HistoricalChristian.Faith

Deuteronomy 32:8: "sons of God," "sons of Israel," or "angels of God"?

Type: supporting argument (off-timeline) · Assessment: sound — as the text-critical claim that the Hebrew reading behind the Septuagint and a Dead Sea Scroll (bene elohim / bene el, "sons of God" / "sons of El") is the earlier reading of Deuteronomy 32:8, and the Masoretic "sons of Israel" a later theological correction that scrubbed the verse of its lesser divine beings. On the manuscript evidence, transcriptional probability, and an undeniable parallel expurgation two verses later (32:43), this is the majority position in modern text-critical scholarship and it holds. What it does not by itself establish is the cosmology built atop it — Heiser's "Deuteronomy 32 worldview" of rebel territorial elohim — which is a further, separately-rated step. The verdict here is about the earlier text, not the whole theology it is made to carry. · advanced by Michael S. Heiser.

Related arguments: the Deuteronomy 32 worldview — see also Psalm 82: real divine beings, or human judges?

The claim

It is far more likely both texts were intentionally altered in MT for the same reason: to eliminate a reference to divine beings in the name of expunging allegedly polytheistic language. It is inconceivable to suppose that a scribe would have done the reverse, altering an innocuous bene yisrael to a potentially explosive bene elohim. The LXX reading therefore sufficiently explains how the MT reading could have arisen, but the alternative fails this test.

— Michael S. Heiser, "Deuteronomy 32:8 and the Sons of God," Bibliotheca Sacra 158 (2001): 52–74. The interpretation being weighed is the narrow text-critical one: that at Deuteronomy 32:8 — "When the Most High gave the nations their inheritance… he fixed the borders of the peoples according to the number of the …" — the original Hebrew read "sons of God / sons of El," not the received "sons of Israel." (The whole "divine council" cosmology leans on this base; whether the base delivers that cosmology is a separate question, taken up below and on the linked pages.)¹

The three readings, and what actually witnesses them

The end of the verse is preserved three ways, but the three are not three equal Hebrew rivals:

• "Sons of Israel" (bene yisrael) — the Masoretic Text, joined by the Samaritan Pentateuch. This is the reading behind most older English Bibles.
• "Sons of God / gods" (bene elohim) — read cleanly by the Qumran fragment 4QDeut^j (4Q37), which unambiguously has bene elohim. A second scroll, 4QDeut^q (4Q44), is broken at this word; P. W. Skehan, who first published it in 1954, judged from the spacing that its line was too short for "sons of Israel" and reconstructed a divine-beings reading — bene el, "sons of El." (Heiser is careful to keep the two scrolls distinct: it is 4QDeut^j, not the fragmentary 4QDeut^q, that carries the clear reading.)
• "Angels of God" (ἀγγέλων θεοῦ) — the majority Greek of the Septuagint. Text critics read this not as a third Hebrew reading but as the Greek rendering of a Hebrew "sons of God" (the Septuagint elsewhere turns "sons of God" into "angels"). A minority of Greek witnesses — including Papyrus Fouad 266 (Rahlfs 848) — keep the more literal "sons of God" (υἱῶν θεοῦ), and Wevers's Göttingen critical edition prints that as the Old Greek.

So the genuine contest is two-way and Hebrew: "sons of God / El" (Qumran + the Greek Vorlage) against "sons of Israel" (MT + Samaritan). "Angels of God" is a downstream Greek softening of the first, not evidence for a distinct third original.²

The evidence for "sons of God" being original

Transcriptional probability runs one way. The governing rule of textual criticism is not to count manuscripts but to weigh them — to ask which reading best explains the rise of the others. A scribe with a live horror of anything polytheistic-sounding has an obvious motive to change an explosive "sons of God" into a safe "sons of Israel"; no scribe has a motive to manufacture "sons of God" out of "sons of Israel." The harder, more offensive reading is therefore the more probably original, and it accounts for its rival; the reverse does not.¹

The parallel expurgation eleven verses later is undeniable. This is the strongest single leg. At Deuteronomy 32:43 the Masoretic Text is demonstrably a shortened text: it lacks a whole colon — "bow down to him, all you gods" — that both a Qumran scroll (4QDeut^q) and the Septuagint preserve, a line naming other divine beings. Once it is granted that MT quietly dropped the "gods" from the end of the Song of Moses, the same hand tidying "sons of God" to "sons of Israel" near its opening is not a conjecture but a pattern. As Jeffrey Tigay puts it, the MT of both verses was altered for one reason: to remove the divine beings.³

