Testing the Irenaean Source Behind Tertullian’s Against Marcion

The strongest evidence connecting Tertullian’s case against Marcion with Irenaeus is a shared set of particular arguments. They do more than quote the same biblical books. They defend some of the same contested readings, combine the same passages, and sometimes offer the same unusual explanation of a difficult sentence.

Table 4 established the shared program: refute Marcion from what his own collection preserves. Table 5 compared the biblical citations and their argumentative uses. The next test is more demanding: do the writers preserve unusual sequences, distinctive arguments, or shared mistakes that point to literary inheritance? I have tested the strongest candidates against the surrounding primary texts, including evidence that works against a simple copying model.

The result is positive for a substantial Irenaean exegetical connection, particularly between Irenaeus and Tertullian. It is less decisive for the more specific proposal that one lost Against Marcion supplied the later accounts. The unusual arguments carry more weight than the sequence comparisons; no exclusive shared error identifying that lost work emerges from the cases examined.

1. What would distinguish an inherited argument from ordinary agreement?

Two writers can quote Luke in Luke’s order because both know Luke. They can agree about a Marcionite reading because both are describing the same textual tradition. Such agreements may be valuable for reconstructing the Gospel, but they do not by themselves show that one account was copied from the other.

A stronger indicator is a shared choice that the biblical text does not dictate: an unusual rearrangement of passages, a particular grammatical repair, or a distinctive objection followed by the same reply. A shared demonstrable mistake can be especially informative, provided that it really is a mistake and is present in both witnesses. These tests must be applied to specific passages; disagreement cannot simply be renamed proof of creative adaptation.

2. The sequence test: the unusual orders do not consistently match

Sequence tested	Earlier or anomalous arrangement	Comparison	Finding
Irenaeus’s catalogue of Lukan material	In AH III.14.3, the woes against the rich (Luke 6:24–26) precede Peter’s great catch (Luke 5:1–11). The catalogue then moves through the bent woman (13:10–17), dropsy and banquet instructions (14:1–14), the midnight request (11:5–8), and the anointing woman (7:36–50).	Tertullian places the catch in AM IV.9.1–2, before the woes in IV.15. He does not reproduce the opening reversal of Irenaeus’s catalogue.	The catalogue contains genuinely unusual order, but this checked feature is not shared. Irenaeus; Tertullian.
A compact bodily-resurrection dossier	AH V.6.2 cites 1 Corinthians 6:15, then 6:13–14: Christ’s members; the body for the Lord; resurrection by God’s power.	AM V.7.4 uses 6:13 → 6:14 → 6:15. The same three verses and bodily argument recur, but in the biblical sequence.	A shared argumentative cluster, without preservation of Irenaeus’s distinctive order. Irenaeus; Tertullian.
A second resurrection dossier	Among the citations in AH V.13.3, the relative order is 1 Corinthians 15:53 → Philippians 3:20–21 → 2 Corinthians 5:4–5.	The corresponding discussions in AM run 1 Corinthians 15 (V.10) → 2 Corinthians 5 (V.12) → Philippians 3 (V.20).	The materials recur, but their relative order changes. This is evidence for comparing a dossier, not for a preserved continuous source sequence.
Epiphanius’s displaced David episode	Gospel scholion 21 in Panarion 42 places David and the showbread (Luke 6:3–4) after the transfiguration and Luke 9:44, before the thanksgiving in Luke 10:21.	Tertullian discusses David within the Sabbath controversy at IV.12.5–8, before the transfiguration in IV.22.	This unusual placement in Epiphanius is not shared by Tertullian. It cannot serve as evidence for their preservation of one distinctive sequence. Epiphanius, scholia 17–22.

The broader Pauline order offers a different result. Tertullian’s treatment runs Galatians, 1–2 Corinthians, Romans, 1–2 Thessalonians, Ephesians/Laodiceans, Colossians, Philippians, and Philemon. Epiphanius’s explicit list at 42.9.4 has the same first eight letter positions, allowing for the Ephesians/Laodiceans identification, but puts Philemon before Philippians. He additionally treats Laodicean material separately.

This is substantial agreement about the shape of the Pauline collection. But the comparison is between Tertullian’s order of treatment and Epiphanius’s reported order; Tertullian’s chapter arrangement is not automatically a transcription of a codex’s contents page. Moreover, we lack a corresponding Irenaean list that could identify this order as his. The agreement supports a shared canonical or polemical tradition without deciding its route of transmission.

Tertullian’s documented reordering of Irenaeus in Against the Valentinians shows that inherited material can be rearranged. That makes adaptation possible here. It does not turn a failed order match into positive evidence of dependence. Riley’s discussion of the demonstrable reuse.

