Paul has just said he opposed Cephas to his face because he stood condemned (v. 11). This verse supplies the evidence, and it does so with an exact chronology: two arrivals, one practice, one reversal, one motive.

The grammar carries almost all the weight. συνήσθιεν is an imperfect of settled habit — Cephas was in the practice of eating with Gentile believers at Antioch. ἦλθον is an aorist: the arrival of the men from James is a single point. Then two more imperfects, ὑπέστελλεν καὶ ἀφώριζεν, describe not a decision but a process — a progressive pulling back, caught in the act. And the closing participle names the cause without softening it: φοβούμενος, he was afraid.

Three things are left deliberately unstated, and the apparatus must leave them so. Who the τινες were and what authority they carried (ἀπὸ Ἰακώβου is a source-preposition, not a commissioning formula). What the withdrawal consisted of — table fellowship generally, or the shared meal that included the Lord’s Supper. And who οἱ ἐκ περιτομῆς were: the visitors themselves, a wider Jerusalem constituency, or non-Christian Jews in Antioch or Judea. Each has serious defenders, and each English paraphrase that resolves one of them does so silently.

πρὸ τοῦ … ἐλθεῖν τινας is the standard Koine temporal construction: πρό + genitive articular infinitive, with the subject of the infinitive in the accusative (BDF §403; cf. Matt 6:8; John 1:48; 13:19; Gal 3:23).

Two features for the rendering. First, τινας is the accusative subject of the infinitive, not an object — the literal tier should show the articular infinitive as a nominal unit (“before the coming of certain people”) so the construction stays visible, while smooth registers may resolve to a finite clause. Second, the aorist infinitive is complexive: the coming is viewed as one whole event, which is what sets up the aorist ἦλθον in the second half of the verse.

Note also the word order: γάρ is postpositive and would normally stand second, but πρὸ τοῦ is felt as a single indivisible unit, so the particle is displaced to third position. The literal tier should preserve the Greek order and let the apparatus record the displacement.

Built onbdf · bdag

Paul’s habit in Galatians is to name opponents by indefinite pronoun and leave them unidentified (1:7 τινές; 2:12; 5:10 ὁ ταράσσων; 6:12-13 ὅσοι θέλουσιν). Per corpus policy, no identification is supplied in any tier: “the Judaizers,” “James’s men,” and “emissaries from James” each settle a contested historical question by translation.

ἀπὸ Ἰακώβου is the crux. ἀπό + genitive is the source-preposition, kept distinct throughout this corpus from παρά (personal source), διά (mediation), and ὑπό (agency). It says the men came from James — from Jerusalem, from his circle — and says nothing about commission. Paul has available the vocabulary of sending (ἀποστέλλω, πέμπω) and the preposition of agency, and he uses neither.

Note that Ἰάκωβος here is the man of 1:19 and 2:9; per the standing entry, render “James” in reading tiers and keep “Jacob” available in the literal tier only where the patriarchal echo is exegetically live, which it is not here.

Built onbdag · bdf

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.

συνήσθιεν is σύν + ἐσθίω, one of the σύν- compounds this corpus consistently keeps audible: the togetherness is the lexeme’s content, not a contextual inference. All NT occurrences concern shared table (Luke 15:2; Acts 10:41; 11:3; 1 Cor 5:11; Gal 2:12), and in Jewish-Gentile settings the verb is loaded — Acts 11:3 makes συνέφαγες αὐτοῖς the charge against Peter, in the same situation and almost the same words.

The iterative imperfect is decisive. Paul does not say Cephas ate one meal; he says eating with Gentiles was his practice. English must mark the duration (“used to eat,” “was in the habit of eating”); a simple past (“he ate with the Gentiles”) reduces a settled policy to an incident and lets the reversal look like a minor inconsistency rather than the abandonment of a position.

The apparatus flags both; the rendering adjudicates neither. μετὰ τῶν ἐθνῶν is μετά + genitive (accompaniment), to be kept distinct in English from the temporal μετά + accusative of 2:1.

Built onbdag · bdf

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

Per the standing entry, ἔθνη renders the LXX גּוֹיִם and marks outsiders to the covenant people rather than an ethnic category: “the nations” in the literal tier, “Gentiles” in smooth registers, never “pagans” or “heathen.”

