This is the hinge of the chapter’s argument. Verses 10-12 have left two statements standing side by side: those who are ἐξ ἔργων νόμου are under a curse (v. 10), and no one is declared righteous ἐν νόμῳ (v. 11). Verse 13 does not soften either. It says that the curse was not lifted but transferred — Christ bought us out from it by coming to stand under it himself.

Four things carry the weight.

The verb. ἐξαγοράζω is the ordinary Hellenistic verb of purchase, used of buying slaves out of the market. It names a transaction but specifies neither price nor payee. Any rendering that supplies either goes beyond the text.

The repetition. κατάρα occurs twice in eight words, and the second occurrence is anarthrous and predicative. Paul does not say Christ became accursed (which would be the adjective he has to hand two words later, ἐπικατάρατος); he says he became a curse — the abstract noun for the concrete, a known Semitic-flavoured idiom, and one that strains Greek enough to have provoked patristic comment.

The ὑπέρ. The preposition is benefactive with the genitive (‘for the sake of, on behalf of’), not ἀντί (‘in place of’). Substitution is a plausible inference from the whole sentence — someone bore what we were under — but it is not what the preposition alone says, and the rendering must not smuggle it in.

The citation. Deut 21:23 concerns the corpse of an executed criminal hung on a post as a public display of the divine verdict already passed. Paul’s wording differs from the LXX at two points, and both differences are consequential. See the sections below.

The verse ends with a comma in the critical text: the ἵνα clauses of v. 14 complete the sentence. No English full stop belongs here.

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

ἐξηγόρασεν is the aorist of ἐξαγοράζω, ἐκ + ἀγοράζω (‘buy in the ἀγορά’). The prefix marks removal from the sphere of ownership, and it is immediately doubled by the following ἐκ τῆς κατάρας — Paul writes the preposition twice over, which is why a rendering that deletes either (‘freed us from’) loses a feature of the Greek.

Four NT occurrences, two in Galatians (3:13; 4:5) and two in the Colossians/Ephesians idiom of buying up time (Col 4:5; Eph 5:16). The Galatians pair is one argument and must receive one English verb.

The corpus glossary fixes this verb at Gal 4:5 as buy out, with ‘redeem’ reserved for declared liturgical register. The reason stands here with full force: English ‘redeem’ has become almost purely religious vocabulary and no longer conveys a transaction, which is the one thing the Greek word contributes. The live metaphor rule (§8) requires that the purchase remain visible.

The aorist is complexive: one completed act viewed whole, not a process. English simple past is correct; the perfect (‘has redeemed’) is permitted only in liturgical register, where the tradition has fixed it.

Built onbdag · bdf

κατάρα is the standard LXX noun for the covenant curse, rendering Hebrew קְלָלָה (Deut 11:26, 29; 27:13; 29:26). It is the formal counterpart of εὐλογία, which stands as its antonym in this paragraph (vv. 8, 9, 14). The pairing is Deuteronomic and must stay visible in English: ‘blessing’ and ‘curse’, not ‘blessing’ and ‘condemnation’.

ἐπικατάρατος is the verbal adjective, the LXX equivalent of אָרוּר, and it is the word of both texts Paul cites in this paragraph (Deut 27:26 at Gal 3:10; Deut 21:23 here). Noun and adjective must share an English root, or the reader cannot see that the citation supplies the very word the preceding clause used.

The interesting datum is that Paul writes κατάρα, the noun, where the adjective was available in the clause and arrives three words later. He does not say Christ became accursed. He says Christ became a curse.

This is abstract-for-concrete, a construction at home in Semitic idiom and not unknown in Greek (cf. 2 Cor 5:21, ἁμαρτίαν ἐποίησεν). The rendering must keep the noun. ‘Became accursed for us’ reads more smoothly and is exactly the smoothing to be avoided: it substitutes the adjective Paul declined, and in doing so it quietly removes the offence the sentence creates.

