Paul has just said that he laid his gospel before the Jerusalem leadership (v. 2). Verse 3 is the first piece of evidence that the conference went his way, and he offers it not as a report of a decision but as a fact about a person standing beside him: Τίτος, uncircumcised, was not compelled.

Four things carry the weight of the verse.

1. ἀλλʼ οὐδέ — an ascensive pile-up. Not ‘and Titus was not circumcised either,’ but ‘not even Titus’: the hardest conceivable test case, a Greek brought by Paul into the middle of a Jerusalem meeting, was left alone.
2. Ἕλλην ὤν — a bare participle with no concessive marker. Greek says only ‘being a Greek.’ English nearly has to say ‘though he was a Greek,’ but that ‘though’ is an inference from the argument, not a word on the page.
3. ἠναγκάσθη — a passive with no agent, and the verb Paul elsewhere uses of the pressure applied at Antioch (2:14) and by the agitators (6:12). The denial is narrowly framed: nobody forced it. What was or was not done voluntarily is a separate question, and the verse does not answer it.
4. The sentence does not finish. Verse 4 opens with a dangling διὰ δέ and v. 5 with a relative οἷς that has no clean antecedent; by v. 6 Paul has restarted the construction altogether. The syntax breaks exactly where the subject is hottest, and that break is data — not a defect to be repaired by translation.

Proposed — pending reviewF. F. Bruce, The Epistle to the Galatians: A Commentary on the Greek Text, New International Greek Testament Commentary, Grand Rapids: Eerdmans, 1982.

Two particles, both load-bearing, and both rendered (§10).

ἀλλά is the strong clause-initial adversative already fixed in this corpus (Phil 3:7; Gal 5:6). Its counterweight is not the immediately preceding clause but the whole apprehension of v. 2 (μή πως εἰς κενὸν τρέχω): Paul feared his running might prove empty; on the contrary, here is the proof it did not. Rendering it ‘yet’ or ‘however’ for style is excluded where a second ἀλλά stands nearby (v. 7, ἀλλὰ τοὐναντίον).

οὐδέ here is ascensive, not continuative. The continuative use (‘nor,’ as at Gal 1:17) coordinates a second limb of an existing negation; the ascensive use adds a further and harder item to a point already made (BDF; cf. οὐδὲ γάρ at Gal 6:13, already fixed as ‘for not even’). The signal is that no preceding negative clause stands available for οὐδέ to continue — v. 2’s negation is inside a subordinate clause of apprehension. So: ‘not even.’

The effect is a superlative by implication. Titus is produced as the limiting case: if circumcision were going to be required of anyone, it would be required of him, in Jerusalem, in Paul’s own company.

Τίτος is a Latin praenomen (Titus) transliterated into Greek — itself a small datum about the man’s background. He is named nowhere in Acts, which is one of the standing puzzles of Pauline chronology, but is heavily present in 2 Corinthians (2:13; 7:6, 13-14; 8:6, 16, 23; 12:18) and is the addressee of the letter bearing his name, where he is called γνήσιον τέκνον (Titus 1:4).

What matters for Galatians 2:3 is that Paul names a person rather than citing a ruling. The argument of 1:11 - 2:10 is that his gospel owes nothing to Jerusalem; here the demonstration is a body, present and uncircumcised, rather than a minute of a meeting. The same move recurs at 2:11-14, where the issue is again not a principle but what a named man did at a named table.

Note that Paul does not call Titus a convert of his, a delegate, or a test case. He says only ὁ σὺν ἐμοί — ‘the one with me.’ The reticence is deliberate and should not be filled in.

Article + prepositional phrase in attributive position, the construction already fixed at Phil 4:21 (οἱ σὺν ἐμοὶ ἀδελφοί) and 1 Pet 5:2 (τὸ ἐν ὑμῖν ποίμνιον). The article converts ‘with me’ into a defining modifier of the name, so the literal tier keeps the wooden ‘the [one] with me’ to display the construction, and smooth registers resolve to a relative clause.

