What Textual Evidence Would Count as Arian?

An accusation against a text must finally point to words.

It is not enough to observe that Arians preferred a reading, that an orthodox proof is shorter, or that a modern edition differs from the received text. Those facts may be important, but each establishes a different proposition. The project must say what the reading does before it says who caused it.

The opposite mistake is equally serious. A critic may answer every doctrinal concern by saying that no major doctrine disappears from the critical text. That response ignores the local loss of inspired testimony. A doctrine can remain provable elsewhere while a particular witness has still been weakened or removed.

The proposition of this article is:

A textual reading is Arian in the strict sense only when its words and context formally entail a defined Arian proposition; readings that merely weaken, accommodate, or contradict such a proposition must be classified separately.

This article supplies the scale. It also establishes what would be required to move from a local reading to a claim about NA28, UBS6, BHS, BHQ, HBCE, or the complete critical corpus.

Begin with the Variation Unit

The smallest ordinary unit of judgment is not “the critical text.” It is a defined variation unit.

A proper record identifies:

• the biblical passage and exact words under review;
• the received Hebrew or Greek reading;
• the reading printed in the named critical edition;
• punctuation or segmentation when it affects syntax;
• relevant manuscript, versional, and patristic evidence;
• the proposition allegedly affected;
• the doctrinal effect;
• the historical use and provenance claim;
• counterevidence in the immediate context and wider corpus.

Without this record, several layers are easily confused.

An apparatus may report a variant without printing it. An editor may prefer a conjecture without placing it in the main text. A translation may adopt an apparatus reading that the Hebrew edition itself does not print. A patristic writer may use wording that does not survive in the manuscripts from which a modern edition is constructed.

The charge must name the object. “BHQ teaches” is false if BHQ merely reports a versional alternative. “The Septuagint corrects the Hebrew” is false if a commentator mentions Greek evidence but retains the Hebrew. “NA29 reads” is premature while the forthcoming Nestle-Aland apparatus remains unpublished, even though UBS6 supplies its intended main text.

Exact edition control is part of doctrinal honesty.

The Martin Rule Applied to Readings

The Martin Rule asks for the exact proposition and all its indispensable elements.

Suppose a reading is accused of teaching AR-02: the Son belongs to the created order. Three things must be shown:

1. the subject is the personal Son;
2. the predicate concerns his being rather than his human nature, office, or manifestation;
3. the reading places him among creatures or gives him creaturely origin.

If the reading says that Christ was made flesh, the second and third elements are absent. If a poetic figure says “I was created,” the first element remains to be proved. If an explicit title is omitted, no positive predicate of creaturehood has yet appeared.

The same discipline applies to each proposition.

A reading may touch one row and leave the others untouched. That resolution must be preserved. The Nicene anathemas provide the historical boundary summarized by these rows.¹

The Five Doctrinal-Effect Classes

Packet Zero fixed five classes.

DU_UNRESOLVED is not a sixth doctrinal class. It is a research state used when the text, syntax, referent, or effect has not yet been closed.

These classes prevent two kinds of erasure.

They prevent the defender of the critical text from saying that a lost witness does not matter because the doctrine survives elsewhere. D1 records the loss.

They also prevent the critic of the critical text from saying that every loss formally teaches the opposite. D3 is reserved for entailment.

Four Preliminary Examples

These are pilot illustrations, not substitutes for the full dossiers assigned to later packets.

A Shorter Incarnation Witness

The received text of 1 Timothy 3:16 reads, “God was manifest in the flesh.” The modern critical text reads with a relative pronoun commonly translated “who was manifested in the flesh.”

The critical reading removes the explicit noun God at that point. That is a material loss of local testimony relevant to AR-04 and belongs at least in the D1 inquiry. But the relative pronoun does not say that Christ is not God. It does not state creaturehood, alien essence, or non-eternity. Calling it D3 Arian would import the conclusion.

The full dossier must still examine antecedent, syntax, manuscript forms, nomina sacra, patristic use, and the complete Pauline corpus. The preliminary distinction is enough: removal of an explicit affirmation is not identical to assertion of its denial.

“Only Begotten Son” and “Only Begotten God”

At John 1:18 the received Greek identifies the revealer as the only-begotten Son. NA28 prints the reading commonly rendered “only-begotten God.”

