Book or Tree? Revelation 22:19 and the Myth of Doctrinal Neutrality

The last warning in Scripture is also the site of a textual dispute. The received text of Revelation 22:19 warns that if a man takes away from the words of the prophecy, God will take away his part “out of the book of life, and out of the holy city.” The modern critical text warns that God will take away his share “from the tree of life and from the holy city.” The difference is one Greek noun: bibliou, book, or xylou, tree.

Wesley Huff introduces the variant after describing the limited materials available to the first printed Greek editors. Erasmus, he says, filled gaps by translating the Latin Bible back into Greek; for that reason, phrases such as “book of life” in Revelation 22 are found in neither the majority nor the oldest Greek manuscripts. The history, together with other examples, is then offered as a reasonable explanation for the differences between the King James Version and modern Bibles. Modern translations may be trusted as accurate representations of what the authors wrote.¹

The historical core of this account is stronger than some received-text defenses have admitted. Erasmus’s only manuscript of Revelation for the first edition—now numbered 2814—lacked the end of the book from late in verse 16 through verse 21. Erasmus supplied Greek from Latin and acknowledged what he had done. The surviving Greek manuscript evidence overwhelmingly reads “tree of life,” while “book of life” has its strongest early support in Latin and some versional or patristic witnesses.

But the conclusion Huff attaches to that history does not follow without a textual and theological argument. Evidence that Erasmus supplied Greek from Latin explains how a reading entered his printed edition. It does not by itself prove that the Greek reading reconstructed by a modern editor is the authentic word God preserved. Nor does it make book and tree doctrinally neutral. One is a register from which a part may be removed; the other is the source of life from which a share may be denied.

The verse does not say the same thing either way. The history may favor one reading. It cannot abolish the difference.

What Erasmus actually did

The history should begin with Erasmus’s own statements rather than with a hostile legend or a defensive counter-legend.

For his 1516 edition, Erasmus borrowed a manuscript of Revelation from Johannes Reuchlin. Its biblical text was interwoven with the commentary of Andreas of Caesarea, and its final extant Greek words stopped near the end of Revelation 22:16. Erasmus did not leave the remaining verses blank. In his annotations and later response to Edward Lee, he said that he had added or supplied the Greek from Latin so that his edition would not end with a lacuna. He also told readers that he had disclosed the procedure so that a person with a Greek copy could correct his wording.²

This admission matters. Some defenders of the Authorized Version have argued that Erasmus must have possessed another Greek manuscript or that the distinctive forms in his final verses prove a Greek exemplar. H. C. Hoskier once explored whether another Apocalypse manuscript might stand behind the text. Later writers repeated the possibility as though it overturned the ordinary history.³

Jan Krans compared those claims with Erasmus’s annotations, his controversy with Lee, the extant manuscript, and the Greek Erasmus printed. He showed that Erasmus’s own report is explicit and that features of the supplied Greek—unusual verb choices, missing articles, and agreement with Latin—corroborate it. Greek manuscripts sometimes cited for the Erasmian form are later copies influenced by the printed text or contain marginal material whose date and source do not establish a pre-Erasmian Greek exemplar.⁴

The responsible historical verdict is therefore firm but bounded. Erasmus supplied the missing Greek conclusion from Latin. “Book of life” entered his printed Greek text in that supplied section. The extant independent Greek tradition known today overwhelmingly supports “tree of life.”

That historical finding settles the immediate editorial procedure, but three cautions remain before it can settle the authentic reading.

First, the Vulgate did not invent the concept “book of life.” The phrase is biblical and appears repeatedly in Revelation. A Latin scribe could easily move from ligno—tree—to libro—book, especially in a context saturated with the words book and words. That is the common critical explanation. But “book” has pre-Erasmian Latin, versional, and patristic history. Erasmus did not originate the reading as an idea; he gave a Latin reading a new Greek form.

Second, an editor’s procedure and a reading’s truth are distinct questions. Back-translation is ordinarily weak evidence for the precise Greek wording of an author. Yet a reading is not necessarily false merely because the editor who printed it reached it by a defective route. Providence could bring an authentic reading into an edition through secondary evidence. To establish that claim, however, the received-text advocate must give a warrant beyond the bare fact that the reading became received.

Third, the modern editor also makes a judgment. “Tree” has imposing Greek support and strong contextual fit, but the printed critical word remains the result of weighing evidence. The manuscripts do not assemble themselves into a main text. The critical conclusion may be well grounded without becoming identical to the artifacts on which it rests.