Two cautions.

First, the modifier is restrictive, identifying Titus by his presence in Paul’s party. Rendering it as a non-restrictive aside (‘Titus, who happened to be with me’) turns a credential into a parenthesis. It matters that Titus was there — in the city, in the meeting, available for inspection.

Second, σύν is companionate, not instrumental, and must stay lexically distinct from ἐν (the corpus keeps σὺν Χριστῷ and ἐν Χριστῷ apart on exactly this ground). Paul had μετά + gen available and used it of Barnabas two verses earlier (μετὰ Βαρναβᾶ, v. 1); the switch to σύν with Titus is recorded here rather than pressed, since the two prepositions overlap heavily in Koine.

The emphatic pronoun ἐμοί is the regular form after a preposition and supports no argument about emphasis.

Ἕλλην properly names a Greek by descent or language, but in Jewish Greek it had become the standing counterpart to Ἰουδαῖος — the non-Jew considered as an outsider to the covenant (John 7:35; Rom 1:16; 10:12; 1 Cor 1:24; Gal 3:28; Col 3:11). It is thus not a statement about Titus’ ethnicity so much as about his standing: he belongs to the ἀκροβυστία (v. 7).

The noun is anarthrous and qualitative: ‘a Greek,’ a member of that class, not ‘the Greek.’

Per §5 and the standing ἔθνη policy, no pejorative or evaluative freight may be added. ‘Gentile’ is admissible in smooth registers where English readers would otherwise hear a claim about nationality, but it is a translation of ἔθνη, not of Ἕλλην, and Paul distributes the two words deliberately within this chapter (τὰ ἔθνη at vv. 2, 8, 9, 12, 14; Ἕλλην here). Keeping them distinguishable in English preserves that distribution.

Excluded from the reading text: ‘a pagan’ (imports a later polemical category, §9), ‘uncircumcised’ (supplies the inference the sentence draws and deletes the noun), and ‘non-Jew’ (a negative formulation Greek had available in οὐκ Ἰουδαῖος and did not use).

The participle of εἰμί stands bare, with no concessive particle. Greek had καίπερ (already fixed in this corpus at Gal 4:4 as ‘although indeed’) and καίτοι available for marking concession and uses neither.

This falls under the standing policy on unmarked adverbial participles (ὑπάρχων, Phil 2:6; ὤν, Gal 4:1): where a concessive reading is contextually strong but grammatically unmarked, the literal tier keeps the bare participle and any ‘though’ is confined to the smooth tier with the inference flagged. So the literal line reads ‘[being] a Greek,’ the reading text ‘Greek though he is.’

The concessive reading is strong here — the whole force of ἀλλʼ οὐδέ depends on Titus’ Greekness being the reason one would have expected circumcision — but it is an inference from the argument, not a feature of the syntax. The Vulgate’s cum esset is a reading, and a reasonable one; it is not evidence (§11).

The standing ἀναγκάζω entry (fixed at Gal 6:12) is ‘compel / press’: coercion short of physical force but leaving no honorable exit — social, ecclesial, theological pressure. Galatians uses the verb at three decisive points, and the pattern is the argument of the letter in miniature:

• 2:3 — Titus was not compelled.
• 2:14 — Cephas’ withdrawal compels the Gentiles to Judaize.
• 6:12 — the agitators compel circumcision to avoid persecution.

One English root at all three points is mandatory, or the pattern disappears. ‘Urge’ and ‘encourage’ drain the coercion; ‘force’ in modern English implies physical constraint the verb does not assert.

Three features of the form matter.

The passive leaves the agent unexpressed. Paul does not say who would have done the compelling. Reading tiers must not supply ‘the apostles,’ ‘James,’ or ‘the false brothers’ — all three have been proposed, and the sentence names none.

The aorist is complexive, viewing the whole episode as a single completed matter, not a process of resisted pressure.

The negation is narrow. What is denied is compulsion, not circumcision. This is the pivot on which the verse’s whole history of interpretation turns.

