Septuagint versus Hebrew readings, with manuscript evidence

Comparisons I, essay 2. The Greek is quoted from the public-domain Swete edition of the Septuagint, as printed in the library’s LXX reader; the license-restricted CATSS data of another project is not used anywhere here. Manuscript witnesses are named by siglum in every case.

Most English Bibles follow the Masoretic Text, the Hebrew text fixed by medieval Jewish scribes and printed in modern editions. Most of the early Church, and the New Testament writers, quoted something else: the Septuagint, the Greek translation made in Alexandria centuries earlier. For a long time the difference between the two was described as if one were the original and the other a mistranslation. The Dead Sea Scrolls ended that picture. At Qumran, Hebrew manuscripts were found that agree sometimes with the Masoretic Text, sometimes with the Hebrew text behind the Septuagint, and sometimes with neither. The Hebrew Bible, in the centuries before 70 CE, existed in more than one edition. “Septuagint versus Hebrew” is therefore the wrong title for what follows, kept only because it is the title readers recognise. The real subject is three witnesses — Masoretic Hebrew, Septuagint Greek, and the scrolls — and what each passage shows about how they relate.

How to read the cases. Each case gives the Masoretic reading, the Septuagint reading in Greek with a translation, and the manuscript evidence, in that order. “The Septuagint agrees with the scrolls against the Masoretic Text” does not mean the Septuagint reading is certainly original; it means the Greek translator did not invent it — a Hebrew text like it existed. Where the scroll evidence is fragmentary, mixed, or limited, the case says so. Nothing in this essay is argued from a manuscript the author has not named.

1. Deuteronomy 32:8 — whose sons set the boundaries of the nations?

The Song of Moses describes the Most High dividing humanity into nations. The Masoretic Text ends the verse:

לְמִסְפַּר בְּנֵי יִשְׂרָאֵל — “according to the number of the sons of Israel.”

The Septuagint ends it differently:

“ὅτε διεμέριζεν ὁ ὕψιστος ἔθνη, ὡς διέσπειρεν υἱοὺς Ἀδάμ, ἔστησεν ὅρια ἐθνῶν κατὰ ἀριθμὸν ἀγγέλων θεοῦ.”

“When the Most High divided the nations, when he scattered the sons of Adam, he set the boundaries of the nations according to the number of the angels of God.”

Here the scrolls settle the question of whether the Greek translator invented his reading. The Qumran Deuteronomy manuscript 4QDeut-j (4Q37) reads, at this verse, “according to the number of the sons of God” (bene elohim). The Septuagint’s “angels of God” is the Greek rendering of that Hebrew phrase — the Septuagint regularly renders “sons of God” as “angels”, as in Job — not a free paraphrase. The Masoretic “sons of Israel” is the outlier: a reading that makes sense only after Israel exists, in a verse about the primeval division of humanity, and that removes the divine council from the picture. The scholarly consensus, reflected in modern translations’ footnotes, is that “sons of God” is the older reading and the Masoretic Text has been revised here, presumably for theological reasons.

Justin Martyr already knew both readings, in the second century. In chapter 131 of the Dialogue with Trypho, he quotes the verse in the Masoretic form (“according to the numbers of the children of Israel”) and then notes, fairly: “The Seventy have translated it, ‘He set the bounds of the nations according to the number of the angels of God.’” He adopts the Hebrew exposition because his argument does not need the Greek one. The variant, in other words, was not discovered at Qumran; Qumran confirmed what readers of both texts had known for two thousand years, and added the Hebrew witness that proves the Greek was translating, not inventing.

2. Deuteronomy 32:43 — the longer song, and a New Testament quotation

The same chapter gives a second case, at its end. The Masoretic Text of verse 43 is short:

“Rejoice, O nations, with his people: for he will avenge the blood of his servants, and will render vengeance to his adversaries, and will be merciful unto his land, and to his people.”

