Quis Dives Salvetur, Carpocratian Exegesis, and the Letter to Theodore

Clement Defending Mark against a Matthew-First Tradition

By Stephan Huller

Clement makes a consequential choice at the beginning of Quis dives salvetur (Who Is the Rich Man That Shall Be Saved?). He quotes the encounter between Jesus and the rich man, identifies his text as Mark, and works through its demands in order. I propose that this choice gives Mark authority within a dispute over possessions, righteousness, and perfection.

Even before that quotation, QDS 2.2 and 3.1 use the characteristic Matthean expression “kingdom of heaven.” This makes the subsequent choice of Mark worth investigating. It does not securely identify the opening saying as Matthew: the usual text of Matthew 19:24 itself reads “kingdom of God.”¹

The case for identifying a Carpocratian element in that dispute has a particularly strong starting point. A central argument in QDS reappears in the Stromata, where Clement expressly directs it against Carpocrates. Epiphanius supplies a reported connection between Carpocrates and Matthew; Irenaeus helps explain the Christological stakes. Together these comparisons give the relationship between QDS and the Letter to Theodore a more specific historical setting.

1. Clement explicitly uses the same argument against Carpocrates

At Stromata III.54.1, Clement introduces his refutation of Ἡ δὲ Καρποκράτους δικαιοσύνη, “the righteousness of Carpocrates.” Jesus’ command to give, he explains, is followed by the command not to turn away a borrower. Giving requires someone who possesses, gives, and lends. Clement then cites feeding the hungry, giving drink to the thirsty, welcoming strangers, and clothing the naked.

QDS 13.1 poses the corresponding question:

τίς γὰρ ἂν κοινωνία καταλείποιτο παρὰ ἀνθρώποις, εἰ μηδεὶς ἔχοι μηδέν;

“What sharing could remain among human beings if nobody possessed anything?”

At QDS 13.4–7, Clement follows this question with the same obligations toward the hungry, thirsty, naked, and homeless. Universal abandonment of the resources needed to help them would make Jesus command both sharing and its opposite. The parallel consists of an argument followed by a recognizable cluster of Gospel examples. One occurrence expressly names Carpocrates as its target.²

The issue is the meaning of κοινωνία, sharing or fellowship. In the passages Clement quotes from Epiphanes’ On Justice, divine righteousness means community with equality, while human laws introduce “mine” and “yours.” QDS 13–14 develops another account of fellowship: possessions can serve the needs of others when their owner masters the passions that corrupt their use.³

The next paragraph of the Stromata makes the connection more specific. At III.55.2 Clement turns to “If you wish to become perfect,” selling possessions, and giving to the poor. His named refutation of Carpocratian righteousness leads directly into the rich-man episode that governs QDS. The interpretations are not identical, as we shall see, but the shared argument places the two discussions in the same field of controversy.⁴

2. The Gospel Clement chooses is Mark

After quoting the rich-man episode, Clement writes at QDS 5.1:

Ταῦτα μὲν ἐν τῷ κατὰ Μᾶρκον εὐαγγελίῳ γέγραπται.

“These things are written in the Gospel according to Mark.”

He acknowledges verbal differences among the other received Gospels while affirming their agreement in meaning. His awareness of alternatives makes his selection worth examining: Mark supplies the passage being explained.⁵

I treat the expressions in Clement’s quotation as authentic Markan material for Clement. My proposal does not imagine him inserting Matthean and Lukan “bits” into a Gospel he otherwise called Mark. The following canonical parallels identify the readings under discussion; they do not establish the direction in which those readings travelled.

Wording in QDS 4.6	Canonical parallel	Function in the narrative
ἐμβλέψας ἠγάπησεν αὐτόν	Mark 10:21	Jesus looks upon the man and loves him
εἰ θέλεις τέλειος εἶναι	Matthew 19:21	The condition concerning perfection
διάδος πτωχοῖς	Luke 18:22	The command to distribute to the poor

The distinction between possessing wealth and trusting in it also has a basis in Clement’s quoted text. At QDS 4.9, Jesus speaks of τοὺς πεποιθότας ἐπὶ χρήμασιν, “those who trust in riches.” This form of Mark 10:24 already directs attention toward attachment.⁶

Clement resumes the perfection condition at QDS 10.1 with τέλειος γενέσθαι, “become perfect,” rather than the τέλειος εἶναι, “be perfect,” of his initial quotation. Stromata III.55.2 also uses γενέσθαι, followed by δὸς πτωχοῖς; IV.28.6 quotes selling and giving without the perfection condition. Those passages do not name an evangelist. QDS supplies the explicit Markan attribution.⁷

Other witnesses preserve related combinations. Irenaeus combines perfection with divide pauperibus. The transmitted Arabic Diatessaron combines perfection, the one thing lacking, and Jesus’ loving look. Cyprian expressly assigns a related sequence to Matthew. These witnesses have different attributions and transmission histories. Clement’s distinctive testimony is his assignment of the extended quotation to Mark.⁸

3. QDS follows the passage through its successive clauses

QDS can be described as the earliest surviving sustained exposition of a passage expressly identified as Mark. It is an extended interpretation of one episode, rather than a commentary on the whole Gospel.⁹ Clement repeatedly returns to successive clauses of the narrative:

QDS	Narrative clause or problem under discussion
6.1–3	The question about inheriting eternal life
6.4–8.2	The address to Jesus as good and the knowledge of the good God
8.2–9.2	The commandments observed from youth and Jesus’ love for the man
10.1–2	If you wish to be perfect and the freedom of the person addressed
10.3–7	The one thing lacking and the comparison with Martha and Mary
11–15	Sell and give interpreted through possessions, their use, and the passions
16–19	Follow me, spiritual poverty, different treasures, and heavenly wealth
20.1–2	The man’s departure and the difficulty of salvation
20.3–6	The disciples’ astonishment and their question about who can be saved
21.1–3	What is impossible for human beings becomes possible through God
21.4–7	Peter’s declaration that the disciples have left everything and followed
22–24	Leaving family and possessions for Jesus and the Gospel
25	Present recompense with persecutions and life in the coming age
26–27.2	The first and last, concluding distinctions, and a return to the camel

This sequence matters because the discussion corresponding to Matthew 19:21 remains inside the exposition of the passage Clement called Mark. Chapter 10 explains the condition concerning perfection; chapters 11–19 unfold the commands that answer it. “Matthew added,” at 17.5, appears within that extended discussion.¹⁰

Not every chapter begins with a fresh quotation. A single command can occupy several chapters, with other sayings brought in to explain it. Modern cross-references identifying Matthew 19:21 in an English translation do not amount to a new source attribution by Clement.¹¹

4. “Poor in spirit” appears before “Matthew added”

The stream of scriptural citations through QDS 15–20 is especially revealing. Clement moves from following Jesus to purity of heart, spiritual poverty, treasure, and entry into the kingdom:

QDS	Scriptural wording or close paraphrase	Canonical comparison
16.1	Come follow me	Mark 10:21 and parallels
16.2	The way for the person pure in heart and divine grace excluded from an impure soul	John 14:6 and Matthew 5:8 with Wisdom 1:4 also echoed
16.3	Poor in spirit and heir of the kingdom of heaven	Matthew 5:3
17.1	Where the human mind is there is its treasure	Matthew 6:21 and Luke 12:34 with order and wording altered
17.2	Good and evil treasure and the mouth speaking from the heart	Closest to Luke 6:45 with Matthew 12:34–35 also relevant
17.5	Matthew named for blessed poverty in spirit and hunger and thirst for God’s righteousness	Matthew 5:3 and 5:6
18.1	The rich who enter the kingdom with difficulty	Mark 10:23–24 and parallels
18.2–4	Destruction of God’s temple and enduring a blow to the face	1 Corinthians 3:17 and Matthew 5:39 or Luke 6:29
19.3	Become pure in heart and see God equated with entering the kingdom	Matthew 5:8
19.4–6	Sell the harmful possessions of the soul and receive heavenly treasure	Interpretive restatement of the command quoted in QDS 4.6
20.1–5	The departure, use of the world, who can be saved, and exceeding astonishment	Mark 10:22–26 with 1 Corinthians 7:31 also echoed

The table records quotations, close paraphrases, and allusions; their canonical parallels do not establish that every supporting saying stood in Clement’s Mark.¹² Two features nevertheless deserve emphasis. “Poor in spirit” and “kingdom of heaven” already appear at 16.3. The explicit attribution to Matthew at 17.5 does not introduce them for the first time. Between those occurrences, the good-treasure saying is closest in several features to Luke 6:45, with Matthew 12:34–35 also relevant.

Clement then writes:

τὸν αὐτὸν δὲ τρόπον καὶ πτωχεία μακαριστὴ μὲν ἡ πνευματική. διὸ καὶ προσέθηκεν ὁ Ματθαῖος· μακάριοι οἱ πτωχοί· πῶς; τῷ πνεύματι.

“In the same way, the poverty that is blessed is spiritual. That is why Matthew also added: ‘Blessed are the poor.’ In what respect? ‘In spirit.’”

The question and answer make “in spirit” carry the explanatory weight. Clement immediately follows it with the blessing of those who hunger and thirst for the righteousness of God.¹³

I favor reading Matthew’s qualifying language as confirmation of the spiritual meaning Clement has already developed under Markan authority. The Greek does not expressly say “added to Mark”: προσέθηκεν can also introduce an additional scriptural witness. The contextual argument is stronger than the grammatical claim. Matthew enters to support an interpretation whose governing narrative is Mark.

5. Epiphanius connects Carpocrates with Matthew

In Panarion 30.14.2, Epiphanius names Cerinthus and Carpocrates and says that they seek to establish Christ’s descent from Joseph and Mary:

ἀπὸ τῆς ἀρχῆς τοῦ κατὰ Ματθαῖον εὐαγγελίου διὰ τῆς γενεαλογίας

“From the beginning of the Gospel according to Matthew, through the genealogy.”