3. The strongest positive case: the same punctuation of 2 Corinthians 4:4

Irenaeus, AH III.7.1–2, resists taking “the god of this age” as a designation of a lower deity. His solution is to pause after “God” and attach “of this age” to the unbelievers. Tertullian, AM V.11.9, proposes the same grammatical operation.

Irenaeus, AH III.7.1	Tertullian, AM V.11.9
in quibus Deus, deinde subdistinguens et modicum diastematis faciens English: “in whom God,” then making a pause and a small interval	Nos contra sic distinguendum dicimus: In quibus deus, dehinc: aevi huius excaecavit mentes infidelium English: We, however, say it should be punctuated thus: “In whom God”; then, “has blinded the minds of the unbelievers of this age.”

This is more informative than a common quotation. Both writers respond to the same theological difficulty with the same particular syntactic repair. A change of punctuation and a transposed construction neutralize the phrase that their opponents use to distinguish gods. The repair may be strained, but its source-critical significance does not depend on calling it an error.

This is the strongest local evidence for an Irenaean exegetical connection in the present comparison. Its limit is equally concrete: the explanation survives in Against Heresies. Tertullian could have taken it from that work, from an Irenaean dossier, or from a wider tradition that transmitted the argument. The match cannot uniquely identify the lost Against Marcion.

Tertullian also adds a simpler alternative at V.11.11: “the god of this age” can mean the devil. He preserves the Irenaean-looking repair without limiting himself to it. That resembles adaptation of an argument more closely than mechanical reproduction of an entire discussion. Irenaeus’s Latin; Tertullian’s Latin.

4. A larger shared argument: “flesh” as conduct rather than bodily substance

In AH V.9–11, Irenaeus interprets “bearing the heavenly image” ethically and denies that “flesh and blood cannot inherit” condemns the body’s substance. At V.10.2 he brings 1 Corinthians 15:50 together with Romans 8:8. Tertullian’s AM V.10.10–12 makes the same combination: heavenly image, present conduct, flesh and blood, and the inability of those “in the flesh” to please God.

Part of the argument	Irenaeus	Tertullian or Epiphanius
The preferred verbal form	portemus et imaginem English: let us also bear the image	Portemus enim, inquit, non portabimus, praeceptive, non promissive English: For he says “let us bear,” not “we shall bear”: as a command, not as a promise.
The bodily substance is retained	non substantiam rejiciens carnis English: not rejecting the substance of flesh	operibus ergo carnis, non substantiae carnis, in nomine carnis denegatur dei regnum English: Therefore God’s kingdom is denied, under the name of flesh, to the works of flesh, not to the substance of flesh.
Epiphanius’s related defence	The Irenaean and Tertullianic distinction explains “flesh” through morally culpable activity.	οὐ γὰρ πᾶσαν σάρκα αἰτιᾶται English: For he is not condemning all flesh. Panarion 42.12.3, elenchus on Galatians, scholion 6.

The connection between Irenaeus and Tertullian involves a preferred reading, an interpretation, and a cross-reference to another Pauline letter. That conjunction is stronger evidence than their isolated use of a standard resurrection verse. Epiphanius has a related ethical defence in his commentary on the Galatian works of the flesh, adding Mary as an example of flesh that cannot justly be condemned.

The limits matter. Epiphanius’s passage does not independently reproduce every component of the Irenaeus–Tertullian cluster. Nor is its cross-reference to 1 Corinthians evidence that Marcion’s copy of Galatians contained a displaced Corinthian verse: it occurs in Epiphanius’s rebuttal. This is a shared mode of exegesis, with different deployments and expansions.

The Irenaean resurrection passages address a broader anti-heretical controversy. Their appearance in Tertullian’s anti-Marcionite case can indicate the transfer of an argument between polemical settings. It does not locate their source specifically in Irenaeus’s projected separate work.

5. Galatians 2:5: a real agreement, but not an exclusive shared mistake

The transmitted Latin of Irenaeus, AH III.13.3, has positive yielding:

Ad horam cessimus subjectioni
English: For a time we yielded in submission.

Tertullian quotes the negative wording and then argues against it in AM V.3.3–5:

Ergo cesserunt, quia fuerunt propter quos cederetur.
English: Therefore they did yield, because there were people on whose account they had to yield.

Both use Paul’s accommodation to defend continuity with the other apostles and with Acts. Tertullian develops that defence with Timothy’s circumcision and the temple episode. This is a meaningful shared textual and argumentative position.

But it fails the exclusivity test for a shared error. The affirmative tradition also has manuscript and other patristic support: the apparatus records omission of the negative expression in the original hand of Codex Claromontanus, some Old Latin witnesses, and authors including Ambrosiaster and Marius Victorinus. Agreement here could reflect a broader textual and exegetical tradition. The apparatus also places Epiphanius among witnesses to the negative form; that is evidence about his wider citation tradition, not an extra reading supplied by his numbered Marcionite scholia. Compare the apparatus for Galatians 2:5.