Here the referent is specifically Gentile believers at Antioch — the men Cephas had been eating with — but the noun does not say so, and no qualifier may be supplied in the reading text. Paul uses the same bare noun of the mission sphere in vv. 2, 8, 9 and of the parties pressured in v. 14, and the unbroken vocabulary is part of the argument: the people whose table Cephas left are the people whose gospel Paul had laid before the pillars.

Built onbdag

ὑπέστελλεν is from ὑπό + στέλλω, and the root image is furling — drawing in sail, hauling something back under cover. Four NT occurrences: Acts 20:20, 27 (Paul did not ὑπεστειλάμην from declaring the whole counsel of God — hold back, keep something in reserve), Heb 10:38 (quoting Hab 2:4 LXX, ἐὰν ὑποστείληται, of shrinking back), and here.

Two components of the Greek must survive. The verb denotes retraction of something previously extended, which is why it fits a man abandoning an established practice; and it carries the connotation of acting from reserve or timidity, which is why φοβούμενος sits so naturally with it. ‘Withdraw’ is the serviceable reading-tier gloss; ‘draw back’ keeps the retraction visible.

Excluded: ‘stop eating with them’ (deletes the verb and supplies an object), ‘avoid them’ (names the outcome, drops the pulling-in), ‘hold himself aloof’ (imports a social attitude the verb does not carry).

Built onbdag · bdf

The corpus has already fixed ἀφορίζω as ‘set apart’ in positive contexts and ‘separate / exclude’ in negative ones, with one English root across both — and Galatians is precisely why the rule exists. At 1:15 God ἀφορίσας Paul from his mother’s womb; at 2:12 Cephas ἀφώριζεν ἑαυτόν from the Gentile table. Same verb, same letter, opposite direction, and the reader should be able to see it.

The lexeme is ἀπό + ὁρίζω, ‘mark off by a boundary.’ Its LXX home is cultic setting apart (Exod 13:12; Lev 20:26, where Israel is ἀφωρισμένοι from the nations), and the negative face is exclusion from a company (Luke 6:22; Matt 25:32). Here the boundary-drawing is the whole point: Cephas re-erects the line between Jew and Gentile that his own practice had treated as down.

The active with a reflexive object is marked — Greek had the middle available — and underlines agency: no one excluded him; he drew the line himself. English must keep the reflexive (‘separating himself’), not a passive or an intransitive (‘he kept apart’). The iterative imperfect again marks a process under way, not a finished policy.

Some note the possible resonance with the root behind Φαρισαῖος (Aramaic pərîšayyāʾ, ‘separated ones’). It may be recorded as a resonance; per §11 it supplies no evidence for the rendering, and nothing in the clause activates it.

Built onbdag

The verse is built on an aspectual contrast that English must reproduce:

• συνήσθιεν (impf) — the prior settled practice
• ἦλθον (aor) — the arrival, a single point
• ὑπέστελλεν … ἀφώριζεν (impf, impf) — the reversal, in progress

The aorist between two sets of imperfects is doing work. It locates the cause precisely and makes the withdrawal a response to it. And the imperfects after it are not merely durative but inceptive-progressive: Paul describes the retraction as something he watched happening, which is also why his own response (ἀντέστην, v. 11) was immediate and public.

English renderings that flatten all five verbs to simple pasts (‘he ate with the Gentiles; but when they came he withdrew and separated himself’) produce a tidy narrative and lose the argument. ‘Used to eat … began drawing back and separating himself’ keeps both the prior habit and the process, and the literal tier should mark the duration woodenly (‘was eating together with … was drawing back … was separating’).

Built onbdf

φοβούμενος is the ordinary verb of fear, middle/passive in form with no passive force. The participle is a present adverbial participle with no causal marker in Greek — per standing policy on unmarked adverbial participles (ὑπάρχων, Phil 2:6; ὤν, Gal 4:1), the literal tier keeps the bare ‘fearing’ and the causal resolution (‘because he was afraid’) is confined to smooth registers and recorded here.

The lexical point is that Paul does not say Cephas was persuaded, pressured, or mistaken. He names fear. Per §6 the rendering must not soften it: ‘out of concern for’, ‘being wary of’, and ‘under pressure from’ all relocate a named emotion into circumstance or prudence. Equally, nothing may be added — Paul does not say what Cephas feared would happen.

With the accusative the sense is fear of someone (as against φοβέομαι μή + subjunctive, apprehension that something may be so). Compare the prolepsis entry at Gal 4:11, a different construction: here τοὺς ἐκ περιτομῆς is a plain object.