This is an external report connecting Carpocratian interpretation with Matthew. Its specific subject is Christology: the genealogy is made to support natural generation from two human parents. Epiphanius is a later witness, and his attribution does not establish exclusive Carpocratian use of an unaltered Matthew.¹⁴

Clement’s own replies prevent a simple allocation of the Gospels to opposing parties. In the explicitly anti-Carpocratian passage at Stromata III.54, he answers through sayings parallel to Matthew 5:42 and 25:35–40. Matthew-related material can support the disputed interpretation and its correction. QDS 17.5 belongs naturally in that setting: Matthew’s qualification supports the proper understanding of poverty within an exposition governed by Mark.¹⁵

6. Irenaeus and the “bare man”

Irenaeus helps explain what is at stake in the genealogical argument. At Against Heresies III.19.1, he opposes those:

qui nude tantum hominem eum dicunt ex Joseph generatum

“Who say that he was simply and solely a man, begotten by Joseph.”

He describes them as nondum commixti Verbo Dei Patris, “not yet united with the Word of God the Father.” The “bare man” concerns humanity without the divine union necessary for salvation in Irenaeus’s account. In this sentence nude is an adverb; physical undress is not the claim being disputed.¹⁶

His Carpocratian report at I.25 describes Jesus as Joseph’s son, distinguished by his soul and by power received from above. The account of Cerinthus separately describes Christ descending upon Jesus and departing before the passion. At III.11.7, unnamed interpreters who distinguish Jesus from Christ are said to prefer Mark. At III.16.2, Matthew’s precise wording, “the birth of Christ,” is used to exclude such separation. These reports show rival readings crossing Gospel boundaries. Irenaeus does not identify every Mark-preferring interpreter as Carpocratian.¹⁷

QDS establishes the speaker’s divine authority before explaining his command. In chapter 6, Clement calls Jesus the Life, Saviour, Truth, Word, and Perfect, and says that he foresaw the question and answers as God. “Why do you call me good?” introduces instruction concerning the Father and the Son. It does not reduce the Saviour to ordinary humanity. This would answer a reductive Christology, although Clement does not name Carpocrates here.¹⁸

7. A Matthew-based reading of the rich man

Irenaeus supplies an earlier example of the interpretation QDS can be read as challenging. At Against Heresies IV.12.5, his quotation includes neighbour-love and the condition concerning perfection. He questions the man’s obedience and treats distribution as exposing and undoing covetousness. The wording and argument are strongly Matthew-shaped, although he does not name the evangelist.¹⁹

QDS 9.1 takes another approach. Clement expressly says that Jesus does not reprove the man for failing to fulfil the law. Jesus loves him and welcomes his obedience, while declaring him incomplete with respect to eternal life. The question becomes how someone already obedient advances toward perfection.²⁰

Clement also knew and used the other interpretation. At Stromata III.55.2 he says the man had not fulfilled neighbour-love; IV.29.3 also makes neighbour-love important. This prevents turning Clement’s preference in QDS into an exclusive Gospel allegiance throughout his writings. The explicitly anti-Carpocratian setting immediately before III.55.2 makes the difference especially instructive: Clement can employ the same social argument while interpreting the rich man differently.²¹

8. Origen preserves the continuing controversy

Origen’s Commentary on Matthew provides a later comparison. Clement cannot be responding to that surviving work, which Eusebius places in Origen’s later career. Its value is that it preserves the problems and competing readings against which Clement’s choice of Mark would have force.²²

Mark’s loving look

At XV.14, Origen considers whether the neighbour-love clause was added. Mark and Luke lack it; Mark’s statement that Jesus loved the man supports the sincerity of his obedience, much as it does in QDS. Origen also investigates a solution retaining the clause. Even a commentary organized around Matthew can make Mark decisive.²³

Crates and renunciation

QDS 11–12 invokes Crates to show that an outward renunciation performed before Christ cannot exhaust Jesus’ distinctive saving instruction. Origen XV.15 invokes Crates against someone who turns to allegory because literal renunciation seems too difficult. The same example appears in opposed arguments.²⁴

Selling does not automatically remove desire

QDS 12.4–5 argues that someone who abandons property can still covet it and regret its loss. Origen XV.16 raises the corresponding objection: sale alone does not remove anger, fear, desire, or grief. Origen therefore cannot be reduced to a merely literal interpreter.²⁵

Material gifts and spiritual assistance

At XV.17, Origen considers an exchange in which material gifts obtain the prayers and spiritual assistance of the poor. This resembles the reciprocal support advocated in QDS 31–35. Origen presents a proposed interpretation for consideration.²⁶

Giving discarded passions to the poor

Origen XV.18 investigates reading possessions as passions and the poor as powers deficient in goodness, while recognizing the strain this puts on selling and giving. Clement’s distinctions between kinds of wealth address the same difficulty; his Stromata also registers objections to giving discarded passions to the poor.²⁷

Peter’s “everything”

QDS 21.5–7 asks what Peter had abandoned if his property amounted to little, and develops the abandonment of inward possessions. Origen XV.21–22 defends a small material sacrifice through the widow’s gift, while also pursuing further interpretation. The same objection elicits a different first answer.²⁸

Poverty, treasure, and purity of heart

In Against Celsus VI.16, Origen rejects indiscriminate praise of material poverty: poor people can also live wickedly. QDS 11.3 and 16–19 make the corresponding distinction. Origen does not identify the beatitude there as Markan.²⁹

Other comparisons extend the shared vocabulary. Origen’s Commentary on John XIII.20–22 reads hunger and thirst for righteousness through living water; On First Principles I.1.9 connects seeing God with the purified mind; Commentary on Matthew X.14 interprets heavenly treasure spiritually, while XV.10 uses the good-treasure saying to distinguish human goodness from God’s. Compare QDS 17.5, 19.3, 13.3, and 17.1–2.³⁰

These are signs of a continuing exegetical debate. They allow us to compare Clement’s Mark-governed exposition with a later Matthew-centered commentary without claiming direct dependence or making Origen Clement’s historical opponent.³¹

9. An earlier Markan tradition behind Matthew and Luke

My further hypothesis is that Matthew and Luke appropriated material from the earlier Markan tradition represented by Clement’s Gospel. The perfection condition and distribution verb are authentic Markan elements for Clement because they belong to the quotation he expressly calls Mark.

The beatitudes raise a wider possibility. Luke 6:20–21 blesses the poor without “in spirit” and those who hunger “now.” Matthew qualifies poverty and specifies hunger and thirst for righteousness. If Clement’s “Matthew added” reflects comparison with his Markan tradition, Luke could preserve less qualified forms inherited from that earlier material. Matthew would not need to be expanding Luke itself.³²

That wider reconstruction remains inferential: Clement does not expressly quote the unqualified beatitude and label it Mark. Tertullian’s discussion of Marcion supplies another witness to an unqualified blessing of the poor, but no independent attribution of it to Mark.³³

“Alexandrian Mark” here means a Markan tradition associated with Clement’s setting, not the modern category of the Alexandrian text type. Matthew and Luke need not have consulted Clement’s later physical manuscript for that manuscript to preserve earlier material.

10. The Letter to Theodore in the same controversy

The Letter to Theodore describes Mark bringing material to Alexandria and enlarging his earlier work for readers advancing toward deeper understanding. Mark supplies narratives and sayings whose interpretation leads the learner toward knowledge; the Alexandrian church preserves the work and governs access to it.³⁴

The anti-Carpocratian comparison gives this defense of Mark a concrete setting. The letter names Carpocratians and describes their claims of freedom as slavery to desire. Clement’s acknowledged Stromata distinguishes the sharing Jesus requires from the sexual fellowship he attributes to Carpocrates. QDS demands interpretation beyond the fleshly surface and the removal of passions from the soul. These concerns already coexist in Clement’s acknowledged writings.³⁵

The language of instruction also converges. QDS 3.2 uses initiation language; 5.2–4 insists that even apparently explained sayings require deeper inquiry; the exposition then leads from obedience toward perfection. This makes the letter’s account of spiritual interpretation intelligible within Clement’s existing practice.³⁶

The letter’s wealthy young man, who loves Jesus and receives instruction concerning the kingdom, appears after the passion prediction corresponding to Mark 10:32–34. This places him near the rich-man episode expounded in QDS. At QDS 16.1, the soul must become poor, pure, and naked before following Jesus. These correspondences invite interpretation, without establishing that the two wealthy men are the same person or that a particular ritual occurred.³⁷

Irenaeus’s “bare man” adds a Christological question to investigate. It cannot settle the manuscript reading or meaning of the letter’s disputed nakedness expression: the vocabulary and immediate contexts differ.

Epiphanius’s Matthew report and the letter’s corrupted Mark need not be contradictory. Carpocratian readers could use more than one Gospel or interpret overlapping traditions. But neither report identifies canonical Matthew as Theodore’s “mixed Gospel.” The letter names Mark as the copied and corrupted text, never names Matthew as the resulting mixture, and quotes additions absent from canonical Matthew.³⁸

I regard the coherence with QDS and the Stromata as positive evidence favoring Clementine authorship. It does not independently exclude a knowledgeable imitator. Even an authentic letter would still transmit a Gospel-origin account requiring historical assessment. Authorship, Gospel history, and the letter’s surviving manuscript remain related but distinct questions.³⁹

11. What the identification establishes

The strongest evidence is internal: QDS repeats an argument Clement explicitly directs against Carpocrates, followed in the Stromata by the rich man and perfection. Epiphanius supplies the reported Matthew connection; Irenaeus illuminates the Christological dispute. This supports identifying Carpocratian teaching among the positions QDS answers.

QDS never names its opponents, and material renunciation was not peculiar to Carpocratians. An exclusively Carpocratian target for the entire work therefore remains unproved. Clement’s use of Matthew against Carpocrates also shows why the issue must be framed as competing interpretations, rather than fixed Gospel parties.

Gospel chronology presents another question. Eusebius attributes to Clement a tradition that the genealogical Gospels came first. The usual reading places Matthew and Luke before Mark; Stephen Carlson argues that προγεγράφθαι concerns public promulgation. Clement’s local preference for Mark in QDS does not require resolving that dispute, but my broader reconstruction of earlier Markan material must address it.⁴⁰

My proposal is that QDS develops a Mark-governed response to interpretations with a demonstrable Carpocratian counterpart. That finding gives the Letter to Theodore a more substantial Clementine setting: the defense of Mark, progress toward perfection, and the rejection of readings that legitimate the passions already belong together in Clement’s acknowledged work.