The agreement is secure for the received Irenaean Latin. Because Irenaeus wrote in Greek, its precise relation to his original wording remains a separate transmission question.

6. Distinctive Gospel readings: agreement does not identify the intermediary

Two Gospel examples demonstrate what a textual agreement can establish—and what it leaves open.

Passage	Tertullian	Epiphanius	Finding
Luke 5:14: testimony “to you”	Ut sit vobis in testimonium English: So that it may be a testimony to you [plural].	ἵνα ᾖ μαρτύριον τοῦτο ὑμῖν English: So that this may be a testimony to you [plural]. Gospel scholion 1.	A distinctive agreement. Both also use the instruction to observe Moses’s prescription against separation from the Creator. It supports a shared reading and argument, but cannot distinguish consultation of related copies from transmission through an earlier dossier.
Luke 24:25: Jesus’s words replace the prophets’ words	omnibus quae locutus est ad vos English: all that he spoke to you	ἐπ᾿ οἷς ἐλάλησα ὑμῖν English: in what I spoke to you Gospel scholion 77.	Both point to Jesus’s speech rather than the prophets’ speech, but Tertullian has “he,” Epiphanius “I.” This is substantial agreement with a visible grammatical difference.

The Adamantius Dialogue also quotes a first-person Emmaus form:

οἷς ἐλάλησα πρὸς ὑμᾶς
English: the things I spoke to you

That makes the form worth investigating across witnesses. However, it occurs in the dialogue’s later anti-docetic discussion; it is not automatically another report of a Marcionite manuscript. More fundamentally, no matching Irenaean quotation of this distinctive wording has been established here. These agreements cannot acquire an Irenaean source merely because the proposed lost work would be a convenient intermediary. Greek text of the Dialogue.

7. Testing the alleged shared errors

The best candidate for a transferred polemical charge is the use of Matthew 5:17 against Marcion. Irenaeus, AH IV.34.1, deploys the saying in an anti-Marcionite argument about fulfilment of the prophets. In AM IV.7.4, Tertullian makes it a deletion charge:

Hoc enim Marcion ut additum erasit.
English: For Marcion erased this as something added.

In the Adamantius Dialogue, Marcus is represented as asserting the reverse form:

οὐκ ἦλθον πληρῶσαι τὸν νόμον ἀλλὰ καταλῦσαι
English: I did not come to fulfil the Law, but to destroy it.

The pattern could reflect the reuse and sharpening of an older anti-law argument. Matthew 5:17 is absent from ordinary Luke, so Tertullian’s charge requires explanation. Yet this is not a demonstrated identical mistake in all three writers. Irenaeus uses a proof text without making Tertullian’s specific deletion claim; the dialogue supplies a contrary formulation. Harmonized Gospel traditions, disputed dominical sayings, and inherited polemic remain possible explanations. We should distinguish the anomalous charge from a proven shared erroneous report. Irenaeus IV.34.1; Tertullian IV.7.4; Dialogue II.

Two other conspicuous textual claims run in opposite directions:

Test case	Tertullian	Epiphanius	Result
Division of Jesus’s garments	AM IV.42.4 explicitly says Marcion removed the division and casting of lots.	Gospel scholion 71 explicitly includes the division of the garments.	A conflicting report about retention, not a shared mistake.
Philemon	AM V.21.1 says the letter’s brevity allowed it to escape Marcion’s falsifying hands.	Panarion 42.12.3, Philemon notice, says it stands in a wholly distorted form.	A direct conflict between general textual assessments. It does not identify which witness, if either, was using a particular copy.
Moses at the transfiguration	AM IV.22.16 contrasts Marcion’s standing Moses with a conversing Moses.	Gospel scholion 17 explicitly has Elijah and Moses talking with Jesus in glory.	The appeal to their presence can support a similar theological argument while the reported wording differs. Shared argument and identical textual report must be assessed separately.

These differences do not rule out literary relationships: copying, abbreviation, independent consultation, and revision can coexist. But they do rule out describing the witnesses as giving one uniform account. Nor can we select one version as a proven error without further evidence about the underlying text. Tertullian IV; Tertullian V; Epiphanius’s corresponding entries.

8. What the test establishes

Proposition	Assessment after the comparison
A shared anti-Marcionite program	Explicitly attested. Irenaeus announces refutation from retained Gospel and apostolic material; Tertullian and Epiphanius repeatedly adopt that strategy.
Particular Irenaean-looking arguments in Tertullian	Substantial positive evidence. The 2 Corinthians 4:4 repair and the combined ethical-resurrection argument are more diagnostic than ordinary agreement in biblical wording.
A distinctive inherited sequence	Not established by these tests. Several actual non-biblical arrangements fail to match; much remaining agreement follows the biblical material or the broad Pauline collection.
An exclusive shared error identifying one lost work	Not established in the examined cases. The strongest anomalies prove either wider textual variation, different uses of a proof text, or conflicting reports.
The specific lost Against Marcion as the common source	A viable hypothesis, strengthened by the positive parallels, but not uniquely identified. Surviving Against Heresies and a wider inherited dossier remain alternatives.
No later writer consulted Marcionite books	Does not follow. Inherited argument and textual consultation can coexist; Epiphanius expressly claims to have taken the books in hand at 42.10.2.