Huff’s history is thus substantially right at the point where exaggeration is least necessary. Erasmus supplied these verses from Latin. Huff’s larger conclusion—that this history establishes the superior accuracy of the modern reconstruction—requires premises he has not yet supplied.

Book and tree do not issue the same warning

The received reading gives the sanction this form:

God shall take away his part out of the book of life, and out of the holy city, and from the things which are written in this book.

---

1. Wesley Huff, “One Bible, Many Versions”, October 13, 2020, under “The Text of the Modern Translations.” ↩
2. Desiderius Erasmus, Annotationes on Revelation 22:20 and Responsio ad annotationes Eduardi Lei. Erasmus’s statements and their publication history are documented and translated in Jan Krans, “Erasmus and the Text of Revelation 22:19”, Ephemerides Theologicae Lovanienses 86 (2010): 141–59. ↩
3. Herman C. Hoskier, Concerning the Text of the Apocalypse, 2 vols. (London: Bernard Quaritch, 1929), apparatus and discussion at Revelation 22:19. Hoskier’s collations remain valuable; later appeals to his exploratory remarks should be distinguished from a demonstrated Erasmian exemplar. ↩
4. Krans, “Erasmus and the Text of Revelation 22:19,” especially the comparison of minuscule 2814, Erasmus’s supplied Greek, Hoskier’s temporary proposal, the later manuscripts influenced by print, and the Latin witnesses. Krans argues that Erasmus’s Greek corroborates his own account of supplementation. ↩

The critical reading says, in substance, that God will remove his share from the tree of life and from the holy city described in the prophecy.

Both are covenantal sanctions. Both threaten exclusion from eschatological blessing. Both stand beside verse 18’s warning against adding to the prophecy. Both condemn tampering with the revealed words. At that level, the readings converge.

Their objects nevertheless differ. The book of life is the heavenly register of those belonging to life. Revelation speaks of names written in it, of the book being opened, and of those absent from it being cast into the lake of fire. To lose one’s “part” from the book of life is unusual phrasing—elsewhere a name is blotted out—but its field of reference is enrollment and acknowledged participation among the living.

The tree of life is the Edenic and eschatological source of life. Revelation promises access to it, places it within the New Jerusalem, and describes its fruit and leaves. To lose one’s share in the tree is to be barred from the life and blessing it signifies. The connection with the holy city is immediate because the tree stands within the city’s final vision.

The critical reading therefore has an internal advantage: Revelation 22 has just described the tree of life, and verse 14 links obedience with right to the tree and entry into the city. “Tree and city” repeats that paired blessing in verse 19. The received reading has a different canonical resonance: Revelation repeatedly names the book of life, and the warning about removing words from “the book” is answered by removal from the “book of life.” Repetition may explain the reading as scribal assimilation; it may also be deliberate rhetoric.⁵

Internal fit does not make the nouns synonymous. The Martin ledger shows the difference by separating the shared sanction from the distinct object threatened in each reading.

This result is exactly what the phrase “no major doctrine is affected” tends to conceal. The doctrine of final judgment remains. The doctrine of the New Jerusalem remains. Both the book and the tree appear elsewhere. But the warning’s object changes at this verse.

A theologian may still conclude that the two images converge upon the same ultimate exclusion. That is a synthetic conclusion. It does not make the local wording indifferent. Scripture chose a noun. Textual criticism is deciding which noun belongs to the warning against altering Scripture.

The irony should not be exaggerated into an argument—Revelation 22:18–19 does not tell an editor which manuscript reading to choose—but it should sober the discussion. A variant in the warning about variants is not rendered harmless because both forms warn.

The perseverance question requires more than a noun

The received reading has sometimes been accused of threatening the perseverance of the saints. If God can take someone’s part from the book of life, does the verse teach that an elect person may be erased from salvation? Defenders answer that the language is conditional warning, that a professed part is not necessarily an eternal decree, or that the threat describes what belongs to the offender’s appearance rather than God’s immutable election.

The critical reading faces a parallel question. If a person has a share in the tree of life that God can remove, does he possess saving life and then lose it? The shift to tree does not eliminate the theological task. Both readings speak of a “part” or “share” being taken away.

The Martin Rule begins by refusing to place the full doctrine of perseverance into one warning. Westminster Confession 17 teaches that those effectually called and sanctified can neither totally nor finally fall away, grounding the conclusion in the immutability of election, Christ’s merit and intercession, the abiding Spirit and seed of God, and the covenant of grace. Revelation 22:19 does not state all those premises in either form.⁶

Nor does the verse identify the threatened person as one of the elect. It says “if any man.” Scripture addresses visible hearers with warnings that God uses as means of preservation. A threat may describe the covenantal claim, profession, or apparent share of one who proves false. It may state the loss that would follow an act without asserting that the regenerate will finally perform it. The exact relation must be worked out from the whole canon.

