuscripts date from more than a thousand years after composition.
Comparison across manuscripts allows a text to be reconstructed with reasonable confidence for most passages. What it reconstructs is the earliest recoverable wording, not a verified copy of an original, and a number of familiar passages are later additions.
The evidence supports a text with a traceable history of copying and change. It does not support the stronger claim that the wording was preserved unaltered.