The result strengthens the case for Irenaean mediation of the argument more clearly than it establishes the precise route by which that argument reached later writers. The distinction matters. We can identify a shared exegetical core without pretending to possess the contents of the lost work.

The surviving refutations should therefore be examined as potentially related presentations of an older controversy. Their overlapping arguments cannot simply be counted as separate acts of discovery. At the same time, each explicit textual claim deserves its own assessment. The evidence is strongest when a passage, a distinctive reading, and a particular argument travel together. It becomes weaker when a proposed source is asked to explain every agreement and every disagreement equally well.

Texts and references

• Irenaeus, Adversus haereses: III.7.1–2; III.12.12; III.13.3; III.14.3; IV.34.1; V.6.2; V.9.2–3; V.10.2; V.11.1–2; V.13.3. Ancient Latin version in Harvey’s edition: Book III and Book V. English controls: III.7, III.14, V.10, V.11, V.13. The old English translation’s future form at 1 Corinthians 15:49 must not replace the Latin portemus, “let us bear.”
• Tertullian, Adversus Marcionem: Book IV and Book V, with loci specified above. Order refers to the treatment within these books unless an explicit ancient list is identified.
• Epiphanius, Panarion 42: 9.4; 10.2; Gospel scholia 1, 17, 21, 71, 77; Pauline elenchus 6 and the Philemon notice. Holl’s Greek text in First1KGreek; Frank Williams’s translation for comparison and scholion numbering. Quoted Greek is translated directly here. Numeral slips in the digitized Greek do not override the identifiable episode or the usual scholion numbering.
• Adamantius Dialogue: W. H. van de Sande Bakhuyzen, ed., Der Dialog des Adamantius, GCS 4 (1901), II, printed pp. 88–90, and V, p. 198. Greek text. The work is anonymous; its literary debates must not automatically be treated as transcripts of actual encounters.
• Textual control: STEP Bible, VarApp, Galatians 2:5, for the wider distribution of the affirmative and negative traditions. Codex Claromontanus is the Pauline D witness; it should not be confused with the Gospel manuscript Codex Bezae.
• The documented adaptation model: Mark T. Riley, Q. S. Fl. Tertulliani Adversus Valentinianos: Text, Translation, and Commentary (1971), introduction, especially the source discussion.

Method note: This is a targeted follow-up to the strongest candidates in Tables 4 and 5, with controls drawn from the earlier textual comparison. It is not a statistical census of the complete works. Latin and Greek excerpts have been checked against the retained primary texts; English translations of those excerpts are supplied here. Spelling, accents, ligatures, and line divisions are normalized for readability. Irenaeus’s ancient Latin version is distinguished from his original Greek composition.

Does Tertullian’s Book II derive from Theophilus of Antioch?

 The Theophilus connection is credible; a whole-book attribution is not demonstrated. Several specific interpretations connect Tertullian with Theophilus’s surviving To Autolycus. The comparison also finds substantial Irenaean alternatives. The results support inherited second-century exegetical material, while leaving its route into Tertullian open.

What the historical claim actually is

Eusebius, Church History IV.24.3, independently attests a work of Theophilus against Marcion. That establishes the existence of an eligible lost source. It does not supply its text or establish Tertullian’s use of it. Eusebius IV.25 also names other anti-Marcionite writers, including Irenaeus, Philip, and Modestus.

Loofs’s 1930 Theophilus von Antiochien adversus Marcionem und die anderen theologischen Quellen bei Irenaeus reconstructs an Irenaean source designated IQT and identifies it with Theophilus’s lost work. His discussion of Theophilus’s triad, hands, freedom, and human development is especially relevant here; see particularly pp.46–66. This audit consulted the digitized volume’s OCR, with the Greek checked independently in the surviving To Autolycus.

The more specific distribution of Tertullian’s first three books among Irenaeus, Theophilus, and Justin is associated with Gilles Quispel’s De bronnen van Tertullianus’ Adversus Marcionem (1943), especially pp.34–55 for Book II. Ernest Evans accepts the broad pattern while warning that precise attribution exceeds the evidence. James Haupt’s 2019 dissertation, p.186, reports this genealogy explicitly, linking Quispel to Zahn and Loofs. The fifteen comparisons below are this audit’s findings, not a recovered quotation of Loofs’s or Quispel’s own list. Quispel’s complete chapter was not obtained for direct examination.