The variant nevertheless affects that work. “Book of life” brings the warning into conversation with names, inscription, and blotting. “Tree of life” brings it into conversation with access, right, fruit, and the city. The theologian may reach the same confessional conclusion from either corpus, but he reaches it through different local material.

This is another negative control. The received-text advocate may not say that “tree” destroys perseverance while “book” presents no difficulty. The critical-text advocate may not say that “tree” removes all appearance of loss or that “book” alone introduces theology. Both readings require careful construction. Neither reading alone overturns the confession.

What changes is the scriptural route. A doctrine that survives both routes is not thereby proof that the routes are identical.

History is evidence, not a new rule of faith

Revelation 22:19 gives Huff one of his strongest popular historical examples. Unlike the claim that John 5:4 began in a margin, Erasmus’s supplementation is not merely a possible genetic story. Erasmus tells us that he supplied Greek from Latin. The surviving page and the Greek he printed agree with his account.

That fact should be conceded without evasion. A defense of the received text is weakened when it treats every embarrassing episode in a printed edition as a myth. Confessional Bibliology does not teach that Erasmus was inspired, that his first edition was perfect, or that every step by which a reading entered print was exemplary.

But history does not become self-authenticating because the history is well documented. The conclusion “Erasmus back-translated this section” does not entail “therefore tree is God’s word.” It supplies important evidence against the Erasmian Greek form. To reach the textual conclusion, one must add the Greek witnesses, versional history, patristic use, internal relations, and a rule for weighing them. To reach the confessional conclusion, one must also explain how that result relates to a Greek New Testament confessed as kept pure and authentic.

The received-text position carries a corresponding burden. It cannot say merely that providence caused “book” to be printed and therefore all contrary evidence is irrelevant. It must explain why the church’s reception of this reading is a mark of the authentic text rather than the perpetuation of an editorial defect. It should acknowledge that the surviving Greek documentary case for “tree” is exceptionally strong and then state openly whether its warrant for “book” is versional antiquity, canonical coherence, ecclesiastical reception, confessional identity, or a combination of those.

That is a harder and more honest defense than exposing a mistake in a blog post. Huff’s error lies chiefly in what he makes the history accomplish. He moves from an identified defect in Erasmus’s materials to confidence in the modern text as an accurate recovery of the authorial wording. The inference is plausible at this verse, but it does not establish the general method, and it does not prove that every critical reading elsewhere enjoys the same kind of evidence. John 5:4 is not Revelation 22:19. Mark 16 is not Revelation 22:19. A method that has a strong case once is not thereby vindicated everywhere.

The verdict at Revelation 22:19 must therefore be more discriminating than either popular side usually permits. Historically, Erasmus supplied the missing Greek conclusion from Latin, and “book of life” entered his printed Greek text within that work. The claim is directly documented.

Textually, the extant independent Greek evidence and immediate context strongly favor “tree of life.” Pre-Erasmian Latin and other non-Greek evidence show that “book” is older than Erasmus, but they do not supply the exact Greek main-text ancestry that his edition lacked.

Doctrinally, both readings preserve the warning against corrupting prophecy and the threat of final exclusion. They do not preserve the same image or proposition. Neither reading alone disproves perseverance; each enters that discussion differently.

Confessionally, accepting “tree” means that the received printed text and Authorized Version contain a secondary reading at the Bible’s final textual warning. That consequence cannot be hidden by saying the doctrine remains. Retaining “book” means defending a reading whose documented Greek editorial route is irregular and whose surviving independent Greek support is gravely deficient. That consequence cannot be hidden by saying providence remains.

Book or tree is therefore an unusually useful test. It punishes theatrical certainty on both sides. It requires the received-text defender to face history and the critical-text defender to face authority.

Huff is right that there is a reasonable historical explanation for the variant. He is not right if “reasonable explanation” becomes “doctrinally neutral change” or if one powerful case is made to certify an indefinitely revisable critical text. The manuscript history may identify how a word entered print. It cannot tell the church, without a doctrine of Scripture, when an editorial reconstruction has become the authentic voice of God.

Notes and Further Reading

Series navigation: Previous: “God or He Who?” · Part 8 of 14 · Next: Part 9 — Erasmus, Vaticanus, and the Comma