Sources and notes

1. Clement, Quis dives salvetur 2.1–4; 3.1–6; 4.4–10; 5.1–4. QDS designates Quis dives salvetur throughout. Greek follows the University of the Aegean Digital Patrologia transcription (2006). “Kingdom of heaven” is characteristic Matthean diction; the usual text of Matthew 19:24 itself nevertheless has “kingdom of God.” The opening wording therefore cannot securely identify Clement’s source. Back to text
2. Clement, Stromata III.6.54.1–4, especially the question at 54.2 and the quotations of Matthew 25:35–40 at 54.3; compare QDS 13.1–7. The Greek question in QDS 13.1 is τίς γὰρ ἂν κοινωνία καταλείποιτο παρὰ ἀνθρώποις, εἰ μηδεὶς ἔχοι μηδέν; The shared argument and scriptural examples establish reuse of an argument Clement expressly directs against Carpocrates elsewhere; they do not establish the exclusive identity of QDS’s unnamed interlocutors. For earlier comparisons of QDS 12 with Clement’s Carpocratian polemic, see Moeser, Jesus the Oracle (2019), 68–69, discussing Denise Kimber Buell and L. William Countryman; those observations are distinct from the more specific sequence compared here. Back to text
3. Clement, Stromata III.2.6.1–8.1, quoting Epiphanes, On Justice. Divine justice is characterized as κοινωνίαν τινὰ ... μετ’ ἰσότητος, while human laws introduce “mine” and “yours.” Distinguish Clement’s account of Epiphanes from Epiphanius of Salamis, the later author of the Panarion. No inference that all reported Carpocratian groups practised identical arrangements is required. Back to text
4. Clement, Stromata III.6.54.1–55.2; QDS 9.1 and 13.1–7. The rich-man interpretation at Stromata III.55.2 immediately follows the explicitly anti-Carpocratian discussion at 54. Clement there questions the man’s fulfilment of neighbour-love; in QDS 9.1 he accepts his fulfilment of the law. The shared social argument survives this exegetical difference. No chronology or change of position can be inferred from the comparison alone. Back to text
5. Clement, QDS 4.6; 5.1. Compare Mark 10:21, Matthew 19:21, and Luke 18:22. The attribution at 5.1 concerns the extended narrative immediately quoted. The claim that its distinctive wording was authentic Markan material for Clement is distinguished here from a reconstruction of its earlier transmission. Back to text
6. Clement, QDS 4.8–9, quoting the form of Mark 10:24 containing τοὺς πεποιθότας ἐπὶ χρήμασιν. This qualification is absent from many modern critical texts of Mark. The argument concerns the Markan text Clement actually presents, not an assumption that all witnesses to Mark have this wording. Back to text
7. Clement, QDS 10.1; Stromata III.6.55.2 and IV.6.28.6–29.3. The perfection condition in QDS 10.1 and Stromata III.55.2 has γενέσθαι; QDS 4.6 has εἶναι. Both Stromata citations use δός, rather than the διάδος of QDS 4.6. Neither Stromata passage expressly names an evangelist. Back to text
8. Irenaeus, Against Heresies IV.12.5 (Latin: Si vis perfectus esse; divide pauperibus); Arabic Diatessaron XXVIII.48–49, trans. Hope W. Hogg; Cyprian, Testimonies to Quirinus III.1. The Latin divide does not establish an underlying Greek διάδος. The Arabic harmony is a later transmitted witness, not Tatian’s surviving second-century Greek text. Cyprian expressly introduces his sequence as Matthew. Back to text
9. Annelies Gisela Moeser, Jesus the Oracle: Reading Mark 9:33–11:19 as an Oracular Procession in Second/Third Century Roman Egypt (PhD diss., Brite Divinity School, 2019), 55. “Earliest surviving sustained exposition of a named Markan passage” preserves the relevant qualification: QDS is not a continuous commentary on the whole Gospel, and earlier lost expositions cannot be excluded. Back to text
10. Clement, QDS 6.1–27.2. Compare Jana Plátová, “Clement of Alexandria’s Homily Quis Dives Salvetur? and Its Pastoral Challenges for Alexandrian Christians,” Vox Patrum 85 (2023): 7–22, appendix at 20. The table follows Clement’s narrative resumptions. In the Greek text used here the departure is discussed at 20.1, the disciples’ alarm at 20.3–6, and the first and last at 26.1; the camel is revisited at 26.7–8. Back to text
11. Clement, QDS 9.1–10.7; 11.1–19.6. The Matthew 19:21 label attached to QDS 10 in the New Advent translation is an editorial cross-reference. Clement does not introduce a new Gospel source there. The command concerning possessions is resumed at 11.2 and receives its extended explanation through chapter 19. Back to text
12. Clement, QDS 15.1–20.6, especially 16.1–3; 17.1–5; 18.1–4; 19.3–6; 20.1–5. The table distinguishes quotation, paraphrase, and allusion. Its canonical parallels do not turn every supporting saying into an explicit citation of Matthew or Luke, or establish that every saying stood in Clement’s Markan exemplar. Back to text
13. Clement, QDS 17.5. The transmitted words explicitly name Matthew and use προσέθηκεν, but do not state “to Mark.” The quotation’s punctuation is editorial. The question πῶς; and answer τῷ πνεύματι focus the interpretation on the qualification. Clement’s next saying includes δικαιοσύνην τοῦ θεοῦ, “righteousness of God,” rather than the usual Matthew 5:6 wording δικαιοσύνην. Back to text
14. Epiphanius, Panarion 30.14.2, in Karl Holl, Epiphanius I, GCS 25 (Leipzig, 1915), 351, lines 9–12. The Greek names Cerinthus and Carpocrates and their appeal to the beginning of Matthew through the genealogy. A gap is marked after εὐαγγελίῳ; δῆθεν also marks the reporter’s distancing. The passage attests Epiphanius’s association, not independently verified exclusive use of an unaltered Matthew by second-century Carpocratians. Back to text
15. Clement, Stromata III.6.54.1–3, quoting the giving-and-lending saying parallel to Matthew 5:42 and the judgment saying parallel to Matthew 25:35–40; QDS 17.5, explicitly naming Matthew. Compare M. David Litwa, Carpocrates, Marcellina, and Epiphanes: Three Early Christian Teachers of Alexandria and Rome (London and New York: Routledge, 2022), chapter 2, discussion of the Carpocratian Gospel citation and Panarion 30.14.2. Litwa questions Epiphanius’s inference about Matthew; it should therefore remain a late external witness rather than the sole basis of the identification. Back to text
16. Irenaeus, Against Heresies III.19.1: qui nude tantum hominem eum dicunt ex Joseph generatum; nondum commixti Verbo Dei Patris. Nude is an adverb in the surviving Latin; “bare man” paraphrases the unqualified-humanity claim, rather than quoting a phrase nudum hominem here. Compare the explicit nudum hominem in Tertullian, De carne Christi 14. The language does not by itself establish physical nakedness or determine the reading or meaning of the disputed nakedness expression in the Letter to Theodore. Back to text
17. Irenaeus, Against Heresies I.25.1–2, on the human Jesus and the power descending upon him in the Carpocratian report; I.26.1, on Cerinthus’s distinction of Jesus and Christ; III.11.7, on unnamed interpreters who distinguish Jesus from Christ and prefer Mark; III.16.2, on Matthew’s “birth of Christ” wording as protection against separation. Irenaeus does not name the Mark-preferring interpreters at III.11.7 as Carpocratians, and these reports should not be collapsed into one identical Christology. Back to text
18. Clement, QDS 6.1–4, especially 6.3, on Jesus foreseeing the question and responses as God; 7.1–8.2, on knowledge of the Father, the greatness of the Saviour, and the law’s preparation for grace. This is compatible with a corrective to a merely-human interpretation of Jesus, but Clement does not name Carpocrates in these sections. Back to text
19. Irenaeus, Against Heresies IV.12.5. His quotation includes the neighbour-love command found in Matthew 19:19 and the condition concerning perfection in 19:21, although the passage does not name Matthew. He questions the man’s claim of obedience and interprets distribution as undoing covetousness. These features establish a Matthew-shaped reading without proving an exclusive Matthean source. Back to text
20. Clement, QDS 8.2–9.2. Compare Mark 10:19–21 and Matthew 19:18–21. Clement’s quoted commandment list does not contain Matthew’s neighbour-love clause, and QDS 9.1 expressly denies that Jesus reproves the man for failing to fulfil the law. The claim here concerns the local argument of QDS. Back to text
21. Clement, Stromata III.6.55.2 and IV.6.29.3. At III.55.2 Clement says that the man had not fulfilled “you shall love your neighbour as yourself.” These passages are a necessary qualification of any claim that Clement consistently opposed the neighbour-love interpretation. No relative chronology of these passages and QDS is assumed here. Back to text
22. Eusebius, Church History 6.36.1–2, places Origen’s twenty-five books on Matthew in his later career, alongside Against Celsus, when he was over sixty. Origen’s surviving commentary is consequently used as a later comparison, not a work to which Clement’s QDS could literally be responding. Back to text
23. Origen, Commentary on Matthew XV.14, PG 13:1289c–1296a. Origen considers, as one possible solution, whether the neighbour-love clause was added; Mark and Luke, and especially Mark’s loving look, support that proposal. He also considers an explanation retaining the clause. This is not an unqualified report of a manuscript in which he had found the clause absent. Back to text
24. Clement, QDS 11.4–12.2; Origen, Commentary on Matthew XV.15, PG 13:1296b–1300a. For the comparative treatment, see Jana Plátová, “Výklad Mk 10,17–31 (Mt 19,16–30 a Lk 18,18–30) u Klementa z Alexandrie a Origena,” Studia theologica 14.4 (2012): 36–61, at 50–51. The different uses of Crates are explicit; direct literary dependence is an inference. Back to text
25. Clement, QDS 12.4–5; Origen, Commentary on Matthew XV.16, PG 13:1300b–1301b. Origen develops an objection rather than claiming that renunciation and moral transformation are incompatible. He then considers possible answers in XV.17–18. Compare Plátová, “Výklad,” 51–52. Back to text
26. Origen, Commentary on Matthew XV.17, PG 13:1301c–1304b, considers material assistance exchanged for the poor’s spiritual assistance, invoking 2 Corinthians 8:13–15. Compare Clement, QDS 31.6–35.1, especially 32.1–4. The Origen passage is a proposed interpretation submitted for judgment, not a simple rejection of almsgiving or endorsement of every form of exchange theology. Back to text
27. Origen, Commentary on Matthew XV.18, PG 13:1304b–1308a; Clement, QDS 17.2–19.6; Stromata IV.6.28.6–29.2. Origen recognizes the strain involved in interpreting the poor as powers deficient in goodness. Clement’s treatment elsewhere also registers the problem of giving discarded passions to the poor. Both writers investigate alternatives; neither can be reduced to a single literal or allegorical position. Back to text
28. Clement, QDS 21.5–7; Origen, Commentary on Matthew XV.21–22. Origen appeals to the widow’s gift when defending the significance of Peter’s small material sacrifice, while also developing further interpretation. The contrast concerns the immediate answer each gives to the objection about Peter’s limited possessions. Back to text
29. Origen, Against Celsus VI.16; compare Clement, QDS 11.3; 16.3–19.6. Origen links the camel saying with blessed poverty and condemned wealth and rejects indiscriminate praise of the materially poor. This passage does not expressly expound the words τῷ πνεύματι or identify the beatitude as Markan. Back to text
30. Origen, Commentary on John XIII.20–22 (also numbered XIII.4.20–22), on Matthew 5:6; On First Principles I.1.9, on Matthew 5:8; Commentary on Matthew X.14, on Matthew 6:19–21, and XV.10, using the good-treasure saying parallel to Matthew 12:35 and Luke 6:45. Compare Clement, QDS 13.3; 16.2; 17.1–5; 19.3–6. These are agreements or differences of application, not evidence that every supporting saying is contested between the authors. Back to text
31. Plátová, “Výklad,” 60–61, identifies shared interpretive problems and differing solutions while declining to infer Origen’s dependence on Clement merely from their agreement. The present proposal builds a hypothesis about Gospel authority on the primary comparisons; it does not attribute that hypothesis to Plátová. Back to text
32. Luke 6:20–21; Matthew 5:3, 6; Clement, QDS 17.5. Luke directly witnesses unqualified poverty and hunger “now.” Clement does not name Luke here. The proposal that Luke preserved material belonging to an earlier Markan tradition is the historical reconstruction advanced in this paper. Back to text
33. Tertullian, Adversus Marcionem IV.14.1 and IV.14.13. His discussion witnesses an unqualified blessing of the poor in his treatment of Marcion’s Gospel; his later repetition uses “kingdom of heaven.” Neither passage explicitly assigns the saying to Mark. No identification of Marcion’s Gospel with Clement’s Gospel is established by this parallel alone. Back to text
34. Letter to Theodore I.15–II.2. Mark brings his and Peter’s notes to Alexandria and transfers material promoting knowledge into his earlier work. The letter attributes enlargement to Mark; it does not describe borrowing by Matthew or Luke. Back to text
35. Letter to Theodore I.2–8 and II.6–9; Clement, QDS 5.2–4; 16.1–3; 18.1; Stromata III.6.54.1–3. The correspondence concerns the opposition between spiritual interpretation and the use of Gospel language to legitimate passions. The letter’s account of textual falsification is additional evidence within its own narrative, not an accusation independently made in QDS. Back to text
36. Clement, QDS 3.2; 5.2–4; 10.1–2; 12.1; 16.1–3; Letter to Theodore I.18–II.2. Compare Alain Le Boulluec, “La lettre sur l’Évangile secret de Marc et le Quis dives salvetur? de Clément d’Alexandrie,” Apocrypha 7 (1996): 27–41. The comparison concerns the interpretation of Gospel wording and the formation of the learner. Back to text
37. Letter to Theodore II.21–III.11; compare Mark 10:17–34 and Clement, QDS 16.1. The first additional narrative is placed after the passion prediction corresponding to Mark 10:32–34, not immediately after the rich man’s departure. Its wealthy young man is not explicitly identified with the man in Mark 10:17–22. Back to text
38. Letter to Theodore I.11–14 distinguishes true and false claims about Mark; II.6–9 alleges corruption of a copy of the secret Gospel. Neither passage names Matthew. The two quoted additions at II.23–III.16 are not passages in canonical Matthew. Epiphanius, Panarion 30.14.2, does not connect his Matthew report with the Letter to Theodore or its Gospel additions. Carpocratian use of more than one Gospel would reconcile the reports without identifying their texts. Back to text
39. For a critical argument against Clementine authorship, Francis Watson, “Beyond Suspicion: On the Authorship of the Mar Saba Letter and the Secret Gospel of Mark,” Journal of Theological Studies 61 (2010): 128–170. For editions and the wider debate, Michael J. Kok, “Secret Gospel of Mark,” e-Clavis Christian Apocrypha, updated April 2024. Literary coherence supports an attribution without independently establishing the date or authorship of the surviving inscription. Back to text
40. Eusebius, Church History 6.14.5–7, from Clement’s Hypotyposeis. The usual chronological reading places the genealogical Gospels first. For public promulgation rather than earlier composition as the force of προγεγράφθαι, see Carlson, “Order of the Gospels,” 122–125. Back to text

