James has just pronounced blessing on the man who endures πειρασμός (v. 12). Now he blocks the inference a sufferer under pressure is most likely to draw: if the trial comes from God, and the trial is pulling me toward collapse, then God is the one pulling. Verse 13 forbids the sentence, and vv. 14-15 relocate the pull to ἡ ἰδία ἐπιθυμία.

Three things carry the verse.

1. One Greek verb, two English words. πειράζω covers both the proving of a person (vv. 2, 12) and the enticement toward evil (vv. 13-14). James is exploiting the single word, and English cannot follow him without either flattening the distinction or hiding the pun. The apparatus keeps one root and flags the shift.
2. A hapax whose force is genuinely open. ἀπείραστος appears nowhere else in the Greek Bible and is rare outside it. Verbal adjectives in ‑τος may be passive-potential (‘cannot be tempted’), passive-factual (‘is not tempted’), or — much more rarely — active (‘does not tempt’). The bare genitive κακῶν leans a different way under each reading.
3. A frame of negatives. μηδείς opens the verse and οὐδένα closes it. What no one may say about God is answered by what God does to no one.

The rendering keeps the negatives at both ends, keeps the source-preposition ἀπό distinct from the agency-preposition ὑπό in v. 14, and does not resolve the hapax in the reading text.

The verb spans a range the corpus policy on πειρασμός has already mapped: proving or putting to the proof (Gen 22:1 LXX, of Abraham; Exod 20:20; John 6:6), attempting to trip someone up (Mark 8:11; Matt 22:35), and enticing toward evil (1 Cor 7:5; 1 Thess 3:5, where ὁ πειράζων is a title). All three are ordinary first-century usage; the lexeme itself does not distinguish them.

James uses the word four times in fourteen words (πειραζόμενος, πειράζομαι, ἀπείραστος, πειράζει) and had just used the noun twice (vv. 2, 12). This is not carelessness. He has affirmed that trials are to be counted joy and that endurance under πειρασμός ends in the crown; now the same root is denied of God. The argument works only because one word covers both fields — the sufferer’s mistake is precisely to slide from the first sense to the second.

The policy adopted here follows the standing πειρασμός entry: the §9 diachronic finding is that modern English ‘temptation’ has narrowed almost entirely to enticement, so ‘test’ is the base gloss and the enticement sense is flagged rather than substituted. The literal and reading tiers therefore keep ‘test’ throughout, and the study tier records that vv. 13-14 shift the weight toward enticement. The liturgical tier keeps the traditional ‘tempt’, where four centuries of English usage have fixed it.

Note the tenses. πειραζόμενος and πειράζομαι are present: the man is speaking mid-trial. πειράζει is a gnomic-durative present: not ‘God did not tempt in this instance’ but ‘God tempts no one, ever, as a matter of what he is.’

μηδείς stands first in the clause, before the participle and the verb, which is the emphatic slot. λεγέτω is a third-person present imperative, and both features are load-bearing.

The third person is a discourse choice. James has been addressing ἀδελφοί μου directly (vv. 2, 16), and here he steps back into the impersonal ruling — the same move made in 1 Pet 4:15-16 and 4:19. English lacks a third-person imperative, so the ‘let’ periphrasis is unavoidable, but converting to a second-person exhortation (‘don’t say…’) erases the shift from address to ruling.

The present aspect is durative: what is forbidden is the practice of saying it, the standing explanation, not one desperate sentence uttered once. James is not policing a slip of the tongue.

μή rather than οὐ is grammatically required with the imperative and carries no independent emphasis, per the corpus policy on μή-forms. Do not read prohibitive force out of the negative’s shape.

ὅτι after λέγω is regularly recitative, functioning as Greek’s quotation mark before direct speech (Mark 1:37; John 1:20). NA28 prints Ἀπὸ θεοῦ with a capital, which is the editors’ judgment that the clause is directly quoted; the earliest majuscules have no such marking and the decision is editorial, not manuscript evidence.

The corpus policy after verbs of confessing renders ὅτι ‘that’ by default and permits direct-quotation punctuation where the clause is a recognizable formula. Here the clause is a formula — a sentence people actually said — so the reading text quotes it. The literal tier keeps ‘say that “…”’ so both the complementizer and the quotation are visible.