The Septuagint is nearly twice as long:

“εὐφράνθητε, οὐρανοὶ, ἅμα αὐτῷ, καὶ προσκυνησάτωσαν αὐτῷ υἱοὶ θεοῦ· εὐφράνθητε, ἔθνη, μετὰ τοῦ λαοῦ αὐτοῦ, καὶ ἐνισχυσάτωσαν αὐτῷ πάντες ἄγγελοι θεοῦ· ὅτι τὸ αἷμα τῶν υἱῶν αὐτοῦ ἐκδικᾶται, καὶ ἐκδικήσει, καὶ ἀνταποδώσει δίκην τοῖς ἐχθροῖς· καὶ τοῖς μισοῦσιν ἀνταποδώσει, καὶ ἐκκαθαριεῖ Κύριος τὴν γῆν τοῦ λαοῦ αὐτοῦ.”

“Rejoice, heavens, with him, and let the sons of God worship him; rejoice, nations, with his people, and let all the angels of God strengthen themselves in him; for he will avenge the blood of his sons, and will render vengeance, and recompense justice to his enemies, and will reward them that hate him; and the Lord shall purge the land of his people.”

The Qumran manuscript 4QDeut-q (4Q44) preserves the longer form, agreeing with the Septuagint against the Masoretic Text — including the call for the angels to worship, which is the line Hebrews 1:6 quotes (“let all the angels of God worship him”). That quotation has puzzled readers who check it against the Masoretic Text, where the line does not exist; it is a straightforward quotation of the longer Hebrew edition behind the Septuagint. A necessary qualification: Qumran does not speak with one voice here. Another Qumran copy of Deuteronomy is shorter at this verse, closer to the Masoretic form. Both editions were therefore in circulation in the same community. The conclusion is not “the scrolls prove the long text original” but the more modest and more interesting one: two editions of the Song circulated side by side, and the New Testament writer used the longer one.

3. Jeremiah — not a verse variant but two editions of a book

Jeremiah is the case that breaks the verse-by-verse habit altogether. The Septuagint Jeremiah is about one-eighth shorter than the Masoretic Jeremiah, and after chapter 25 the order of chapters diverges completely: the oracles against the nations, which the Masoretic Text places at the end of the book (chapters 46–51), stand in the middle of the Septuagint, immediately after 25:13. A translator does not do that by accident, and for a century scholars argued whether the Greek translator had abridged and rearranged a text like the Masoretic one.

Qumran answered that question. Two Jeremiah manuscripts from Cave 4, 4QJer-b (4Q72) and 4QJer-d (4Q73), display the shorter text and the Septuagint’s arrangement: they are the Hebrew edition behind the Greek translation. Two others, 4QJer-a (4Q70) and 4QJer-c, follow the longer, Masoretic arrangement. Both editions therefore existed in Hebrew, in the same caves, before either became canonical. The Septuagint translator was not shortening Jeremiah; he was translating a shorter Jeremiah. Which edition is earlier, passage by passage, remains a matter of detailed argument — the shorter edition is generally thought to represent an earlier stage of the book’s growth, but each difference has to be judged on its own — and this essay does not pretend otherwise. What the manuscripts establish is the framework: Jeremiah circulated in two Hebrew editions, and the Greek and Hebrew Bibles of later centuries each standardised a different one. Any account of “what Jeremiah says” that quotes only one edition is quoting half the evidence.

4. 1 Samuel 17:4 — how tall was Goliath?

The most famous single-number variant in the Bible is also one of the best attested. The Masoretic Text gives Goliath’s height as:

גָּבְהוֹ שֵׁשׁ אַמּוֹת וָזָרֶת — “his height was six cubits and a span” (about nine feet nine inches on the usual cubit).

The Septuagint gives:

“καὶ ἐξῆλθεν ἀνὴρ δυνατὸς ἐκ τῆς παρατάξεως τῶν ἀλλοφύλων, Γολιὰθ ὄνομα αὐτῷ, ἐκ Γέθ· ὕψος αὐτοῦ τεσσάρων πήχεων καὶ σπιθαμῆς.”