Primary texts online: Clement, QDS; Greek of Stromata III.54–55; Stromata III in English; Greek of Panarion 30.14.2; Irenaeus I.25; III.11; III.16; III.19; Greek transcription of the Letter to Theodore. Translations of Greek and Latin excerpts in this post are my own.

Conclusions from 'the Myth of Tertullian's Eyewitness of Marcion's Canon'

The central conclusion of the study is that Tertullian’s descriptions of Marcion’s texts require examination as historical testimony; they cannot automatically be promoted into a record of personal inspection. His quotations give us evidence about readings attributed to Marcion, together with the arguments Tertullian builds around them. Establishing how he obtained a particular reading is a further historical task.

The earlier post examined twenty-two examples against a broader comparison of 904 indexed Gospel and Pauline clusters throughout all five books of Adversus Marcionem. Those clusters include quotations, paraphrases, narrative references, allusions and references to books or collections. Their number is not a count of verbatim transcriptions or proven errors. The important result concerns the character of the evidence: precise textual allegations coexist with abbreviation, variable Latin rendering, mixed material and theological inference.

By eyewitness here I mean someone who personally inspected the relevant Marcionite books. The comparison does not establish that Tertullian never did so. It establishes that personal inspection of a complete collection cannot simply be assumed from the confidence with which he accuses Marcion of corruption.

1. Knowledge of a reading and direct inspection are separate questions

Tertullian sometimes reports a very specific difference. His discussion of Ephesians 3:9 identifies the preposition in as the point at issue. He contrasts a mystery hidden in God with the reading he attributes to the heretic, in which the omission changes the sense to hidden from God (Adversus Marcionem V.18.1–3). This is detailed textual testimony and deserves serious attention.

The detail tells us what difference Tertullian alleges. It does not, by itself, tell us how he learned of it. A manuscript, an excerpt, a report from another reader or an earlier polemic could all communicate a small verbal difference. The precision of the information and the route by which it reached him must be evaluated separately.

A precise allegation is stronger evidence for the particular reading reported than for a particular scene of document inspection. A reconstruction should preserve that distinction instead of allowing the first claim to settle the second.

2. An accusation of erasure can incorporate theological inference

The discussion associated with Matthew 5:17 is especially revealing. Tertullian argues that Jesus’ instruction to the healed leper agrees with the law. He then reasons that Jesus could have prefaced the instruction with the saying that he came to fulfil the law. The argument culminates in this statement (Adversus Marcionem IV.9.15):

Constat ergo dixisse illum, quia et fecit

“It is therefore established that he said it, because he also did it.” Tertullian goes on to charge Marcion with removing the utterance rather than admitting that his own side inserted it. The reasoning explicitly offered proceeds from Christ’s action, as Tertullian interprets it, to a conclusion about the saying and its erasure.

He may have had additional textual information. The paragraph nevertheless does not supply a report of a manuscript comparison. Its expressed argument cannot be replaced with an unstated eyewitness observation. This is a concrete reason to distinguish an alleged deletion from the evidence offered to establish it.

3. A missing phrase in Tertullian is not sufficient evidence for a missing phrase in his source

The Ephesians 2:20 comparison provides a particularly effective control. In Book IV, Tertullian invokes Paul and quotes the foundation of the apostles without adding the prophets (Adversus Marcionem IV.39.6). In Book V, he quotes the apostles clause and then declares (Adversus Marcionem V.17.16):

Abstulit haereticus, Et prophetarum

“The heretic removed, ‘and prophets.’” The shorter wording therefore occurs in Tertullian’s own earlier citation as well as in the form he later condemns. The earlier citation does not explain whether its brevity comes from abbreviation or its source.

This observation does not prove that the earlier citation came from a Marcionite copy. It does not prove that he used a longer copy either. It shows why the omission alone cannot discriminate between those possibilities. Absence from a patristic quotation and absence from a manuscript are different claims. Every reconstruction that infers a shorter text from a shortened citation must address this problem.

4. Verbal variability does not decide whether a manuscript was present

Galatians 4:4 appears in two forms within Book V. At Adversus Marcionem V.4.2, Tertullian writes:

Cum autem evenit impleri tempus, misit deus filium suum

At Adversus Marcionem V.8.7, the temporal construction changes:

At ubi tempus expletum est misit deus filium suum

The fulfilment-of-time construction differs, while the sending clause remains the same. First Corinthians 15:50 likewise appears with non possidebunt, non consequentur and consequi non possunt (Adversus Marcionem V.10.11; Adversus Marcionem V.10.15; Adversus Marcionem V.14.4). These are demonstrably variable Latin formulations.

A writer consulting a Greek manuscript can translate or restate the same sentence in different Latin. A writer using notes or memory can do the same. The variations weaken the assumption of verbatim reproduction; they do not identify the source of the information. Assigning each formulation to a different Greek exemplar, or concluding that no exemplar was consulted, would require additional evidence.