Proposed — pending reviewJ. Louis Martyn, Galatians: A New Translation with Introduction and Commentary, Anchor Bible 33A, New York: Doubleday, 1997.

Aorist passive infinitive, complementary to ἠναγκάσθη. Per the standing περιτομή / περιτέμνω policy, the term is rendered plainly as circumcision and the περι- element is kept recoverable in notes, since Paul coins κατατομή against it at Phil 3:2 and the wordplay depends on the morphology staying visible.

The aorist is complexive: the rite as a single act, not a state entered. Contrast the deliberate presents at Gal 5:2-3 and 6:13 (περιτέμνεσθαι, περιτεμνόμενοι), which the corpus already fixes as naming a course being entered on rather than a completed event — the men now undergoing or pressing the rite, not the already-circumcised. Paul’s switch of aspect between 2:3 and 5:2 is recorded rather than pressed.

The passive is genuine: Titus would have been the recipient, and the agent of the circumcising is as unexpressed as the agent of the compelling. Keep the English passive; ‘to undergo circumcision’ is permitted in the study tier, where it makes the recipient’s role visible without supplying an agent.

Verse 3 does not end a sentence so much as leave one hanging. The critical text punctuates with a high point, and what follows is among the roughest Greek in the Pauline corpus:

• v. 4 opens with διὰ δὲ τοὺς παρεισάκτους ψευδαδέλφους — a causal prepositional phrase with no verb to attach to, and NA28 closes the verse with a dash.
• v. 5 opens with οἷς, a dative relative whose antecedent is the false brothers — but the syntax of v. 4 never reached a point where a relative could be hung on it.
• v. 6 restarts with ἀπὸ δὲ τῶν δοκούντων, another prepositional phrase, and then abandons that construction too, resuming with ἐμοὶ γὰρ οἱ δοκοῦντες after two parenthetical remarks.

This is anacoluthon on a scale the corpus has already had to handle at Jas 3:8 and Gal 1:20, and the policy is the same: the break is the text and is not repaired in translation. Mark it with dashes, supply no verb unbracketed, and do not close an English sentence where Greek leaves the construction open.

The textual tradition is itself evidence here. Some witnesses omit the οὐδέ at the head of v. 5, which has the effect of reversing the sense (‘to whom we yielded for a time’) and producing a narrative in which a concession was made. That reading is almost certainly an ancient attempt to resolve the difficulty of vv. 3-5 — and Marcion was reported in antiquity to have read the passage so as to make Titus’ circumcision the thing conceded. Where scribes smooth, the harder text is the text, and the harder text is one that breaks off.

Proposed — pending reviewHans Dieter Betz, Galatians: A Commentary on Paul's Letter to the Churches in Galatia, Hermeneia, Philadelphia: Fortress, 1979.

The verse itself is textually stable in NA28; the instability lies immediately after it.

v. 3. No variant of consequence. Some minuscules read Ἕλλην ὑπάρχων for Ἕλλην ὤν, an assimilation to Paul’s more emphatic being-verb (cf. Ἰουδαῖος ὑπάρχων at 2:14 in the same chapter) and secondary; the harder and better-attested ὤν stands. A few witnesses add an article before Ἕλλην, converting a qualitative predicate into a definite designation — the same direction of scribal clarification the corpus records elsewhere, and equally to be avoided.

v. 5, οὐδέ. This is the significant seam. The critical text reads οἷς οὐδὲ πρὸς ὥραν εἴξαμεν; a Western strand (reflected in part of the Latin tradition) omits οἷς or οὐδέ, yielding ‘to whom we yielded for a time.’ The omission produces an intelligible narrative out of an unintelligible one, which is exactly why it is secondary. Metzger’s committee treats the omission as a Western attempt to relieve the anacoluthon; the retention of both words is the harder reading and is also the reading that fits Paul’s argument, since a concession at Jerusalem would concede the Galatian case.

Proposed — pending reviewJ. B. Lightfoot, Saint Paul's Epistle to the Galatians, 10th ed., London: Macmillan, 1890.