What is quoted matters for word order: Ἀπὸ θεοῦ stands before πειράζομαι. The fronting is the point of the sentence. The speaker is not reporting that he is being tested; he is asserting where it comes from.

James writes ἀπὸ θεοῦ here and ὑπὸ τῆς ἰδίας ἐπιθυμίας in v. 14. The corpus keeps the prepositions distinct: ἀπό marks source or origin, ὑπό marks the personal agent of a passive.

The rendering does not adjudicate: ‘from God’ in the reading text preserves the source-sense that is uncontested, and any agency overtone is left to be heard rather than written in. What is excluded is ‘by God’, which would adopt the drift-reading and erase the v. 14 contrast in the same stroke.

Note that θεοῦ is anarthrous here and articular (ὁ θεός) in the following clause. Anarthrous after a preposition is regularly definite by idiom (BDF §255) and carries no indefiniteness.

NT hapax, absent from the LXX, and rare enough in surviving Greek that the lexica work from a thin base. It is a verbal adjective of the ‑τος type built on the πειρα‑ root with α‑privative.

Verbal adjectives in ‑τος are notoriously ambiguous in Greek: they may be passive-factual (‘untested’, like ἄγραφος ‘unwritten’), passive-potential (‘untestable, cannot be tested’, like ἀόρατος ‘invisible’), or occasionally active (‘not testing’, like ἄπταιστος in some usages). All three have been read here, and each pairs with a different construal of κακῶν.

The reading text uses ‘beyond the reach of evils’, which is hospitable to the two passive readings and does not exclude the argumentative function the active reading wants. The literal tier keeps the wooden ‘untemptable [by] evils’; the study tier records all three. What is excluded from every tier is a rendering that reads as settled — ‘God cannot be tempted by evil’ presents the contested potential-passive as fact.

The versional evidence is worth stating precisely because it points the other way from the modern consensus: Jerome and the Syriac tradition both took the word actively. That is reception, not lexical evidence (§11), but it shows the reading is ancient and cannot be dismissed as a modern conjecture.

Proposed — pending reviewPeter H. Davids, The Epistle of James: A Commentary on the Greek Text, New International Greek Testament Commentary (Grand Rapids: Eerdmans, 1982)

A bare genitive plural of the substantival neuter κακά, ‘evil things’. Its case-function follows from whichever reading of ἀπείραστος is adopted, so the two questions cannot be separated.

• With the passive-potential adjective: genitive of agent or means — ‘cannot be tested by evils’. Genitive of agent without ὑπό is unusual but occurs with verbal adjectives.
• With the passive-factual adjective: genitive of respect or separation — ‘untested in the matter of evils’, ‘without experience of evils’.
• With the active adjective: objective genitive or genitive of content — ‘not a tempter with evils’.

English ‘of evils’ in the literal tier holds all three, which is why the wooden rendering keeps it. The reading text’s ‘beyond the reach of evils’ leans to the first two without excluding the third.

On the noun itself: the corpus fixes κακός as ‘evil / harmful’, covering both moral badness and the causing of injury. The plural neuter substantive here names evils as things — instances, not the abstract quality — and should not be singularized to ‘evil’, which would abstract what James left concrete and plural.

Proposed — pending reviewJ. B. Mayor, The Epistle of St. James: The Greek Text with Introduction, Notes and Comments, 3rd ed. (London: Macmillan, 1913)

Two postpositive particles carrying the argument, both of which must be rendered per §10.

γάρ grounds the prohibition: the reason no one may say it is that the claim is false about God. The corpus policy renders γάρ ‘for’ and reserves ‘because’ for cases where English syntax makes ‘for’ ungrammatical; ‘for’ stands here.

δέ joins the second clause to the first. It is not adversative in force — nothing in the second clause opposes the first — but continuative and mildly correlative: the two clauses are the two halves of one denial, one about what God is and one about what God does. ‘And’ is the right English; ‘but’ would manufacture a contrast, which is the same fault the corpus already flags for ascensive δέ at Phil 2:8.

The structure is worth seeing whole:

μηδείς … λεγέτω — prohibition γάρ … ἀπείραστός ἐστιν κακῶν — ground, stated of God’s nature … πειράζει δὲ αὐτὸς οὐδένα — ground, stated of God’s action

αὐτός in the nominative with a finite verb that already carries the person is marked, and in predicate position (not under an article) it is intensive, not the identity-adjective ὁ αὐτός. Render ‘he himself’.

