Paul has just finished reporting the Jerusalem accord: the pillars gave him the right hand of partnership, and nothing was added to his gospel (vv. 6-10). Now, abruptly, the scene shifts to Antioch and the man who had given that right hand is the man Paul opposes publicly. The verse is the hinge of the whole autobiographical section — the point at which Paul’s argument stops being a defense of his independence and becomes a demonstration that the gospel’s truth outranks even the most senior apostolic standing.

Three things carry the weight.

The verb is positional, not violent. ἀντέστην is the second aorist of ἀνθίστημι, ‘take a stand opposite’ — the image of two parties drawn up facing each other, whether armies, litigants, or opponents in debate. Paul is not describing a shouting match but the assumption of a public position.

κατὰ πρόσωπον is openly public, not merely direct. The phrase is a fixed LXX idiom (‘to the face’), and v. 14’s ἔμπροσθεν πάντων confirms that the confrontation was in front of the whole assembly. English ‘to his face’ is close but slightly risks a merely personal reading.

The ground is a verdict, not Paul’s opinion. κατεγνωσμένος ἦν is a periphrastic pluperfect: a condemning judgment had been passed and its result stood. Who passed it — the facts of the case, the gospel itself, God — is left unexpressed, and that reticence is the most contested feature of the verse. Paul does not say ‘I thought he was wrong’; he says the man already stood condemned, and his own stand merely declared what was already the case.

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.

ἀνθίστημι = ἀντί + ἵστημι. The second aorist (ἀντέστην) and the perfect are intransitive: the subject takes up a position facing an adversary. The LXX fixes the register — οὐκ ἀντιστήσεται οὐδεὶς κατὰ πρόσωπόν σου (Josh 1:5), of no one standing against Joshua; and the NT keeps it for facing an opponent in debate (Acts 6:10), in legal-prophetic opposition (2 Tim 3:8), and for taking a stand against the devil (Jas 4:7; 1 Pet 5:9).

Two English candidates must be excluded. ‘Resist’ (Vulgate restiti, and the whole English tradition after it) has drifted in modern English toward enduring pressure applied by someone else — which inverts the clause, since Paul is the one moving. The corpus already confines ‘resist’ to declared liturgical register and reserves it for ἀνθίστημι in paraenetic contexts (Jas 4:7). ‘Oppose’ abstracts the posture away and belongs to the ἀντίκειμαι family (Phil 1:28), which names a resulting state rather than the act of taking the position.

‘Took a stand against’ is used in reading tiers. The aorist is ingressive-complexive: one decisive assumption of position, not a running quarrel — which matters, because Paul is reporting a single public act, not a season of friction.

Note also what the verb is not: it is neither κατηγορέω (‘bring a charge’) nor ἐπιτιμάω (‘rebuke’) nor ἐλέγχω (‘convict’). Paul had all three available. He chose the word for standing in the way.

Built onbdag · bdf

κατὰ πρόσωπον is a fixed prepositional idiom calquing the Hebrew לִפְנֵי / עַל־פְּנֵי and overwhelmingly LXX in flavor. With κατά + accusative in its confrontational-directional sense (BDAG κατά B.1), the phrase means ‘facing, in the presence of, to the face of’ — Acts 3:13 (κατὰ πρόσωπον Πιλάτου), Acts 25:16, 2 Cor 10:1 (κατὰ πρόσωπον ταπεινός), Luke 2:31.

The decisive internal control is v. 14: ἔμπροσθεν πάντων, ‘in front of everyone.’ Whatever κατὰ πρόσωπον contributes by itself, Paul’s own narration makes the event a public one. That blocks ‘privately’ and blocks any reading in which Paul took Peter aside.

What it contributes beyond publicity is harder. Two forces are usually distinguished:

• Face-to-face, in person — Peter was present and addressed directly rather than by letter or report. This is the sense of 2 Cor 10:1, where it is set against ἀπών (‘absent’).
• Openly, in the presence of the assembly — the LXX judicial register, where standing κατὰ πρόσωπον is standing before a party with standing to judge.

These are not exclusive and the clause probably carries both. The corpus policy on πρόσωπον (keep the face audible; cf. εὐπροσωπέω at Gal 6:12 and the προσωπολημψία family) requires that the organ survive in English: ‘face to face’ in reading tiers, ‘against [the] face’ woodenly in the literal tier. ‘Publicly’ and ‘openly’ are excluded from the reading text — each resolves the phrase into one of its implications and deletes the noun.

Built onbdag · bdf · muraoka-lxx

καταγινώσκω = κατά (adverse) + γινώσκω: to reach an adverse judgment about someone, hence ‘condemn.’ Only three NT occurrences, all with a personal object and all forensic in flavor (Gal 2:11; 1 John 3:20, 21, of the heart condemning). LXX: Deut 25:1 (the judge condemning the guilty), Prov 28:11, Sir 14:2.

The form is periphrastic pluperfect: imperfect εἰμί plus perfect passive participle. The corpus already fixes the policy for this construction (ἤμεθα δεδουλωμένοι, Gal 4:3): it names a completed act with a state persisting from it, distinct from both the simple imperfect (‘used to be’) and the aorist (‘was’). So three features must survive in English:

1. The verdict is complete. Not ‘was being condemned,’ not ‘was coming under criticism.’
2. The result stands at the time of Paul’s action. The condemnation was in force when Paul took his stand. ‘He had been condemned’ loses this; ‘he stood condemned’ carries it.
3. The passive has no expressed agent, and none may be supplied.