Built onbdag

Two standing decisions apply. περιτομή is retained as the abstract noun ‘circumcision’ even where it designates persons, since the abstract-for-concrete idiom is the same construction Paul parodies with κατατομή at Phil 3:2. And ἐκ + genitive with an articular substantivized phrase marks derivation and group membership (οἱ ἐκ πίστεως, Gal 3:7; οἱ ἐκ νόμου, Rom 4:14; οἱ ἐκ τῆς Καίσαρος οἰκίας, Phil 4:22) — ‘of’ or ‘belonging to’, not motion away.

So: ‘those of the circumcision’, a group defined by it.

Per §5 the rendering adjudicates none of these: ‘the circumcision party’, ‘the Jewish Christians’, and ‘the Jews’ each adopt one, and the last additionally risks an ethnic verdict the Greek does not make. Keep the bare phrase and flag the range.

Built onbdag

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

γάρ grounds the preceding clause and must always be rendered (§10). The question is what it grounds. The nearest candidate is κατεγνωσμένος ἦν (v. 11): Cephas stood condemned, for this is what he did. On that reading v. 12 is the indictment’s evidence and v. 13 the aggravation.

A second possibility reaches back to κατὰ πρόσωπον αὐτῷ ἀντέστην: the γάρ explains why a public confrontation was necessary. The two are not exclusive — the public reversal is both the condemnation’s ground and the confrontation’s occasion — and the English ‘for’ leaves both audible, which is why it is retained over ‘because’ or ‘since’.

Note also the ὅτε … ὅτε frame. v. 11 opens with ὅτε δὲ ἦλθεν Κηφᾶς; v. 12 has ὅτε δὲ ἦλθον, of the τινες. Two arrivals with the same verb and the same connective, and the second undoes what the first made possible. The repetition is deliberate and should not be varied in English for style.

Built onbdf

Two variants, both bearing directly on the sense.

ἦλθον (plural) / ἦλθεν (singular). The plural, read by the earliest and best witnesses, takes the τινες as subject: when they came. The singular, supported by a significant body of witnesses including the Latin tradition, has James as subject — when he came — which would place James himself in Antioch. Internally the plural is far stronger: the τινες are the nearest plural antecedent, they are the subject of the infinitive in the first half of the verse, and the singular creates a narrative James’s presence nowhere else attests. The singular is most plausibly a scribal slip occasioned by the proximity of Ἰακώβου, possibly reinforced by the singular ἦλθεν of v. 11.

τινας / τινα. A thinly supported singular would give “a certain man from James,” and coheres with the ἦλθεν variant. It is secondary on the same grounds.

Together the two singular readings constitute a coherent secondary line in which James visits Antioch in person. The modern critical editions print the plurals, and the internal evidence is decisive; but the existence of the variant is worth recording, since it shows how early readers already felt the question of James’s responsibility pressing on the text.

Built onna28 · metzger-tc

The verse was the sharpest exegetical flashpoint in the early Latin West, and the dispute is instructive.

Chrysostom and much of the Greek tradition read the episode as real but Peter’s withdrawal as a measured accommodation, with Paul’s rebuke directed past Peter at the onlookers — a reading motivated by reluctance to attribute genuine fault to an apostle, and resisted by Paul’s own κατεγνωσμένος (v. 11) and φοβούμενος here.

Jerome, in his Galatians commentary, took over a version of this: the confrontation was a staged disagreement (simulatio) for the benefit of the onlookers, neither apostle actually erring. Augustine objected forcefully in their correspondence, insisting that if Paul wrote that Peter was at fault and acted from fear, then he was, and that an exegesis which makes the apostolic record a pretence undermines the credibility of all Scripture. The ensuing exchange is one of the sharpest patristic debates about the honesty of the biblical narrative, and Augustine’s position prevailed in the Western tradition.

The translator’s interest in this is methodological rather than doctrinal. Augustine’s objection is precisely the discipline of §6: if the text says fear, the translation says fear. The pressure to render φοβούμενος as something more prudential — ‘out of deference’, ‘concerned about’ — is the same pressure Jerome felt, and it is to be resisted for the same reason.

Proposed — pending reviewJerome, Commentarii in epistulam Pauli ad Galatas, and Augustine–Jerome correspondence, Augustine, Epistulae 28, 40, 67, 71, 73, 82 (CSEL 34, 44). · John Chrysostom, Commentarius in epistulam ad Galatas, PG 61.