The emphasis does real work. James has just denied something about God’s condition (ἀπείραστος); now he denies something about God’s activity, and the intensive pronoun insists that the denial is about God personally and not about the machinery of providence. Whatever else is at work in a trial, he is not the tempter.

Do not absorb it into the verb (‘and he tempts no one’) — that drops a word James chose. Do not render ‘God himself’, which repeats the noun; the pronoun is doing the pointing.

The verse opens with μηδείς in the emphatic first slot and closes with οὐδένα in the emphatic last slot. The alternation of the negative stem is purely grammatical — μή-forms with imperatives, οὐ-forms with indicatives — and carries no independent weight, per the standing corpus policy on μή/οὐ.

But the placement is rhetorical. The frame sets a prohibited human sentence against a denied divine act, and the two negatives answer each other across the verse:

No one may say it … he tests no one.

English word order permits both to stand where Greek put them, and they should be kept there. Rendering the second clause ‘he does not test anyone’ preserves the sense but loses the closing position of the negative and with it the symmetry.

The verse is stable in the critical tradition; NA28 prints it without a bracketed reading and there is no entry in Metzger’s commentary, which is itself the report.

Two minor points worth recording.

ἀπείραστος / ἀπείρατος. A handful of witnesses read the variant spelling ἀπείρατος (from πειράω rather than πειράζω). The two forms are near-synonymous in Hellenistic Greek and the variation is orthographic-morphological rather than semantic; but ἀπείραστος is the harder form and the one that preserves the paronomasia with the surrounding πειράζω forms, which is decisive for retaining it.

Punctuation of Ἀπὸ θεοῦ. The capital and the quotation are editorial, since the majuscules are unpointed and unpunctuated. NA28’s decision is well grounded in the recitative ὅτι, but it should be reported as an editorial judgment about discourse, not as a manuscript reading — the same discipline the corpus applies to the accented ἄρα/ἆρα and τι/τί decisions.

The early readings divide along the same lines the lexical question does, which is evidence that the ambiguity is ancient rather than a modern manufacture.

Jerome, in the Vulgate, renders Deus enim intentator malorum est — ‘God is not a tempter of evils’ — taking ἀπείραστος actively. The Old Latin tradition behind him shows the same direction. Whatever one concludes about the Greek, this is the earliest sustained interpretive vote on record and it is not for the potential-passive.

The Peshitta likewise construes the clause as a denial of God’s activity rather than of his vulnerability.

Greek commentators, unsurprisingly, feel the pull of the passive. The tradition represented in the catenae reads the clause as asserting that God stands outside the reach of evil — and then supplies exactly the premise the passive reading requires: what a being does not have, it cannot give.

Proposed — pending reviewDale C. Allison Jr., A Critical and Exegetical Commentary on the Epistle of James, International Critical Commentary (London: Bloomsbury T&T Clark, 2013) · Biblia Sacra iuxta Vulgatam Versionem, ed. R. Weber and R. Gryson, 5th ed. (Stuttgart: Deutsche Bibelgesellschaft, 2007)

The verse is doing something narrower than it is often made to do, and §6 requires that the narrowness be preserved rather than expanded in either direction.

What is denied: that God is the one from whom enticement toward evil proceeds, and that God performs πειράζω toward anyone. The second clause is unrestricted — οὐδένα, no one.

What is not denied: that God brings, permits, or superintends trials. James has said the opposite eleven verses earlier: πειρασμοῖς περιπέσητε ποικίλοις is to be counted joy, and the crown of life is promised to the one who endures πειρασμόν. Nor is Gen 22:1 LXX (ὁ θεὸς ἐπείραζεν τὸν Ἀβραάμ) contradicted, nor Deut 8:2, nor Matt 6:13.

The distinction James draws is not between two vocabularies — he uses one verb for both — but between a purpose and an agency. The trial that proves is God’s; the pull toward collapse is not, and v. 14 names where it comes from instead.

The rendering is deliberately calibrated so that neither the metaphysical nor the merely relational reading is closed off. ‘Beyond the reach of evils’ can be heard either way, which is where the Greek leaves the reader.

Proposed — pending reviewRalph P. Martin, James, Word Biblical Commentary 48 (Waco, TX: Word Books, 1988)