Built onbdag · muraoka-lxx

The genitive is open, and the readings are not trivially equivalent.

• Source — the curse that law pronounces (so v. 10’s γέγραπται supports: the curse is the law’s own written sentence).
• Possession — the curse belonging to the law-regime, i.e. the condition of being ὑπὸ κατάραν (v. 10).
• Epexegetical — the curse which is the law, in the sense that for those ἐξ ἔργων νόμου the law functions as curse.

The first two are close and both well supported; the third is a minority reading and sits awkwardly with v. 21 (ὁ νόμος κατὰ τῶν ἐπαγγελιῶν; μὴ γένοιτο).

English holds a bare genitive, so the rendering keeps one. ‘The law’s curse’ and ‘the curse the law pronounces’ each adopt a reading.

Note also the article. Paul writes τοῦ νόμου here, against anarthrous ἐν νόμῳ (v. 11), ἐκ νόμου (v. 18), and ὑπὸ νόμον (4:4-5). The standing corpus policy withholds the article where Greek does; here Greek supplies it, and English must too. The article is anaphoric, pointing back to the τὰ γεγραμμένα ἐν τῷ βιβλίῳ τοῦ νόμου of v. 10 — the specific written sanction just quoted.

Built onbdf · bdag

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)

The aorist participle is unmarked. Greek supplies no conjunction, and at least three construals are live:

• Means — ‘by becoming a curse’ (the commonest reading, and the one that makes the participle answer the question the verb leaves open).
• Coincident/attendant circumstance — ‘becoming a curse’, the two actions one event seen from two sides.
• Antecedent — ‘having become a curse’, the becoming preceding the buying-out.

The corpus policy for unmarked adverbial participles (fixed at ὑπάρχων, Phil 2:6, and ὤν, Gal 4:1) applies: the literal tier keeps the bare participle, and any resolution is confined to the smooth tier and flagged. The modern rendering uses ‘becoming’, which in English is as open between means and coincidence as the Greek is.

On the verb itself: γίνομαι, not εἰμί. The corpus already fixes γίνομαι in Christological contexts as ‘come to be / become’ precisely to preserve the contrast with ὑπάρχω and the entry-into-a-new-state force (Phil 2:7; Gal 4:4). The same applies here. ‘Being made a curse’ (the liturgical form, following the Vulgate’s factus) supplies a passive and with it an unnamed maker; the Greek middle names none.

Built onbdf · bdag

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

ὑπέρ + genitive is benefactive: ‘for the sake of, on behalf of’. The corpus entry is already fixed (Phil 1:7; Gal 2:20) and distinguishes it typographically from ὑπέρ + accusative (‘above, beyond’, Phil 2:9).

The question is whether it means in place of. Greek has a preposition for that — ἀντί — and Paul does not use it here. ὑπέρ can shade toward substitution in particular contexts (Phlm 13 is the clearest case), and Hellenistic usage does attest the drift; but the drift is contextual, and the preposition by itself establishes only benefit.

Policy: render ‘for us’ or ‘for our sake’ in every tier; flag the substitutionary construal; do not adopt ‘in our place’ in the reading text.

Built onbdag · bdf

The first-person plural occurs twice and its referent is genuinely contested.

• Jews under law. Paul has used ‘we’ of Jewish believers at 2:15-16, and the law-curse of v. 10 is pronounced by a covenant document addressed to Israel. On this reading the ἵνα clause of v. 14 marks the movement outward: he bought us out, so that the Abrahamic blessing might reach the nations (εἰς τὰ ἔθνη).
• All believers. The second ἵνα of v. 14 switches to λάβωμεν (‘that we might receive the promise of the Spirit’), and that ‘we’ plainly includes the Galatians, who are Gentiles. If the two ‘we’s of v. 13 and the λάβωμεν of v. 14 are one group, the inclusive reading follows.