5. The identity of the comparison text must be established locally

The sword/division passage shows why a reconstruction cannot assume that every accusation compares Marcion with a text otherwise identical to canonical Luke. At Adversus Marcionem IV.29.13, Tertullian gives the division saying and then says that sword is what is written, but Marcion corrects it. Yet sword and division also distinguish the received parallels in Matthew 10:34 and Luke 12:51.

The passage makes the comparator a question requiring investigation. Memory, a harmonized comparison or other textual material could be relevant. The accusation itself does not establish which explanation is correct, nor does it establish that Marcion invented the Lukan wording.

A modern verse coordinate identifies a correspondence; it does not independently establish an ancient book’s title, wording or history. The same restraint applies when a saying with a Matthean counterpart is said to have been removed from the Gospel attributed to Marcion. The accusation cannot supply all the missing steps in the history of that Gospel.

6. Direct access, faithful reproduction and access to the complete collection are distinct claims

The expression “Tertullian had Marcion’s canon before him” can conceal several propositions. It may mean that he consulted a Marcionite Gospel, that he consulted a Pauline collection, that both were available throughout composition, that he reproduced them faithfully, or that the copies represented Marcion’s own edition. These propositions do not stand or fall together.

Consultation of one book would not establish consultation of every book discussed. Consultation at one stage would not establish that the book remained open during every later quotation. Even a demonstrably Marcionite copy would leave its date, provenance and relationship to Marcion’s own editorial work to be investigated.

The collection reports remain valuable. At Adversus Marcionem V.21.1, Tertullian contrasts the acceptance of Philemon with the rejection of two letters to Timothy and one to Titus. That testimony concerns the collection he attributes to his opponent. It does not narrate the circumstances in which he acquired the information. Reports about membership in a collection, reports about individual readings and claims of direct inspection need separate evidentiary judgments.

7. A coherent polemical purpose can coexist with flexible citation and interpretation

Across the passages compared, Tertullian’s defence of the Creator, continuity with prophecy and bodily resurrection remains recognizable. His arguments are often more stable than his wording. This helps explain why several verbal forms can serve one purpose, without making every difference a change of doctrine.

The discussion of the god of this age is a useful example. Tertullian first proposes a punctuation that supports one interpretation, then expressly allows the simpler interpretation identifying the figure with the devil (Adversus Marcionem V.11.9–11). His later use of the devil interpretation (Adversus Marcionem V.17.9) consequently recalls an alternative already offered. It need not represent an unnoticed reversal.

This flexibility also limits what we can extract as the wording of a document. An explanatory clause, an opponent’s objection and a hypothetical consequence do different work. The stable purpose of the polemic does not guarantee a stable quotation form, and a change of quotation form does not by itself demonstrate a change of belief. Neither fact establishes fabrication.

8. Reconstruction requires a stated chain of evidence

The responsible outcome is to describe what supports each proposed reading. An explicit contrast, such as the disputed preposition in Ephesians 3:9, has a different status from a clause inferred to be absent because Tertullian does not quote it. A reported inventory has a different status from a conjecture about Marcion’s motive. A Latin synonym has a different status from a specified change of letters.

The chain of reasoning should therefore remain visible: the wording of the received polemic; the identification of its speaker; the separation of quotation from explanation; the reading attributed to the opponent; and the grounds for inferring its source and history. Independent corroboration can strengthen a claim, provided the independence of the witnesses is itself assessed. Agreement between two writers cannot simply be counted twice if their information may share a source.

The study establishes the need for these controls. It does not prove exclusive dependence on an earlier polemic, identify a particular intermediary, or demonstrate that Tertullian never handled a Marcionite manuscript. Those remain historical questions to be investigated with evidence beyond the citation differences themselves.

The historical claim that remains defensible

Tertullian is a witness to readings attributed to Marcion and to the arguments constructed around them. Some reports are specific and potentially very valuable. Others are abbreviated, interpretive, composite or entangled with the transmission of Tertullian’s own work. The comparison gives us reasons to distinguish their evidentiary weight.

What it does not give us is a demonstrated scene in which the complete collection, accurately representing Marcion’s own edition, lies open before Tertullian while he transcribes it. That scene cannot be used to guarantee the quotations whose character is supposed to establish it.

Tertullian’s status as a direct inspector of Marcion’s complete collection must be demonstrated independently of the mere fact that he accuses Marcion of corrupting Scripture. His testimony retains its historical importance. Its source, precision and scope must be earned at the particular passages where the evidence survives.

Primary passages discussed

The examples above are drawn from the comparison developed in The Myth of Tertullian’s Eyewitness of Marcion’s Canon. Latin quotations retain the wording used in that study; the English explanations are newly written. The Latin Library text credits Ernest Evans (1972), transcribed by Roger Pearse (2002).

• Adversus Marcionem IV.9.15 — inference from Christ’s conduct and the accusation of erasure.
• Adversus Marcionem IV.39.6 and Adversus Marcionem V.17.16 — the apostles’ foundation and the missing prophets.
• Adversus Marcionem IV.29.13 — sword and division.
• Adversus Marcionem V.4.2 and Adversus Marcionem V.8.7 — two formulations of Galatians 4:4.
• Adversus Marcionem V.10.11, Adversus Marcionem V.10.15 and Adversus Marcionem V.14.4 — the flesh-and-blood saying.
• Adversus Marcionem V.18.1–3 — the disputed preposition in Ephesians 3:9.
• Adversus Marcionem V.11.9–11 and Adversus Marcionem V.17.9 — explicitly permitted interpretations.
• Adversus Marcionem V.21.1 — reports about the Pauline collection.

The Myth of Tertullian's Eyewitness of Marcion's Canon

An influential picture can be created by a very small step in an argument. Tertullian accuses Marcion of removing words from a Gospel or an apostolic letter. The accusation is then treated as though it were a report of a comparison made with the relevant documents open before him. A polemical assertion acquires the authority of an eyewitness observation.

The question is how much of that picture the surviving evidence actually supplies. What did Tertullian quote? Whose wording was he quoting? How consistently did he reproduce it? How did he distinguish a reading from an interpretation, an omission from an abbreviation, and an observed difference from a conclusion about what Marcion must have done?

The myth examined here is the assumption that Tertullian’s accusations give us unmediated access to a canon he personally inspected, with each quotation functioning as a dependable transcription of the text in front of him. His work provides indispensable evidence. Its evidentiary value has to be established passage by passage. The present comparison leaves open the possibility that he directly consulted Marcionite texts; it challenges the privilege sometimes attached to that possibility before the character of his testimony has been examined.

To test the assumption, I grouped the Gospel and Pauline references throughout all five books of Adversus Marcionem by modern verse coordinates and compared the repeated passages. The resulting picture is of a writer whose theological purpose is often more stable than his quotation form. He abbreviates, changes Latin vocabulary, adjusts syntax, joins passages, adopts an opponent’s voice and offers more than one interpretation. In several places, the argument itself tells us that we are watching an inference or a hypothetical exchange.

Citation variability does not demonstrate that Tertullian never saw a Marcionite manuscript. It demonstrates that an accusation of mutilation and a quotation in a polemic cannot automatically be treated as a record of direct inspection. That distinction is the basis of this post.

What the study actually counted

The source concordances contained 904 indexed clusters: 489 Gospel entries and 415 Pauline entries, covering all 146 chapters of the five-book work. “Clusters” matters. Some entries are direct quotations or close paraphrases; others are narrative references, possible allusions, shared Old Testament wording, or references to a book or collection. They are not 904 independent transcriptions of Marcion’s text.

Measure	Result	Meaning
Indexed source clusters	904	489 Gospel and 415 Pauline entries.
Repeated primary-reference coordinates	443	278 Gospel and 165 Pauline candidates, after consolidating overlapping locations.
Coordinates repeated across books	368	A subset of the 443; repetitions within one book were also examined.
Narrow quotation-label candidates	72	Each received a contextual note; the label alone does not prove comparable wording of the same clause.
Contextual verse notes	77	The 72 candidates plus five additional contrasts or source cautions.
Retained Latin subsections	1,188	The full source text, with its printed defects and editorial interventions left visible.

Expanding “Luke 1:26–35” into ten searchable coordinates does not create ten direct quotations. The same restraint applies to a whole-chapter reference in Paul. Editorial Gospel parallels were kept distinct from the closest indexed reference. Three records that merely illustrated the title “Son of Man” with possible verse locations were retained as title references rather than turned into artificial quotations.

The 443 repeated coordinates therefore constitute a field for comparison, not a count of contradictions. The most consequential findings are developed below, with the clearest wording differences in bold. Modern chapter and verse numbers serve as finding aids throughout; they do not independently identify the ancient book from which Tertullian obtained a saying.

Begin with the way Tertullian presents his case

Tertullian’s own framing deserves priority over an imagined scene of document inspection. At Adversus Marcionem IV.4.1, he presents competing claims:

Adversus Marcionem IV.4.1

Ego meum dico verum, Marcion suum. Ego Marcionis affirmo adulteratum, Marcion meum.

“I say that mine is true; Marcion, that his is. I affirm that Marcion’s is adulterated; Marcion, that mine is.” Tertullian then invokes priority in time as the criterion for deciding between them. His charge of corruption belongs to an argument about which text has authority and priority. The historical direction of alteration is a proposition being defended. It should not silently become an independent eyewitness premise.

At Adversus Marcionem IV.2.3, he reports that Marcion assigns no author to his Gospel: nullum adscribit auctorem. Whatever we conclude about his identification of that Gospel with a form of Luke, the title is itself part of the dispute. The surviving polemic should not be supplemented with a conveniently explicit Marcionite title page that Tertullian does not report.

At the beginning of Book V, Tertullian explicitly announces a role he is performing:

Adversus Marcionem V.1.4

Adhuc ego in persona discipuli et inquisitoris conversor

“I am still acting in the character of a disciple and an inquirer.” This is an important warning about voice. A statement introduced by “he says,” or an exchange between “I” and “you,” may belong to Tertullian’s staged interrogation. The reader must identify its argumentative function before extracting it as a transcript of someone else’s words.