What was actually run

The automated comparison used all 29 chapters of Tertullian’s Book II, all 82 chapters of Theophilus’s three books, and all 168 chapters of Irenaeus’s five books. No Irenaean chapter was missing. The English comparison used the nineteenth-century translations reproduced by New Advent. Selected historical arguments were then checked against Tertullian’s Latin, Theophilus’s Greek, and the indicated Irenaean passages.

Corpus	Words in full comparison	After marked quotations and reference spans were removed
Tertullian	18,353	17,109
Theophilus	27,524	21,383
Irenaeus	217,340	179,355

The program used TF–IDF cosine similarity over single words and two-word combinations, English stop-word removal plus a declared list of frequent theological and archaic terms, and sublinear term frequency. This is a retrieval aid for finding passages to inspect. It is not Greek–Latin stylometry, an authorship classifier, or a probability of dependence. The full script and results are embedded below.

In the full-text pass, the highest Theophilus score exceeded the highest Irenaeus score in 6 of the 29 Tertullian chapters. After removing HTML-marked quotations, that happened in 2 chapters. The only chapter in both sets was II.25, the Adam/Cain discussion. These counts are descriptive: Irenaeus is a much larger comparison corpus, translations differ, and chapter lengths vary. They cannot be converted into a percentage of Book II written by either author.

The quotation-removal pass deletes HTML q and blockquote elements and Bible-reference spans. It removes biblical and nonbiblical marked quotations alike; unmarked quotations and allusions remain. It therefore tests sensitivity to quoted material, rather than producing a perfectly Bible-free text. Removal can also shorten a chapter sharply and alter its vector normalization.

Fifteen comparisons and controls

The assessment column separates a substantial parallel from an author-specific source identification. A close Irenaean comparison does not disprove a Theophilan ancestor: that possibility is part of Loofs’s own reconstruction.

No.	Argument or control	Tertullian, Against Marcion	Theophilus, To Autolycus	Irenaeus, Against Heresies	Observed agreement	Assessment
1	Knowledge of God and defective vision	II.2.1–4	I.2; I.5–6	IV.20.5–7	God’s invisibility is answered by attention to the observer and to divine works.	A real apologetic resemblance; the blindness/sun comparison is not sufficiently exclusive to identify a lost book.
2	Creation from nothing	II.5.3	II.4	II.10.2–4	God’s creative power exceeds a human craftsman’s dependence on existing materials.	Common apologetic argument. Tertullian also grants creation from matter for argument’s sake; this is not a shared error or a unique textual reading.
3	Humanity formed by God’s hands	II.4.4	II.18	IV.20.1	The special making of humanity is distinguished from creation by command.	Substantial exegetical parallel. Theophilus names Word and Wisdom; Irenaeus explicitly identifies Son and Spirit; Tertullian foregrounds the Word and a familiar hand. Their formulations are not interchangeable.
4	A better dwelling and progress	II.4.1	II.24	IV.38.3, as a broader developmental parallel	The world is prepared for humanity, followed by a better dwelling and the possibility of advancement.	A worthwhile Theophilus comparison beyond simply quoting Genesis. Irenaeus supplies development toward perfection, though not this exact dwelling formulation.
5	Freedom makes obedience and reward meaningful	II.5.5–6.7	II.27	IV.37.1–2	Humanity has its own power of choice; reward and blame presuppose voluntary action.	Strong shared argument. Irenaeus has both the creation of free agents and the rejection of praise/blame based on an unchosen nature. Direct Theophilus-to-Tertullian borrowing is possible, not established.
6	Humanity brings death upon itself	II.8.1–2	II.27	III.23.1–2, within the argument for Adam’s restoration	The Creator supplies life; disobedience makes humanity responsible for death.	One of the closest original-language parallels. It is better evidence than a generic agreement about goodness, but does not determine whether transmission was direct or mediated.
7	Goodness and justice belong to one God	II.11–13	I.3	III.25.2–3	The same God is kind, judges, and chastens.	Irenaeus is the more specific anti-Marcionite control: he argues that separating goodness from judgment destroys the deity of both proposed gods. Theophilus supplies the compatible theological position.
8	Law, prophets, and an Isaiah quotation sequence	II.17–19, especially II.19.2	III.9; III.11–12	IV.13; IV.16–17	The law benefits humanity; prophets reinforce it. Isaiah 1 is followed by Isaiah 58.	A genuine shared arrangement, partly biblical. The sequences diverge after the two Isaiah blocks. A common collection of proof texts is a live explanation.
9	Adam and Cain questioned to permit confession	II.25.1–3	II.26; II.29	III.23.4–6	God’s questions are explained as opportunities for repentance/confession, rather than ignorance; Cain is compared with Adam.	The strongest specific Theophilus candidate in this audit. Tertullian develops the contrast between confession and denial; Irenaeus also contrasts the responses of Adam and Cain.
10	The visible Word represents the invisible Father	II.27.3–4	II.22	IV.6.7; IV.20.5–7	Old Testament encounters are attributed to the Word acting for the Father.	Substantial shared Logos exegesis. It belongs to a wider second-century tradition and is also present in Irenaeus. The verbal and theological differences require explanation rather than silent harmonization.
11	Breath distinguished from the life-giving Spirit	II.9.1–7	No comparably specific surviving passage established	V.12.2	Tertullian insists on the Greek distinction behind Genesis 2:7 and distinguishes creature from maker.	A strong Irenaean comparison. Irenaeus joins the same breath/Spirit distinction to creature/maker language and 1 Corinthians 15:45. This does not prove that a lost Theophilus lacked the argument.
12	Egypt’s goods as repayment for slavery	II.20	III.9 and III.21 mention the Exodus, but not this full defence	IV.30.1–3	Taking Egyptian goods is defended through unpaid labour and the accuser’s own use of others’ property.	The specific surviving parallel is Irenaeus, not merely the Exodus narrative in Theophilus. Irenaeus himself attributes part of this discussion to a presbyter, showing that the tradition can be older still.
13	The devil’s evil is acquired, not created by God	II.10	II.17 offers the broader goodness-of-creation premise	IV.41.1–2	A creature becomes evil through its own departure from the good.	Theophilus’s passage concerns animals and humanity, not the same detailed argument about the devil. Irenaeus is the closer explicit control; Tertullian’s Ezekiel 28 exposition is additional material.
14	Control: Adam’s infancy and milk before solid food	No equivalent extended infancy/milk exposition in Book II	II.25	IV.38.1–2	Theophilus and Irenaeus explain original humanity through childhood and gradual capacity.	This is important to Loofs’s Theophilus–Irenaeus reconstruction, but must not be counted as a demonstrated Tertullian Book II borrowing. Tertullian’s language of progress does not reproduce the whole argument.
15	Control: Tertullian’s own revision framework	II.1	No counterpart sought in To Autolycus	Not a source-identification test	The opening links Book II to the revision history and argument of Book I.	The surviving work has explicit Tertullianic composition and arrangement. An inherited source can coexist with substantial rewriting; identifying a parallel does not assign every surrounding sentence to that source.

