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.
ἐλεύθερον γὰρ καὶ αὐτεξούσιον / liberum et sui arbitrii.
Freedom and self-determination are paired. Irenaeus IV.37.1–2 provides the essential control.
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.
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.
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
- Eusebius, Church History IV.24–25: independent attestation of the lost anti-Marcionite literature.
- 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.
- 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.
- 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.
- 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.
- 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.
- 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.
- 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.
- 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.