Vulgate: sed neque Titus, qui mecum erat, cum esset gentilis, compulsus est circumcidi.

Three things to note.

neque renders οὐδέ continuatively (‘neither … ’), losing the ascensive force. Latin had ne … quidem available for ‘not even’ and does not use it here. This is the source of the older English ‘But neither Titus.’

cum esset supplies the concessive (or causal — cum is ambiguous in Latin too) that the Greek participle leaves unmarked. A reasonable reading, and recorded as such; not evidence (§11).

gentilis for Ἕλλην substitutes the covenant-outsider category for the ethnic-linguistic noun, effectively translating ἔθνη rather than Ἕλλην. It is a good guide to the sense and a poor guide to the word — precisely the distinction the rendering above tries to preserve by keeping ‘Greek’ in the main text.

Peshitta likewise renders with a form of the standard Syriac term for the non-Jewish nations, and like Latin supplies a subordinating conjunction before the predicate. Both versions resolve the participle because neither language has a comfortable equivalent for a bare adverbial participle; the resolution is a translational necessity in those languages and carries no weight for English.

The verse became a patristic battleground almost entirely because of the narrow scope of ἠναγκάσθη.

Jerome and Augustine quarrelled over this chapter at length, though their dispute centred on vv. 11-14 (whether Paul’s rebuke of Peter was genuine or staged). Jerome’s instinct throughout the chapter was to minimize real conflict among the apostles; Augustine insisted the text means what it says. The same methodological divide governs how each handles v. 3: whether Paul’s careful ‘not compelled’ is read as conceding something or as denying everything.

Chrysostom, in his homilies on Galatians, reads the verse as a flat denial that circumcision occurred, and takes ἀλλʼ οὐδέ as the ascensive climax it is — Titus is produced because he was the hardest case. He also notes that Paul does not say the apostles forbade it, only that no one compelled it, and treats the reticence as deliberate.

Marcion was reported by his opponents to have read the passage in a way that made the ‘false brothers’ the Jerusalem apostles themselves. The report tells us the passage’s roughness was exploited early; it tells us nothing secure about his text.

In the modern period the verse has been pressed into service in the Acts-and-Galatians chronology debate (is Gal 2:1-10 the Acts 15 conference, the Acts 11:30 famine visit, or neither?), and the Titus/Timothy comparison with Acts 16:3 has been argued in every direction. None of that is lexical evidence, and the rendering above is deliberately compatible with each reconstruction.

Proposed — pending reviewJohn Chrysostom, Commentary on the Epistle to the Galatians, in Nicene and Post-Nicene Fathers, First Series, vol. 13, ed. Philip Schaff, Buffalo: Christian Literature Company, 1889.

Paul could have written a sentence that settled everything: Titus was not circumcised. He wrote instead that Titus was not compelled to be circumcised — and then broke off.

The choice is not evasion. Throughout Galatians the thing at issue is never the surgical fact but the ἀνάγκη: whether anything must be added to Christ for a Gentile to stand in the people of God. That is why the same verb returns at 2:14 (Cephas’ withdrawal compels Gentiles to live as Jews) and at 6:12 (the agitators compel circumcision to escape the cross). Paul’s quarrel is with compulsion, which is why he can circumcise Timothy (Acts 16:3) without contradiction and will not have Titus circumcised under pressure.

Two disciplines follow for the translator.

First, do not resolve what the verb leaves open, in either direction. Rendering ‘Titus was not circumcised’ is more useful to a historical reconstruction and less faithful to the Greek. §6 forbids shading a rendering toward whatever we would prefer the text to have settled.

Second, do not repair the broken sentence. The anacoluthon of vv. 3-6 is the most honest feature of the passage: a man writing about a wound still open, who starts three sentences about the false brothers and finishes none of them. The scribal tradition tried to smooth it, the versions had to subordinate it, and every fluent modern English rendering is tempted to tidy it. What is handed on is the roughness — and the rough text says less than we want and exactly what Paul wrote.