Four original-language anchors

These short excerpts identify the particular language being compared. Translations below are explanatory glosses, not claims that Tertullian literally translated these exact surviving Greek sentences.

Freedom: Autol. II.27 / Marc. II.5.5.
ἐλεύθερον γὰρ καὶ αὐτεξούσιον / liberum et sui arbitrii.
Freedom and self-determination are paired. Irenaeus IV.37.1–2 provides the essential control.

Self-incurred death: Autol. II.27 / Marc. II.8.2.
θάνατον ἑαυτῷ ἐπεσπάσατο / homo sibi mortis statum attraxit.
Humanity brought death upon itself. The compact verbal correspondence is closer than the broad doctrinal theme alone.

Confession: Autol. II.26, II.29 / Marc. II.25.3.
ἀφορμὴν μετανοίας καὶ ἐξομολογήσεως / locum sponte confitendi delictum.
An opportunity for repentance and confession / room to confess the offence voluntarily. The paired Adam–Cain application makes this especially useful.

The Father’s representative: Autol. II.22 / Marc. II.27.3.
ἀναλαμβάνων τὸ πρόσωπον τοῦ πατρὸς / semper egisse in dei patris nomine.
The Word assumes the Father’s person or acts in the Father’s name. This belongs to a broader Logos-theophany tradition, also witnessed by Irenaeus.

The ordering test

Selected themes do not recur in one continuous shared order. In Theophilus the visible Word is explained at II.22, Paradise and advancement at II.24, Adam’s invitation to confession at II.26, freedom and self-incurred death at II.27, and Cain’s corresponding opportunity at II.29. Tertullian treats Paradise and advancement at II.4, freedom and death at II.5–8, combines Adam and Cain at II.25, and returns to the visible Word and invisible Father at II.27.

This distribution is compatible with reuse and rearrangement. It is not a demonstration that the surviving To Autolycus supplies Book II’s running outline. Nor can a different order in a lost Against Marcion be verified from the lost work itself. Ordinary dependence on the Genesis narrative also explains some shared ordering.

The clearest shared quotation sequence is more limited: Theophilus III.12 and Tertullian II.19 both move from Isaiah 1:16–17 to Isaiah 58:6–7. After those blocks Theophilus proceeds through Jeremiah, Hosea, Joel, and Zechariah; Tertullian proceeds into a series of Psalms. This is evidence worth preserving for a possible shared collection of proof texts, not a long continuous sequence unique to these two works.