The third point is where the real contest lies. Who condemned Peter?

What must not happen: ‘because he was clearly in the wrong’ deletes the forensic verb; ‘because I condemned him’ supplies Paul as the agent, which makes the ὅτι clause circular and turns a prior verdict into Paul’s own opinion.

Built onbdag · bdf · muraoka-lxx

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

ὅτι here is causal, not complementizing. It gives the reason for Paul’s stand, not the content of what Paul said; the content arrives at v. 14 (εἶπον τῷ Κηφᾷ … Εἰ σὺ Ἰουδαῖος ὑπάρχων …).

The distinction is exegetically load-bearing. If ὅτι were complementizing, the clause would be Paul’s accusation and the verse would read ‘I opposed him, saying that he stood condemned’ — a charge Paul levels. Read causally, the clause is Paul’s justification to the Galatians: the verdict was already in place and his stand was its recognition. That is a markedly different claim, and it is the one the syntax supports, since ἀνθίστημι is not a verb of speaking and does not take a ὅτι-object.

The corpus keeps three ὅτι functions distinct — causal (‘because,’ Gal 4:6), complementizing (‘that,’ after verbs of speech or knowing), and recitative (after verbs of confessing) — and the classification should be recorded at every occurrence rather than left to the English conjunction to settle.

Built onbdf · bdag

Κηφᾶς is the Greek form of Aramaic כֵּיפָא, ‘rock,’ of which Πέτρος is the translation (John 1:42 makes the equation explicit). The corpus policy is to transliterate as written at every occurrence and never substitute ‘Peter.’

The reason is distributional. Within Galatians, Paul writes Κηφᾶς at 1:18; 2:9, 11, 14, and Πέτρος only at 2:7-8 — and precisely at 2:7-8 the style shifts in other ways as well (τὸ εὐαγγέλιον τῆς ἀκροβυστίας, ἀποστολὴ τῆς περιτομῆς), which is part of the evidence that Paul is reproducing a formula or memorandum there. Rendering both names ‘Peter’ hides a datum that bears on the composition of vv. 7-9.

Built onna28 · metzger-tc · bdag

Antioch on the Orontes, the third city of the empire and the Syrian capital — not Pisidian Antioch, which Acts distinguishes (Acts 13:14). Transliterate and supply no identifying qualifier in the reading text.

The realia matter for the verse’s force, and belong in a note rather than in the rendering (§9). Antioch was the base from which Paul and Barnabas had been sent out, a congregation with a substantial Gentile membership and a large Jewish population in the city, and a place where mixed table-fellowship was the established practice (ἦλθεν … μετὰ τῶν ἐθνῶν συνήσθιεν, v. 12). That is what makes Peter’s withdrawal a reversal rather than a continuation: the practice he abandoned was the practice he had himself been sharing.

It also explains why Paul’s stand had to be κατὰ πρόσωπον. A private correction would have left the public practice — and so the public implication about Gentile standing — unaltered.

The δέ is the transition into a new episode, and per §10 it must be rendered rather than dropped. Whether it is transitional (‘now’) or adversative (‘but’) is genuinely open, and the choice shades the whole paragraph.

Read as transitional, the verse simply moves to the next item in the itinerary that has been running since 1:18 (ἔπειτα … ἔπειτα … ἔπειτα … Ὅτε δέ).

Read as adversative, it sets Antioch against Jerusalem: the pillars who extended the right hand are the party one of whose number Paul now has to oppose.

The corpus policy for δέ is to render per context and classify in the note. ‘But’ is used in the reading text because the contrast with vv. 6-10 is materially present in the narrative, while recording that the contrast is carried by the juxtaposition of episodes rather than asserted by the particle. ‘Now’ in the study tier. Dropping it entirely — common in English versions — removes the seam.

Built onbdf

The verse is often read as evidence about apostolic authority, and the claims it makes need to be stated with care.

What Paul asserts: that on one specific occasion he took a public position against Cephas, and that he did so because a condemning verdict already stood against Cephas’ conduct. The ὅτι clause is a justification, not an accusation (see the syntax note): Paul’s warrant is not his own standing over against Peter’s but the prior fact of the verdict.

What the verse does not say: it does not pronounce on Peter’s doctrine, which v. 12 shows Paul regarded as sound (Peter had been eating with Gentiles); the fault is in practice contradicting conviction, which v. 13 names ὑπόκρισις. Nor does it set Paul above Peter in rank — the argument from 1:11 forward is not that Paul is senior but that the gospel is, and that it binds whoever carries it.

It is worth recording that the early Church felt the difficulty keenly. The Greek commentators (Chrysostom most influentially) were prepared to read the whole exchange as a staged disagreement for the instruction of onlookers; the Latin tradition, with Augustine pressing Jerome hard on exactly this point, insisted that the disagreement was real and that Paul’s rebuke and Peter’s fault were both genuine. Augustine’s position is the one the text’s plain sense requires — but it should be noted as reception, not as evidence for the lexical decisions above.

Built onbdag

Proposed — pending reviewJ. Louis Martyn, Galatians: A New Translation with Introduction and Commentary, Anchor Bible 33A, New York: Doubleday, 1997. · John Chrysostom, Commentary on the Epistle to the Galatians (Homiliae in epistulam ad Galatas commentarius), in Patrologia Graeca 61. · Augustine and Jerome, Correspondence on Galatians 2:11-14 (Augustine, Epistulae 28, 40, 82; Jerome, Epistulae 102, 112), in Corpus Scriptorum Ecclesiasticorum Latinorum 34, 54-56.