He also begins the entire work by explaining revision, the loss and unauthorized circulation of an earlier form, and the possibility that differences among circulating copies could confuse readers (Adversus Marcionem I.1.1–2). That history does not explain each surviving citation difference. It does establish that the received five-book polemic stands at the end of a process of composition and transmission, rather than giving us an untouched notebook of observations.

The passages that make the eyewitness assumption difficult

1. Matthew 5:17: an inference about what Jesus said becomes an accusation of erasure

The most revealing example concerns the saying, “I did not come to destroy the law but to fulfil it.” Tertullian discusses Jesus’ instruction to the healed leper and argues that the action accords with the law. He then writes:

Adversus Marcionem IV.9.15

Quocunque modo praecepit, eodem potuit etiam illam praemisisse sententiam, Non veni legem dissolvere sed adimplere.

In effect: however he gave the instruction, he could also have prefaced it with that saying. The reasoning then becomes more forceful:

Adversus Marcionem IV.9.15

Constat ergo dixisse illum, quia et fecit, et te potius vocem domini de evangelio eradicasse quam nostros iniecisse.

“It is therefore established that he said it, because he also did it, and that you removed the Lord’s utterance from the Gospel rather than our people inserted it.”

The expressed argument moves from agreement between action and saying to the conclusion that the saying was spoken and that Marcion removed it. Tertullian may have had additional textual evidence. This paragraph’s actual reasoning is nevertheless an inference from the narrative and from his interpretation of Christ’s conduct. We cannot replace that reasoning with an unstated report that he watched a particular line disappear from a particular manuscript.

The saying’s closest canonical counterpart is Matthew 5:17. Its absence from canonical Luke also makes the identification of the comparison text a real question. The erasure charge does not, by itself, establish that the saying once belonged to a text otherwise identical with our Luke.

2. Ephesians 2:20: The apostles’ foundation: omission is not a transparent source fingerprint

Adversus Marcionem IV.39.6

super fundamentum apostolorum

Tertullian’s appeal secundum Paulum

Adversus Marcionem V.17.16

superaedificati super fundamentum apostolorum. Abstulit haereticus, Et prophetarum

Quotation followed by an explicit allegation of deletion

Book IV lacks and prophets; Book V explicitly says that the heretic removed those words. The earlier short citation does not explain why it is short. It could be Tertullian’s abbreviation and cannot safely identify his source as Marcionite. This is one of the clearest demonstrations that absence in a patristic quotation and an alleged absence from a Gospel/letter copy are different kinds of evidence.

This is a direct test of the method used to reconstruct a shorter text from a quotation. If the Book IV citation were extracted in isolation, its missing prophets could look like evidence for the very shorter form condemned in Book V. Comparison with the author’s own practice exposes the ambiguity. Tertullian’s explicit deletion charge remains evidence for his report about Marcion; the earlier omission remains unexplained by that report. An omission in Tertullian’s sentence and an omission from a manuscript are different observations.

3. Matthew 10:34 / Luke 12:51: Sword and division: the comparator resembles Matthew

Adversus Marcionem III.14.5

machaeram venit mittere in terram, non pacem

Saying applied to the opponent’s Christ

Adversus Marcionem IV.29.13

non, dico vobis, sed separationem. Machaeram quidem scriptum est. Sed Marcion emendat;

Gospel wording followed by an accusation of correction

The received Matthew/Luke parallel itself has sword/division. Tertullian nevertheless invokes sword while alleging that Marcion corrects it. His polemic may use a Matthean-looking comparison rather than a mechanically isolated comparison with the wording of canonical Luke. This is important for evaluating his charges, but does not prove the exact origin of either reading in his materials.

The practical consequence is serious. A reconstruction that assumes every accusation compares Marcion with an otherwise identical canonical Luke will misdescribe what is visible here. The comparator has the sword familiar from Matthew. Several explanations remain possible, including recollection, a harmonized comparison or other textual material. The passage itself does not choose among them. It certainly supplies no warrant for converting an accusation into proof that Marcion invented the Lukan word division.

4. Galatians 4:4: The fullness of time: two forms in Book V

Adversus Marcionem V.4.2

Cum autem evenit impleri tempus, misit deus filium suum

Pauline quotation in the Galatians discussion

Adversus Marcionem V.8.7

At ubi tempus expletum est misit deus filium suum

Quotation introduced by dicente apostolo

Cum autem evenit impleri and At ubi ... expletum est are different temporal constructions. The sending clause is the same. Both arguments identify the sender with the God who ordered and predicted the times. These passages establish verbal variability within a single book, without establishing two Greek copies. The shared omission of other portions of the canonical verse in these short quotations is not, by itself, evidence that those portions were absent from a source.

These two forms appear within Book V, so the difference cannot simply be assigned to a long interval between separate treatises. A single verbatim Latin transcription would not have both constructions at once. Independent translation, quotation from memory and rhetorical restatement are possible explanations. The omission of other parts of the canonical verse in both short quotations is also insufficient, on its own, to prove that those parts were absent from Marcion’s copy.

5. 1 Corinthians 15:50: Flesh and blood: three renderings and two applications

Adversus Marcionem V.10.11

caro et sanguis regnum dei non possidebunt

Quotation introduced as what Paul adds

Adversus Marcionem V.10.15

Caro et sanguis regnum dei non consequentur

Quotation attributed to the apostle

Adversus Marcionem V.14.4

Caro et sanguis regnum dei consequi non possunt

Pauline wording embedded in explanation

Possess/inherit, attain, and be unable to attain are not one fixed Latin sentence. The third also changes future exclusion into inability. The interpretation shifts: V.10.11 and V.14.4 explain flesh and blood through their works, whereas V.10.14–15 stresses the transformation required of the resurrected body. Tertullian distinguishes resurrection from entrance into the kingdom and maintains bodily resurrection throughout. The shift is real interpretive flexibility, not a flat assertion and denial of bodily resurrection.

This is particularly damaging to a method that treats each Latin verb as a transparent window onto the Greek source. The exact Latin wording and the argumentative application both change. The shared commitment is that God will raise the body and fit it for the kingdom. A reader who identifies every local shift with a new manuscript or a new doctrine will multiply differences the argument does not require.

6. Galatians 4:24–26 / Ephesians 1:21: The two covenants: a composite citation in an alteration dispute

Adversus Marcionem III.24.3

apostolus matrem nostram sursum designat

Tertullian’s account of the heavenly mother-city

Adversus Marcionem V.4.8

aliud super omnem principatum generans, vim, dominationem, et omne nomen quod nominatur, non tantum in hoc aevo sed et in futuro, quae est mater nostra

Composite covenant passage in a discussion of what Marcion altered

The later sequence includes recognizably Ephesian principalities/name/age language within the Galatian allegory, followed by mother language and a church-confession phrase. It cannot be presented as an uncomplicated quotation of canonical Galatians. Tertullian introduces it with a claim that Marcion partly changed the passage, but precise quotation/comment boundaries require care; not every connective or added phrase is thereby proved to be Marcion’s. The comparison establishes mixed textual material in the argument, not a second creed of Tertullian.

Here there is recognizably mixed textual material, and it occurs in a setting where Tertullian discusses alteration. That makes the passage important evidence. It also makes boundaries essential: where does the attributed text stop, where does explanation begin, and what belongs to the received transmission? The composite must be preserved for investigation. It should not be regularized into canonical Galatians and then cited as though the irregularity had never existed.

7. Galatians 2:5: The negative in Galatians: keep the textual crux visible

Adversus Marcionem V.3.3

nec ad horam cessimus subiectioni.

Quoted text immediately followed by an announced corruption argument

Adversus Marcionem V.3.4

Ergo cesserunt

Tertullian’s inference

Adversus Marcionem V.3.5

Necessario igitur cessit ad tempus

Tertullian’s historical explanation

The printed sequence moves from we did not yield even for an hour to they yielded and he yielded temporarily. That tension is genuine. However, the first wording appears inside an explicit discussion of vitiatio scripturae. Quotation voice, the disputed negative and the transmission of Tertullian’s text must be resolved before treating this as two inconsistent personal citations. The negative has been retained exactly as the source prints it. The CSEL 47 derivative checked here also retains nec; it does not remove the difficulty.

An eyewitness reconstruction should not solve this difficulty by silently removing nec, nor should it turn the initial negative and the ensuing affirmative argument into two uncomplicated personal readings of Tertullian. His charge of corruption is part of the context. The honest result is an unresolved textual and quotation-voice problem with a clearly stated location, not a confident transcription manufactured by choosing whichever half suits the reconstruction.

8. Luke 16:17; Luke 21:32–33; Matthew 5:18 parallel: The law, the word, and the passing of heaven and earth

Adversus Marcionem IV.33.9

quam unus apex verborum domini

Gospel-related exposition at the law/John boundary

Adversus Marcionem IV.39.18

non transiturum caelum ac terram, nisi omnia peragantur.

Indirect summary in the eschatological exposition

The first passage speaks of a tittle of the Lord’s words where canonical Luke 16:17 speaks of the law. The second attaches until everything is accomplished to heaven and earth, combining neighbouring apocalyptic ideas rather than reproducing Luke 21:33 alone. These are consequential examples of exposition and Gospel language flowing together. They are not, without further evidence, two identifiable copies or a reconstructed sequence of Marcionite verses.

The category needed here is citation intertwined with exposition. Tertullian moves between related sayings and prophetic confirmation. The result cannot be treated as a neutral, continuously copied section of an otherwise recoverable Gospel without first showing which words are quotation and which are his interpretive bridge.

9. Matthew 5:45: Rain and sun: the paired adjectives change

Adversus Marcionem II.17.1

pluentem super bonos et malos et solem suum oriri facientem super iustos et iniustos

Tertullian’s own appeal to Christ

Adversus Marcionem IV.36.3

pluit super iustos et iniustos, et solem suum oriri facit super bonos et malos

Tertullian’s supplementary saying

Rain falls on the good and bad in the first passage, on the just and unjust in the second; the sun receives the other pair. This is a complete interchange in the actual clauses, not a mismatch between index snippets. The shared argument is that the Creator benefits even those who reject him. The separate allegation of deletion in IV.17.6 concerns Marcion’s text and must not be counted as another form adopted by Tertullian.