Which transmission routes remain possible?

Model	What the present evidence supports	What remains unproved
Tertullian directly uses a work of Theophilus	Specific shared Genesis interpretations; an independently attested anti-Marcionite work.	That these passages stood in the lost treatise; that Tertullian read it directly; that it supplied all or most of Book II.
Theophilus reaches Tertullian through Irenaeus	Important parallels occur in all three; this is compatible with Loofs’s proposal that Irenaeus inherited Theophilan material.	The relevant intermediate wording and which Irenaean work Tertullian used.
Tertullian uses Irenaeus and other inherited materials	Specific surviving Irenaean comparisons in II.9 and II.20, alongside Theophilus-related Genesis exegesis.	Whether the additional material came from Theophilus, an anthology, another writer, or oral/exegetical tradition.
All share an earlier exegetical stock	Recurring scriptural problems and solutions, especially when no exclusive wording or peculiar error is shared.	The existence and identity of any particular common written source.

A reconstruction must also avoid circular confirmation: passages assigned to the lost Theophilus on the basis of their occurrence in Irenaeus or Tertullian cannot then be treated as independent surviving Theophilan witnesses to those same authors’ dependence. The extant To Autolycus provides genuinely independent comparison material; a reconstructed lost treatise has a different evidential status.

The name check found no occurrence of the Latin stem for Theophilus in the five downloaded Latin books of Against Marcion, and no Marcion-name stem in the Greek To Autolycus. These are restricted text searches, not proofs of non-use. An anonymous debt is entirely possible, and silence in the surviving apology cannot tell us the contents of its author’s lost polemic.

Conclusion for the larger investigation: the Theophilus hypothesis strengthens the case for treating Tertullian as a transmitter and reviser of inherited argument. It also requires us to qualify an exclusive-Irenaeus explanation of Book II. The data do not decide between direct Theophilan use, Irenaean mediation, and a shared earlier source. They do not establish that Tertullian lacked direct access to Marcion’s writings: inherited argument and direct consultation can coexist, and Books IV–V need their own tests.

All 29 chapters: full text and quotation-removal results

Each cell gives the highest-scoring chapter within the named corpus and its cosine score. A value of 0.10 means a cosine similarity of 0.10 under this particular retrieval model, not a 10 percent likelihood of borrowing. Small score differences should not be treated as historical decisions.

Book II chapter	Theophilus: full	Irenaeus: full	Theophilus: marked quotations removed	Irenaeus: marked quotations removed
2.1	Autol.1.11 0.02259	Irenaeus.2.27 0.02404	Autol.1.11 0.02721	Irenaeus.2.27 0.02545
2.2	Autol.1.4 0.03135	Irenaeus.1.10 0.06740	Autol.1.4 0.03586	Irenaeus.2.13 0.04767
2.3	Autol.1.4 0.03313	Irenaeus.4.38 0.02872	Autol.3.26 0.02245	Irenaeus.4.38 0.03075
2.4	Autol.2.18 0.03780	Irenaeus.4.37 0.03825	Autol.3.14 0.02925	Irenaeus.4.37 0.03928
2.5	Autol.2.9 0.02987	Irenaeus.4.38 0.03751	Autol.2.9 0.03058	Irenaeus.4.38 0.04123
2.6	Autol.2.17 0.03281	Irenaeus.4.13 0.04558	Autol.2.17 0.03606	Irenaeus.4.13 0.04712
2.7	Autol.2.23 0.02165	Irenaeus.4.13 0.03223	Autol.2.23 0.02243	Irenaeus.4.13 0.03340
2.8	Autol.3.11 0.02728	Irenaeus.5.12 0.04023	Autol.2.28 0.02650	Irenaeus.5.12 0.03923
2.9	Autol.2.19 0.03385	Irenaeus.5.7 0.04384	Autol.1.7 0.02881	Irenaeus.5.7 0.04421
2.10	Autol.2.20 0.03542	Irenaeus.3.23 0.04425	Autol.2.25 0.03299	Irenaeus.3.23 0.04536
2.11	Autol.2.21 0.04492	Irenaeus.3.25 0.04066	Autol.1.3 0.02800	Irenaeus.3.25 0.04249
2.12	Autol.2.11 0.06604	Irenaeus.3.25 0.04077	Autol.2.13 0.04032	Irenaeus.3.25 0.04425
2.13	Autol.2.17 0.03162	Irenaeus.3.18 0.03533	Autol.2.17 0.03553	Irenaeus.3.23 0.03825
2.14	Autol.1.7 0.02641	Irenaeus.4.36 0.02963	Autol.1.7 0.02636	Irenaeus.4.36 0.02756
2.15	Autol.3.5 0.02805	Irenaeus.4.36 0.03087	Autol.2.35 0.02545	Irenaeus.3.23 0.02941
2.16	Autol.1.3 0.02277	Irenaeus.4.37 0.03306	Autol.1.2 0.02557	Irenaeus.4.33 0.03442
2.17	Autol.3.9 0.06770	Irenaeus.4.13 0.04960	Autol.2.26 0.02274	Irenaeus.4.33 0.03509
2.18	Autol.2.1 0.02355	Irenaeus.4.27 0.03765	Autol.2.1 0.02441	Irenaeus.4.30 0.02913
2.19	Autol.3.12 0.10026	Irenaeus.4.17 0.08155	Autol.2.35 0.02727	Irenaeus.4.13 0.03320
2.20	Autol.3.21 0.02570	Irenaeus.4.30 0.05281	Autol.3.21 0.02792	Irenaeus.4.30 0.05747
2.21	Autol.2.12 0.02961	Irenaeus.4.8 0.04377	Autol.3.19 0.02699	Irenaeus.4.8 0.04777
2.22	Autol.2.36 0.02803	Irenaeus.4.17 0.07111	Autol.1.12 0.01936	Irenaeus.4.17 0.04101
2.23	Autol.2.17 0.01687	Irenaeus.4.27 0.03460	Autol.2.17 0.01919	Irenaeus.3.25 0.02970
2.24	Autol.2.17 0.02831	Irenaeus.3.12 0.04107	Autol.2.17 0.02606	Irenaeus.3.23 0.03892
2.25	Autol.2.29 0.05558	Irenaeus.3.23 0.04375	Autol.2.29 0.05420	Irenaeus.4.3 0.05166
2.26	Autol.2.34 0.02841	Irenaeus.3.6 0.03323	Autol.2.34 0.03172	Irenaeus.4.27 0.03173
2.27	Autol.2.13 0.02001	Irenaeus.4.6 0.04739	Autol.1.3 0.02055	Irenaeus.4.20 0.04095
2.28	Autol.2.26 0.02169	Irenaeus.3.23 0.03769	Autol.2.26 0.02304	Irenaeus.3.23 0.04061
2.29	Autol.1.3 0.02016	Irenaeus.3.25 0.03318	Autol.1.3 0.02294	Irenaeus.3.25 0.03412