Whatever judgment is finally made about authenticity, grammar, and provenance, the critical wording explicitly predicates deity. It therefore supplies D4 counterevidence against AR-04, the denial that the Son is truly God. The reading cannot be called formally Arian merely because it differs from the received text or because a heretical writer used it.

That does not end the inquiry. The received reading bears directly upon natural sonship. The critical reading’s unusual construction, its relation to AR-08, its early use, and Burgon’s historical claim all require investigation.² One variation unit may therefore contain D4 evidence on one proposition and a possible weakening on another.

The ledger must allow both facts to remain visible.

John 1:3–4 and the Boundary of Creation

The words around John 1:3–4 can be punctuated so that “what was made” closes the preceding sentence or begins the next clause. Ancient manuscripts ordinarily supply less punctuation than modern editions, and patristic writers divided the words differently. Athanasius and Burgon preserve materially different parts of that reception history.³

The doctrinal question is whether a division disturbs the assertion that every made thing came through the Word or permits something made to be located “in him” as life. The syntactical and reception history are significant.

Yet the prologue still says that all things came to be through the Word and that the Word was God. A punctuation difference cannot be called D3 for AR-06 until the proposed syntax actually denies the Son’s uncreated agency. It may prove D0, D1, or D2 after the clause is parsed. The pilot remains open.

Proverbs 8:22

The received Hebrew says that the LORD “possessed” Wisdom in the beginning of his way. The old Greek tradition uses wording ordinarily translated “created me.” Early Arians employed the Greek wording of the Son.

If three premises were established—Wisdom is directly the personal preincarnate Son, the Greek verb denotes production into creaturely being, and the clause governs his divine person—the resulting proposition would be D3 for AR-02. But those premises cannot be smuggled into the label.

There is also an edition-layer issue. BHS and BHQ print the Leningrad Hebrew consonants and vocalization at this point; they do not replace the Hebrew main text with the Greek verb. A commentator’s preference, an apparatus report, an eclectic reconstruction, and a translation’s adoption must be recorded separately.⁴

The passage is important precisely because the strongest possible doctrinal effect depends upon a transparent chain of argument.

A Reading Can Have More Than One Effect

The five classes are not stickers placed once upon an entire verse.

A reading may be D1 relative to one proposition and D4 relative to another. It may weaken an explicit designation of sonship while strengthening an explicit designation of deity. It may be neutral concerning eternity but significant concerning creative agency.

The database should therefore treat the relation between a reading and a proposition as the classified unit:

reading × proposition → doctrinal effect

This avoids an artificial demand for a single verdict such as “orthodox” or “heretical” when the wording has several distinct consequences.

It also protects counterevidence. If the critical reading at one locus calls Christ God, that fact must appear even in a series investigating anti-orthodox tendencies. If another critical reading removes “Lord” or “Son of God,” that loss must appear even if the critical corpus preserves those titles hundreds of times elsewhere.

Collation first; theological judgment second; intention last.

Textual Effect Is Not Scribal Intent

After effect comes provenance.

Packet Zero fixed six grades:

An Arian use of John 1:18 is not P5. A Valentinian interpretation of John 1:3–4 is not P5. A church father’s accusation is not P5 merely because the accuser is orthodox. A reading’s usefulness to a heresy does not prove that a heretic created it.

The strongest language—“corrupted by Arians” or “corrupted by Gnostics”—is reserved for the strongest record.

This restraint does not make motive irrelevant. It prevents motive from being inferred from outcome and then repeated as history.

From One Reading to an Edition

Even a D3 reading would not by itself prove that an entire edition teaches Arianism.

Three levels must be distinguished.

Local-reading claim

The named reading at the named passage entails AR-02.

This is the narrowest and most readily testable claim.

Pattern claim

Several readings in the edition independently weaken or favor the same Arian proposition, beyond what a neutral distribution would lead one to expect, while contrary evidence is fully reported.

This requires a closed census rather than selected examples.

Corpus claim

The edition, read as a complete text, formally teaches or materially reconstructs an Arian doctrine of the Son.

This is the most demanding claim. It must account for every D4 reading that calls Christ God, Creator, immutable Lord, and object of worship. A corpus may contain local weakenings and remain formally anti-Arian. It may also accumulate losses that materially impair its anti-Arian precision without ever stating the complete heresy.