This is a useful control because there is no elaborate new theological position to explain. The interchange makes little difference to the Creator’s universal beneficence, but it makes a clear difference to the wording. A writer can preserve the point of a saying while failing to reproduce the same allocation of its clauses.

10. Luke 10:22; Matthew 11:27 parallel: Knowing the Father: cognovit and scit qui sit

Adversus Marcionem II.27.4

Nemo cognovit patrem nisi filius.

Explicit appeal to the common Gospel

Adversus Marcionem IV.25.10

Nemo scit qui sit pater, nisi filius, et qui sit filius, nisi pater, et cuicunque filius revelaverit.

Gospel wording in the polemical discussion

The perfect cognovit and direct object patrem give way to present scit and the indirect question qui sit pater; the second passage also supplies the reciprocal clause. This is real wording variation. It is not sufficient to establish two physical Gospel copies. An additional Matthean-looking repetition bracketed in some editions of IV.25.10 is absent from the Latin Library text used here; Roth, The Text of Marcion’s Gospel, p. 133, discusses it as a likely later interpolation. It has therefore not been counted as an additional authorial quotation.

The comparison shows verbal change, but the manuscript problem also warns against attributing every difference in a printed edition directly to Tertullian. Before the Father–Son saying is used as a fingerprint of an ancient Gospel, the received text of the polemic must itself be examined.

11. Luke 4:32: Teaching with authority: a different Latin formulation

Adversus Marcionem IV.7.7

in potestate erat sermo eius

Quotation introduced with inquit

Adversus Marcionem IV.13.1

erat enim docens tanquam virtutem habens

Tertullian’s Gospel recollection

The first places the authority in his word; the second describes him teaching as one possessing power. The second also resembles the Synoptic teaching-with-authority formula. The change is visible within Book IV. It establishes variable citation wording, while the claim of authoritative teaching remains compatible in both contexts.

The same book can thus alternate between two recognizable ways of reporting the authority of Jesus’ teaching. The conceptual agreement is stronger than the verbal agreement. An argument for a particular Greek form needs more than the assumption that the Latin has been copied with mechanical consistency.

12. Luke 10:27; Deuteronomy 6:5 and Leviticus 19:18: The love command: vocabulary and order change

Adversus Marcionem IV.25.14

Diliges dominum deum tuum ex toto corde tuo et ex tota anima tua et totis viribus tuis

Gospel command in Tertullian’s exposition

Adversus Marcionem V.8.10

Diliges dominum de totis praecordiis et totis viribus et tota anima tua et proximum tuum tanquam te ipsum.

Command expressly said to be approved by Christ in V.8.9

Corde becomes praecordiis, ex becomes de, and soul and strength change order. The second adds the neighbour command. Both invoke the same ethical requirement. Because the wording also belongs to the Torah, the correspondence does not uniquely identify a Gospel source.

Shared biblical commands are especially liable to reappear through several channels. A Gospel, Paul’s argument about love, and the underlying scriptural command can all enter the discussion. The variation establishes a flexible rendering; it does not assign each form uniquely to a Marcionite or non-Marcionite exemplar.

13. Luke 11:27: The woman’s blessing: direct quotation becomes indirect narration

Adversus Marcionem III.11.3

Beatus venter qui te portavit, et ubera quae hausisti.

Direct speech attributed to the woman

Adversus Marcionem IV.26.13

beatum uterum qui illum portasset, et ubera quae illum educassent.

Indirect report by Tertullian

Venter/uterus and suckling/nurturing are different formulations. The change from direct to indirect speech explains part of the grammar. Both passages support the reality of Jesus’ mother. This is a clear example of citation freedom, with weaker force as evidence for a different underlying written text.

This example sets a limit on what we should call a variant. The direct and indirect reports genuinely differ, but the change of grammatical presentation matters. A catalogue that ignores that distinction will mistake ordinary narration for a collation of competing manuscripts.

14. Romans 11:33: Wisdom: a Greek-derived term and its Latin equivalent

Adversus Marcionem II.2.4

O profundum divitiarum et sophiae dei

Explicit quotation of the apostle

Adversus Marcionem V.14.9

O profundum divitiarum et sapientiae dei

Explicit quotation of the apostle

Sophiae and sapientiae provide a particularly simple control: the Latin wording varies without changing the thought or requiring a different Greek reading. Separately, the webpage’s in investigabiles/investigabiles spelling and spacing must be preserved and checked against editions before being used as evidence for a meaningful textual variant.

The contrast is almost an experiment in miniature. One can see two Latin forms while retaining the same underlying concept. A Greek-derived noun and its ordinary Latin equivalent do not require two Greek readings. This is precisely the kind of control that must precede a reconstruction from Latin quotations.

15. 1 Corinthians 1:21: Knowing God: several verbs and two divine nouns

Adversus Marcionem II.2.5

non agnoscentes in sapientia dei per sapientiam deum

Tertullian’s syntactically adapted Pauline wording

Adversus Marcionem V.5.7

Quoniam in dei sapientia non intellexit mundus per sapientiam dominum

Quotation in the letter’s exposition

Adversus Marcionem V.5.7

in sapientia sua non erat agnitus

Tertullian’s ensuing explanation

The exposition changes between agnoscere, intellegere and, elsewhere in this same paragraph, cognoscere; deum and dominum also alternate. This shows why a Latin synonym or divine title cannot automatically be retroverted into a different Greek reading. The object of recognition remains the Creator in the argument.

Because multiple ways of saying know occur in the same exposition, no single Latin verb should automatically be assigned to the document supposedly lying before Tertullian. The alternation between God and Lord also requires contextual judgment before it is made evidence for an otherwise unattested source reading.

16. 1 Corinthians 1:25: God’s weakness: invalidum/validius and infirmum/fortius

Adversus Marcionem II.2.5

stultum dei sapientius hominibus, et invalidum dei validius hominibus.

Pauline formulation used by Tertullian

Adversus Marcionem V.5.9

Quid infirmum dei fortius homine, nisi nativitas et caro dei?

Pauline wording recast as a rhetorical question

The second pair of adjectives changes, as does singular/plural in the human comparison. This is a rhetorical recasting rather than a second neutral transcription. Both defend the apparently weak works of God; the later passage applies the language to incarnation and flesh.

The interrogative form makes the rhetorical adaptation plain. The later wording is doing work in an argument about incarnation. Its difference from the earlier formulation should be recorded, but it cannot be divorced from that argumentative use and treated as a disinterested transcription.

17. 1 Corinthians 15:53: Putting on incorruption: different necessity formulas

Adversus Marcionem V.10.14

Oportet enim corruptivum hoc

Paul’s requirement, interrupted by Tertullian’s explanatory aside

Adversus Marcionem V.12.3

Necesse est corruptivum istud induere incorruptelam et mortale istud immortalitatem.

Pauline quotation in the later argument

Oportet and Necesse est express the same necessity; hoc and istud also alternate. The longer context of V.10.14 contains both incorruption and immortality. This is variation in Latin rendering, with stable propositional content.

This is an example of real but modest verbal variation. The proposition remains the same. An apparatus built from patristic quotations needs room for this category; otherwise it will turn Latin alternatives for necessity into fictitious differences between ancient copies.

18. Galatians 4:10: Days, months, seasons and years: the order changes

Adversus Marcionem I.20.4

observantes tempora et dies et menses et annos

Indirect account of the disputed observance

Adversus Marcionem V.4.6

Dies observatis et menses et tempora et annos

Direct Pauline wording

Seasons stands first in Book I and third in Book V. The participle also becomes the second-person verb expected in direct speech. The same legal-observance argument continues across the books. The following sabbaths, meals, fasts and great days in V.4.6 are an expansion introduced with ut opinor; they must not automatically be inserted into Paul’s text.

The extra list of observances following the verse is even more instructive. Tertullian signals his own elaboration. A reconstruction that simply annexes everything in the paragraph to the Pauline quotation would create a larger text by ignoring the author’s indication of inference.

19. 1 Thessalonians 4:17: Meeting the Lord: auferemur and rapiemur

Adversus Marcionem III.24.11

Auferemur enim in nubes obviam domino

Quotation expressly marked secundum apostolum

Adversus Marcionem V.20.7

simul rapiemur in nubibus obviam domino

Quotation with Cum illis and an intervening dicit

The verbs differ, and in nubes becomes in nubibus. The common movement to meet the Lord is stable. This is a good warning against inferring distinct Greek exemplars from each Latin lexical choice.

Different Latin verbs for the same movement are not, without further evidence, different Greek verbs in the source. The stable meeting-the-Lord clause is an agreement control within the very example that displays variation. Both facts must be retained.

20. 1 Corinthians 15:45: Last Adam and last Lord: a deliberate difference

Adversus Marcionem V.10.7

novissimus Adam in spiritum vivificantem

Text endorsed by Tertullian

Adversus Marcionem V.10.7

dominum enim posuit novissimum pro novissimo Adam

Reading expressly assigned to the heretic

Two forms are present because the passage is a textual accusation. They are not two interchangeable forms endorsed by the author. The earlier II.9.6 comparison with a living soul is also entangled with Genesis 2:7. The episode is a useful control against counting every difference in the work as authorial inconsistency.

This passage supplies a necessary counterweight to the examples of loose quotation. Tertullian can state a precise contrast and identify its sides. Such a local report has a different status from an omission inferred from silence. It tells us what distinction he alleges; the route by which he learned it remains a separate historical question.

21. Ephesians 3:9: Hidden in God or hidden from God: the preposition is the issue

Adversus Marcionem V.18.1

occulti ab aevis in deo qui omnia condidit

Form defended by Tertullian

Adversus Marcionem V.18.1

Occulti ab aevis deo qui omnia condidit

Form he says the heretic produces by removing in

Adversus Marcionem V.18.3

non deo erat occultatum, sed in deo omnium conditore

Tertullian’s explicit explanation

This is a meaningful textual difference, but its author/opponent distribution is explicit. It demonstrates a carefully argued local contrast, not the author’s inability to maintain his own form of the verse.

This is one of the strongest kinds of local evidence in the work: a specified word, an explicit allegation of its removal and an explanation of the resulting sense. The study’s criticism does not erase that evidence. It requires the precision visible here to be demonstrated locally, rather than assumed for everything Tertullian says.