Sources and verification boundaries

1. Eusebius, Church History IV.24–25: independent attestation of the lost anti-Marcionite literature.
2. Friedrich Loofs, Theophilus von Antiochien adversus Marcionem und die anderen theologischen Quellen bei Irenaeus, TU 46.2 (Leipzig: Hinrichs, 1930), especially pp.46–66. Digitized volume; the archive’s downloadable OCR was retrieved and searched. OCR corruption, especially in Greek, prevents treating every OCR spelling as an edition reading.
3. Gilles Quispel, De bronnen van Tertullianus’ Adversus Marcionem (Leiden, 1943), pp.34–55. Bibliography and attribution checked through L. Rochus’s 1945 review, Evans’s introduction, and Haupt. Quispel’s complete text was not examined, so this report does not claim to have replicated his individual derivations.
4. Ernest Evans, Adversus Marcionem (1972), introduction, p.xx, section v: the broad Irenaeus–Theophilus–Justin distribution and its evidential qualification. His statements about direct Marcionite texts in Books IV–V are his historical assessment, not a finding established by this Book II test.
5. James Haupt, Tertullian’s Text of the New Testament outside the Gospels (PhD dissertation, Birmingham, 2019), p.186 and notes 237–239: the later reception of the source hypothesis.
6. F. R. Montgomery Hitchcock, “Loofs’ Theory of Theophilus of Antioch as a Source of Irenaeus,” JTS 38 (1937): 130–139, 255–266. The criticism of exclusive attribution is reported in the annotated Philonic bibliography and the later literature; the present comparison independently retains Irenaeus as a control rather than assuming his relevant passages are all Theophilus.
7. Tertullian, Book II, Latin text with section anchors; Holmes’s English translation for the automated pass. The modern Evans English translation was not used as the comparison corpus.
8. Theophilus, To Autolycus: Book I, Book II, Book III, English translation by Marcus Dods. Greek checked in Open Greek and Latin / Perseus TEI text, based on William Gilson Humphry’s Libri tres ad Autolycum (Cambridge: Parker, 1852). The TEI edition is distributed under CC BY-SA 4.0; short excerpts here have normalized accent encoding. See its source description and licensing metadata.
9. Irenaeus, Against Heresies, complete five-book English corpus, chapters linked individually above. The numerical screen did not compare Tertullian’s Latin against a complete Greek Irenaeus: that complete Greek text does not survive.

Reproducible analysis script

Machine-readable screening results

Prepared from public scholarly resources and retrieved primary texts. The source hypothesis remains a historical inference; the lost treatise has not been recovered.