The final verdict must use the strongest language the evidence warrants, not the strongest language available.

What Would Prove the Indictment?

To prove that a critical reading is Arian in the strict sense, an article must:

1. transcribe the received and critical readings from named editions;
2. establish the grammar and referent;
3. name the exact AR proposition;
4. demonstrate every indispensable element;
5. classify the result D3;
6. disclose readings and context that bear against the conclusion;
7. state provenance separately.

To prove that a critical edition favors Arianism as a pattern, the project must add:

1. an exhaustive or defined census;
2. consistent classification of favorable and unfavorable changes;
3. a comparison with the received control;
4. an explanation of how counterevidence affects the aggregate result.

To prove heretical corruption, it must add:

1. recoverable historical evidence of use, alteration, and transmission at the grade claimed.

Anything less may still prove a serious loss. It will not prove the whole indictment.

Two Opposite Objections

One objection says that this standard is too strict. Heresy often advances by subtraction rather than explicit denial, so D1 should be called Arian.

The premise is possible; the conclusion does not follow. A subtraction may serve an Arian argument and may even have been motivated by one. Those are effect and provenance claims that can be proved. Calling the remaining words a formal denial when they contain no denial weakens the case by making it inaccurate.

Another objection says that D1 is theologically trivial because doctrine is drawn from the whole Bible.

The canon is indeed read as a whole. Yet every inspired witness belongs to that whole. If a reading removes a true assertion that God gave the church, the survival of the doctrine elsewhere does not restore the lost words. D1 is not D3, but neither is it D0.

The scale gives each injury its proper name.

Verdict

Textual evidence counts as Arian in the strict sense when a defined reading, in its actual context, formally asserts one of the eight Arian propositions. A reading that removes an anti-Arian witness is a weakening. A reading that makes the heresy easier is accommodating. A reading that contradicts Arius is counterevidence.

Those distinctions will prevent the series from announcing its conclusion before doing its work. They will also prevent real textual losses from disappearing beneath the slogan that doctrine survives somewhere else.

What This Does Not Prove

This article does not assign final classes to 1 Timothy 3:16, John 1:3–4, John 1:18, or Proverbs 8:22. The examples establish the method and record obvious preliminary boundaries. Their complete manuscript, versional, patristic, edition, and provenance dossiers remain open.

It also does not prejudge the final corpus verdict. The series may conclude that particular readings weaken anti-Arian testimony, that some favor a proposition, that none formally entail Arianism, or that one or more cross the D3 threshold. The ledger, not the desired ending, will decide.

Text and Edition Record

• Received New Testament control: Scrivener’s representation of the Greek underlying the Authorized Version.
• Current critical New Testament control: NA28.
• Newest committee main text: UBS6, identical in main text to the intended NA29; the NA29 apparatus is not yet the published control.
• Received Old Testament control: Bomberg Second Rabbinic Bible (1524–25).
• Modern Hebrew controls: BHS and BHQ as Leningrad-based diplomatic main texts with apparatuses; HBCE separately where an eclectic volume exists.
• Septuagint evidence: admitted only where it is used to displace, reconstruct, or translate against the received Hebrew.
• Preliminary examples: not final apparatus verdicts.

Bibliography

• Athanasius. De Decretis.
• Athanasius. De Synodis.
• Burgon, John William. The Causes of the Corruption of the Traditional Text of the Holy Gospels, chapter 13.
• Council of Nicaea. Creed and anathemas, A.D. 325.
• German Bible Society. Biblia Hebraica Quinta project description.
• German Bible Society. Novum Testamentum Graece, 28th edition.
• German Bible Society. The Greek New Testament, 6th edition.
• Ussher, James. A Body of Divinity.

Revision History

• 2026-07-24: First Packet One draft; D0–D4 and P0–P5 applied to Arian proposition testing, edition and corpus claims separated, and John 1 / Proverbs 8 pilot boundaries recorded without premature closure.

Notes and Sources

Source-control note: D0–D4 classify textual effect, not moral blame or scribal intention. P0–P5 classify the historical record, not doctrinal consequence. A local, pattern, edition, and corpus claim require successively larger evidence sets. The examples remain provisional records pending their assigned dossiers.

Previous in the series: “Subordination Without Arianism?”

Next in the series: “Why ‘Gnosticism’ Must Remain Plural.”