22. 2 Corinthians 4:4: The god of this age: alternatives openly offered

Adversus Marcionem V.11.9

Nos contra sic distinguendum dicimus: In quibus deus, dehinc: aevi huius excaecavit mentes infidelium

Punctuation proposed against Marcion

Adversus Marcionem V.11.11

Simpliciori responso prae manu erit esse huius aevi dominum diabolum interpretari

Alternative interpretation openly offered

Adversus Marcionem V.17.9

Hic erit diabolus

Later recall, with a conditional reference to the Pauline reading

Tertullian first makes the Creator the agent who blinds the unbelievers of this age, then expressly allows the simpler devil reading. The later devil interpretation is already licensed in the earlier discussion. This shows tactical plurality in exegesis, not an unnoticed reversal of a single fixed explanation. The dispute initially concerns punctuation and attachment, not necessarily different words.

The later devil interpretation therefore cannot be presented as a discovery that contradicts Tertullian’s earlier settled view. He had already offered it. What remains stable is his effort to block Marcion’s argument for another god. The range of interpretations he permits serves that effort, and the passage discusses punctuation before it discusses any change of letters.

The passages that agree are part of the argument

A study designed to expose variation must retain the cases of agreement. Otherwise it would simply select the evidence required by its title. Tertullian sometimes repeats a short formulation with striking stability.

Verse	Locations	Shared wording	Finding
Luke 9:20	IV.21.6; IV.22.6; IV.34.16	Tu es Christus	The short confession is repeatedly used for the same identification of the scripturally expected Christ.
Luke 16:16	III.23.3; IV.33.7; V.2.1; V.8.5	Lex et prophetae usque ad Ioannem	The verbal core remains stable while its applications include a historical boundary and the prophetic gifts.
1 Corinthians 15:11	I.20.4; IV.4.5	Sive ego ... sive illi, sic praedicamus	The wording continues to support the unity of apostolic preaching.
Philippians 3:20	III.24.3; V.20.7	municipatus ... in caelis	A distinctive Latin rendering persists, with the Greek term also supplied in Book III.

The result is not an author incapable of remembering anything he has written. It is an author whose citation practice permits both close repetition and considerable flexibility. His defence of the Creator, continuity with prophecy and bodily resurrection is recognizable across the passages compared. That is why the differences require classification rather than a single verdict of reliability or unreliability.

The control works in both directions. Verbal agreement in a short clause does not prove that the writer transcribed a complete verse. Verbal disagreement does not prove that he changed his theological position or possessed a different physical exemplar. The scope actually quoted is the scope that can be compared.

“Marcion’s canon” contains several different historical questions

The title of this post uses canon as a convenient description of the Gospel and apostolic collection discussed in the polemic. The evidence has several levels. A report that a community accepts a letter is different from a report about one word in that letter. A charge that someone removed the word adds a further claim about direction and agency. A claim of personal inspection adds yet another question: how the polemicist acquired his information.

Evidence	What it directly gives the reader	Further question
A book or collection is named	Tertullian’s report about its acceptance, rejection or identity.	What informed that report, and which community or textual form does it describe?
A saying or letter is quoted	The wording of the received polemic, with its quotation context.	How much is direct quotation, translation, paraphrase or remembered formulation?
A word is absent from the quotation	An absence in this particular citation.	Was the source shorter, or did the writer abbreviate?
Marcion is accused of deletion	An explicit claim about textual difference and responsibility.	Is the comparison identified, and what establishes the direction and agent of change?
A critic presents a detailed dispute	Evidence that he had access to information about the dispute.	Did it come from direct inspection, interlocutors, excerpts, an earlier polemic, or a combination?

The final chapter of Book V is useful here. In discussing Philemon, Tertullian says that the letter’s brevity saved it from Marcion’s falsifying hands; he then contrasts its acceptance with the rejection of two letters to Timothy and one to Titus (Adversus Marcionem V.21.1). Those are valuable reports about the collection he attributes to his opponent. They should be recorded as such.

Adversus Marcionem V.21.1

Affectavit, opinor, etiam numerum epistularum interpolare.

“He aimed, I suppose, to tamper even with the number of the letters.” The opinor marks the proposed motive as an inference. The inventory report and the explanation of Marcion’s intention do not have identical evidentiary status. Neither, by itself, narrates the circumstances in which Tertullian personally inspected a collection.

The same distinction applies to quotations. An earlier anti-Marcionite dossier, reports from other readers, excerpted material, direct consultation and recollection are hypotheses about access that require their own evidence. The present comparison cannot identify the author of a hypothetical dossier or establish exclusive dependence on one source. It does show why such questions remain necessary even when Tertullian sounds completely certain.

What a defensible reconstruction should preserve

A reconstruction can use Tertullian extensively and still preserve the distinctions the evidence demands. The crucial step is to stop making one kind of testimony stand in for every other kind.

1. Preserve the Latin before retroverting it. Different renderings of a Greek word, and different Latin constructions of the same thought, must remain visible. They cannot all be treated as separate Greek readings by default.
2. Identify the speaker. A line attributed to Marcion, an opponent’s objection, Tertullian’s reply and an imagined consequence belong to different voices. The staged persona at the opening of Book V makes this especially clear.
3. Mark the source of a textual claim. Explicitly alleged variants such as the missing in in Ephesians 3:9 deserve a different label from omissions inferred because Tertullian fails to quote a clause.
4. Keep Synoptic comparison visible. Matthew’s sword and Luke’s division are a warning against assuming a comparison between two otherwise identical copies of Luke. Modern verse coordinates locate a parallel; they do not prove an ancient title or source.
5. Retain quotation and interpretation as separate layers. A clause explaining what flesh and blood means is not another line in Paul’s letter. A statement that Christ’s action proves what he said is not a report of visual comparison between two manuscripts.
6. Keep the difficult readings difficult. The negative in Galatians 2:5, the mixed covenant passage and the transmission issue at IV.25.10 should not be silently repaired to produce the desired historical text.
7. State confidence at the level of the individual claim. A clear local contrast can support a specific allegation more strongly than a shortened recollection supports the omission of a whole clause. Neither requires a universal verdict on everything in the work.

Several smaller source cautions belong to the same discipline. The supplied text of V.12.6 has flesh and blood in the cleansing language associated with 2 Corinthians 7:1, rather than the conventional flesh and spirit. V.13.12 includes per corpus Christi in an editorial restoration. The Romans gap discussed at V.14.6 does not provide exact deleted verse boundaries. These problems must be attached to the particular claims affected by them. They cannot be resolved by treating the polemicist’s confidence as a substitute for the missing evidence.

The conclusion: direct access must be argued, and precision must be demonstrated

The comparison establishes a substantial gap between the wording of Adversus Marcionem and the documentary precision sometimes assigned to it. Tertullian can omit words he elsewhere says Marcion removed. He can accuse Marcion of substituting a word found in the received Lukan parallel. He can turn agreement between an action and a saying into an argument that the saying was spoken and subsequently erased. He can quote one Pauline verse in several Latin forms and deploy the same phrase through different interpretations.

These observations make the effortless eyewitness reconstruction untenable. They do not make the work useless. The explicit contrasts in 1 Corinthians 15:45 and Ephesians 3:9 show how valuable a clearly located textual allegation can be. The stable repetitions show that verbal agreement is also possible. The study gives us reasons to discriminate among kinds of evidence rather than to accept or reject a whole author at once.

The historical fact to be demonstrated is how Tertullian obtained a particular reading; the textual fact to be demonstrated is how faithfully he reproduced it. Neither fact follows automatically from his accusation that Marcion corrupted Scripture. Direct access, if established, would still leave translation, abbreviation, recollection, exposition and transmission to be assessed. Indirect access would still permit the preservation of genuine information.

That is the force of calling this an eyewitness myth. The imagined scene of Tertullian with Marcion’s canon open before him cannot do the work that only the surviving testimony can do. His polemic supplies that testimony in particular, uneven and sometimes very precise forms. Its authority should be earned at those particular locations, where the words can be read and the argument can be checked.

Sources and method

This post adapts the verse-by-verse comparison of the Gospel and Pauline material in all five books of Adversus Marcionem. The numbered examples retain the study’s distinctions between authorial wording, reported opponent readings, mixed material and textual cruxes. The count of indexed entries is not a count of directly quoted verses or of proven contradictions. English renderings and explanations in this post are newly written; Latin quotations are from the ancient text.

• Tertullian, Adversus Marcionem, Book 1, The Latin Library. The site credits Ernest Evans (1972), transcribed by Roger Pearse (2002). Exact book, chapter and subsection references accompany the examples.
• Tertullian, Adversus Marcionem, Book 2, The Latin Library. The site credits Ernest Evans (1972), transcribed by Roger Pearse (2002). Exact book, chapter and subsection references accompany the examples.
• Tertullian, Adversus Marcionem, Book 3, The Latin Library. The site credits Ernest Evans (1972), transcribed by Roger Pearse (2002). Exact book, chapter and subsection references accompany the examples.
• Tertullian, Adversus Marcionem, Book 4, The Latin Library. The site credits Ernest Evans (1972), transcribed by Roger Pearse (2002). Exact book, chapter and subsection references accompany the examples.
• Tertullian, Adversus Marcionem, Book 5, The Latin Library. The site credits Ernest Evans (1972), transcribed by Roger Pearse (2002). Exact book, chapter and subsection references accompany the examples.
• Dieter T. Roth, The Text of Marcion’s Gospel (Brill, 2015), especially the discussion of IV.25.10 on p. 133. This is the source for the specific interpolation caution, rather than a claim established by the comparison of the Latin Library text alone.
• Benjamin Haupt, Tertullian’s Text of the New Testament outside the Gospels (PhD thesis, University of Birmingham, 2019), for the broader problem of language, translation and citation practice.
• CSEL 47 Latin derivative of V.3, consulted as a check on the retained negative nec in the Galatians 2:5 passage. The site’s English translation was not used.

The numerical results describe the supplied index and its explicit screening rules. They cannot be used as statistical rates of error or compared directly with counts from a differently constituted Clement corpus. The primary historical question here concerns access to Marcionite texts; the citation comparison provides controls on how that access can be inferred, rather than a proof that settles every source relationship in the five-book work.