The structure of v. 14 can be read either way: as a two-stage movement (us → the nations → all of us) or as a single inclusive first person throughout.

The rendering settles nothing. ‘Us’ is as open in English as ἡμᾶς is in Greek, and no qualifier (‘us Jews’, ‘us who were under the law’) may be supplied in any tier.

Paul cites Deut 21:23 (LXX 21:23; MT 21:23) and his wording differs from the LXX at two points.

LXX (Rahlfs): κεκατηραμένος ὑπὸ θεοῦ πᾶς κρεμάμενος ἐπὶ ξύλου. Paul: ἐπικατάρατος πᾶς ὁ κρεμάμενος ἐπὶ ξύλου.

(1) ἐπικατάρατος for κεκατηραμένος. Paul substitutes the verbal adjective that stands in Deut 27:26, which he has just cited at Gal 3:10. The effect is that the two Deuteronomy texts now share a word, and the argument of vv. 10-13 turns on that shared word: the curse pronounced in 27:26 is the curse Christ came under in 21:23. Whether Paul made the substitution or found it in a text-form available to him cannot be determined; the Greek witnesses to Deuteronomy do not support ἐπικατάρατος here.

(2) The omission of ὑπὸ θεοῦ. The LXX has ‘cursed by God’, rendering the Hebrew קִלְלַת אֱלֹהִים תָּלוּי. Paul drops the agent phrase.

πᾶς. Paul retains the LXX’s πᾶς, which has no counterpart in the Hebrew (תָּלוּי, participial, with no quantifier). The universal is load-bearing: it is what makes the sanction applicable without exception, and so applicable to the one hanged at Golgotha. The same πᾶς had already done service at Gal 3:10 (ἐπικατάρατος πᾶς ὃς οὐκ ἐμμένει).

The article ὁ before κρεμάμενος is Paul’s addition (the LXX has bare πᾶς κρεμάμενος), converting the participle into an attributive substantive: ‘everyone who is the one hanging’. The sense is not materially changed.

Hebrew (WLC) Deut 21:22-23, for comparison: וְכִי־יִהְיֶה בְאִישׁ חֵטְא מִשְׁפַּט־מָוֶת וְהוּמָת וְתָלִיתָ אֹתוֹ עַל־עֵץ ׃ לֹא־תָלִין נִבְלָתוֹ עַל־הָעֵץ כִּי־קָבוֹר תִּקְבְּרֶנּוּ בַּיּוֹם הַהוּא כִּי־קִלְלַת אֱלֹהִים תָּלוּי. The Hebrew concerns the corpse of one already executed, hung up as public display; the hanging is not the means of death. Paul is applying to a mode of execution a text originally about post-mortem exposure. That he does so without argument suggests the application was already current — see Reception.

Built onna28 · metzger-tc · bhs

Proposed — pending reviewRichard N. Longenecker, Galatians, Word Biblical Commentary 41 (Dallas: Word, 1990) · John William Wevers, ed., Deuteronomium, Septuaginta: Vetus Testamentum Graecum III.2 (Göttingen: Vandenhoeck & Ruprecht, 1977)

κρεμάμενος is the present middle participle with intransitive force (‘hanging’, not ‘being hung’). The present aspect names the state of suspension rather than the act of hoisting — which matters, since in Deuteronomy the hanging is the display of a body already dead, while in Paul’s application it is the manner of the death.

ξύλον is the plain Greek word for wood or a piece of timber, and in the LXX it renders עֵץ for both living trees and worked wood. Deut 21:22 uses it of the post on which the corpse is exposed.

In early Christian usage ξύλον becomes a near-technical designation for the cross: Acts 5:30; 10:39; 13:29; 1 Pet 2:24. The Deuteronomy text is the evident source of the usage, and Gal 3:13 is the place where the connection is made explicit.