The sense of the verse fits "sons of God," and "sons of Israel" strains it. Verse 8 divides the nations, and verse 9 keeps Israel back as the Lord's own portion — so Israel is set apart from the count, not used as its measure. The seventy nations of the Table of Nations (Genesis 10) map onto the seventy "sons of God," a number with a precise pagan-cosmological analogue: at Ugarit the high god El — himself titled "Most High" — fathers seventy sons. To make "sons of Israel" the measuring-number instead, one must read it as the seventy souls of Jacob's household who went down to Egypt (Genesis 46:27; Exodus 1:5) — but that entity, as Heiser presses, "did not yet exist" when the nations were apportioned, so the correlation is anachronistic and looks like a later back-formation. The divine-council reading needs no such maneuver.³

Adversarial assessment

What holds. As a text-critical judgment the reading is strong and mainstream. It has a clean primary witness (4QDeut^j) and an early Greek one (Papyrus Fouad 266); it satisfies lectio difficilior and the "best-explains-the-rival" test; it is corroborated by the transparent 32:43 minus; and it coheres with the Genesis 10 / Ugaritic seventy-fold logic. This is why the weight of specialist opinion moved to it — Emanuel Tov judges MT a theological correction, Wevers's Göttingen Septuagint prints "sons of God," and the NRSV puts "gods" in its text while the ESV footnotes the variant. None of this rests on ruling the supernatural in or out; it is ordinary comparative and transcriptional argument.³

What the Masoretic defenders still have. The minority case is not frivolous, and honesty requires stating it. First, the numbers still favor MT — the preponderance of manuscripts, plus the Samaritan Pentateuch as a second independent Hebrew witness, read "sons of Israel." Second, the most serious text-critical defense (David E. Stevens, in the same journal) offers a purely accidental mechanism: a scribe's eye slipped and dropped the letters of "Israel," leaving a bare "sons of El" that later filled out. Third, a stream of confessional writers (Heath Henning's book-length rebuttal; the "Reformed Masora" and OTTC treatments) argue the shorter Greek/Qumran reading is itself the innovation — a Hellenistic or Essene angelology — and lean on lectio brevior and Masoretic conservatism the other way. The strongest reply to Stevens is decisive but narrow: accidental loss can yield a bare bene el, but it cannot produce the extra consonants of 4QDeut^j's bene elohim, which would take a second, deliberate insertion — possible, but it "stretches credulity." That is why the verdict lands on "sound," not merely "plausible": the rival survives as a real minority position, but it does not have the better of the argument.⁴

The text is the base, not the whole building. The one guard the reading's enthusiasts need is this: winning the variant does not win the cosmology. Granting "sons of God" as original settles that the verse names divine beings among whom the nations were parcelled; it does not, by itself, establish that those beings are rebellious territorial elohim who were judged in Psalm 82, still less that they are the same order as the Genesis-6 "sons of God." Each of those is a further step with its own evidence and its own rating — the identity of Psalm 82's elohim is separately weighed and rated contested, and the identification with the Genesis-6 beings is weighed on the Nephilim timeline's pagan-gods argument (also contested). The text-critical foundation is firmer than the theology raised on it; that is the honest shape of the case.⁴

Assessment

Sound — as the text-critical claim that "sons of God / El" is the earlier Hebrew reading of Deuteronomy 32:8 and the Masoretic "sons of Israel" a later theological correction.

On the manuscript evidence and the standard canons of the discipline this is the reading to prefer, and it is the majority view in critical scholarship: a clean Qumran witness, an early Greek one, the harder-reading and best-explains-the-rival logic, and the undeniable parallel expurgation at 32:43 converge on it, and no rival line defeats that convergence. The Masoretic defense is a real minority position — it keeps the numbers and the Samaritan Pentateuch, and Stevens's accidental-loss mechanism is a genuine argument — but its best form cannot account for the full consonants of 4QDeut^j, so it does not overturn the verdict. Two precisions keep the rating honest. First, the exact earliest Hebrew form is itself open — whether the substratum was "sons of El," "sons of gods," or "sons of God" is debated within the divine-beings camp, and none of those options is "sons of Israel," so that internal question does not rescue MT. Second, this page rates only the earlier text: it is the load-bearing base of the divine-council reading, not its proof. The sub-claim to keep is "the original Deuteronomy 32:8 named divine beings, not the sons of Israel"; the step to weigh elsewhere is what those beings are and what cosmology they carry.

Sources