Rendering: tree is the traditional English and is defensible, since ξύλον covers a living tree and the Acts/1 Peter idiom is fixed in English as ‘the tree’. But the referent in Deuteronomy is a post or stake, not a growing tree, and ‘tree’ risks a picture the text does not supply. Policy here: ‘tree’ in reading and liturgical tiers, where the Acts and 1 Peter concordance link requires it; ‘wood’ or ‘a piece of timber’ recorded in the note. Excluded from the reading text: ‘cross’ (substitutes σταυρός, which Paul has at 5:11; 6:12, 14 and declines here, thereby destroying the citation) and ‘gallows’ (imports a modern instrument).

Built onbdag · muraoka-lxx

Documentary perfect passive: a completed act of writing whose result stands in force. The corpus policy is already fixed — ‘it stands written’ where register permits, ‘it is written’ in liturgical register, never ‘it was written’, which converts a standing verdict into a past event.

The introducing ὅτι is causal here, not recitative: it grounds the assertion of v. 13a in the written sanction. (Contrast Gal 3:10, where ὅτι after γέγραπται introduces the quotation itself.) NA28 punctuates with a colon after γέγραπται, marking the following words as the citation; the punctuation is editorial, since the earliest majuscules are unpunctuated, and should be reported as such rather than cited as evidence.

The copula is elided in the quotation (ἐπικατάρατος πᾶς …). English must supply ‘is’; the literal tier brackets it.

Built onbdag · bdf · na28

The verse is textually stable. Three points worth recording:

ξύλου. A few witnesses add the article (τοῦ ξύλου), assimilating to Deut 21:23 LXX, where the preceding verse has supplied a definite referent. The anarthrous reading is secure.

ὑπὲρ ἡμῶν. Scattered witnesses read περὶ ἡμῶν. The ὑπέρ/περί alternation in such constructions is a known Koine fluctuation (cf. the same variation at Gal 1:4) and carries no semantic weight; ὑπέρ is secure here.

ἐπικατάρατος. No Greek witness to Galatians reads the LXX’s κεκατηραμένος, and none supplies ὑπὸ θεοῦ. The divergence from the LXX is therefore not a transmission accident within Galatians but a feature of the text Paul wrote (or of the text-form he cited). This is worth stating plainly, because an argument built on the omission of ὑπὸ θεοῦ needs to know that the omission is not a scribal matter.

Built onna28 · metzger-tc

Pre-Christian Jewish application. Deut 21:22-23 had already been extended from post-mortem exposure to execution itself before the Christian period. 11QTemple (11Q19) LXIV 6-13 applies the hanging language to crucifixion of traitors, and 4QpNah (4Q169) 3-4 i 7-8 refers to one who ‘hangs men alive’ — generally taken as a reference to Alexander Jannaeus’ crucifixions. Whatever the precise reading of those texts, they establish that the move Paul makes without argument was available in his environment. The Deuteronomy sanction had become a live text about crucifixion, and the charge that a crucified man stood under it would have had purchase.

Patristic. Chrysostom’s homily on the verse presses that the one who bore the curse was himself without sin, so that the curse found nothing in him to rest on and was thereby dissolved; he reads the abstract κατάρα as a way of saying ‘bore the curse’. Jerome, who knew the Hebrew, remarks on the absence of ὑπὸ θεοῦ and on the difference between the Hebrew and Greek forms of the text — one of the earliest observations of the divergence. The Vulgate’s factus pro nobis maledictum preserves the noun, which most English versions have not.

Later. The verse became a locus classicus in Reformation soteriology for penal substitution, and in the modern period a locus for objections to it. Both are reception and neither supplies lexical evidence (§11). What the Greek supports and does not support is set out in the ὑπὲρ ἡμῶν and γενόμενος sections above; the doctrinal history is recorded here and kept out of the rendering.

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, 1889) · Martin Hengel, Crucifixion in the Ancient World and the Folly of the Message of the Cross, trans. John Bowden (Philadelphia: Fortress, 1977)