“And there came out a strong man from the ranks of the Philistines, Goliath by name, from Gath; his height was four cubits and a span” (about six feet nine inches).

The Qumran Samuel manuscript 4QSam-a (4Q51) agrees with the Septuagint: four cubits and a span. So does Josephus, in Antiquities 6.9.1, whose height for Goliath follows the Greek figure. The agreement of a Hebrew scroll, the Greek translation, and Josephus against the Masoretic Text is the strongest pattern this essay can show, and modern translations accordingly footnote the verse (“Septuagint, Dead Sea Scrolls and Josephus: four cubits”). Whether four or six cubits is “original” is a separate question from which is taller: a giant’s height grows easily in retelling, and the shorter figure may itself be a rationalisation. The manuscript evidence establishes that both figures are ancient, not that either is the author’s. Goliath remains formidable on either reading — six feet nine inches was very tall in the ancient world — which is perhaps why the variant matters less to the story than to the history of the text.

5. 1 Samuel 10:27–11:1 — a paragraph in neither Bible

The converse case is just as instructive: a reading preserved in a scroll and in Josephus that is in neither the Masoretic Text nor the Septuagint. In the standard text, chapter 10 ends with Saul going home while “worthless fellows despised him and held their peace”, and chapter 11 opens, without transition, with Nahash the Ammonite besieging Jabesh-gilead. The jump is abrupt, and Nahash’s opening demand — surrender on condition that he gouge out every man’s right eye — arrives unexplained.

The Qumran manuscript 4QSam-a supplies the missing paragraph: Nahash had already been oppressing the Gadites and Reubenites across the Jordan, gouging out the right eye of every man he captured and leaving Israel without a deliverer, until seven thousand men escaped to Jabesh-gilead; “about a month later” he besieged the city. Josephus tells the same story, in the same place (Antiquities 6.5.1). The likeliest explanation is a scribe’s eye skipping between similar phrases (homoioteleuton) in the ancestor of both the Masoretic Text and the Septuagint, dropping a paragraph that the Qumran copy and Josephus’s source retained. The evidence, however, is limited — essentially one Hebrew manuscript and Josephus — and translations differ accordingly: the NRSV prints the paragraph (as 10:27b), while the NET Bible follows the Masoretic Text and notes the longer reading with caution. This essay follows the NET’s caution in wording: the paragraph is probably original, but “probably” is doing real work. A single scroll, however good, is not a consensus.

6. Genesis 47:31 — bed or staff? A case with no scroll verdict

The last case is included because it needs no scroll at all, and shows the other way a Septuagint reading can differ: not a different text, but a different way of reading the same consonants. Jacob, dying, makes Joseph swear to bury him in Canaan, “and Israel bowed himself” — and here the versions part. The Masoretic vocalisation reads the final word as mittah, “bed”:

עַל־רֹאשׁ הַמִּטָּה — “on the head of the bed.”

The Septuagint reads the same consonants as matteh, “staff”:

“εἶπεν δέ Ὄμοσόν μοι. καὶ ὤμοσεν αὐτῷ· καὶ προσεκύνησεν Ἰσραὴλ ἐπὶ τὸ ἄκρον τῆς ῥάβδου αὐτοῦ.”

“And he said, Swear to me. And he swore to him. And Israel worshipped on the top of his staff.”

Hebrews 11:21 follows the Septuagint: Jacob “worshipped, leaning upon the top of his staff.” No Dead Sea Scroll verdict is offered here, because none is needed and none is claimed: the consonantal Hebrew supports both vocalisations, and the choice between them is a choice between reading traditions, not between manuscripts. The case is a reminder that not every Septuagint difference is a textual variant. Some are the Greek translators hearing the same Hebrew words differently — and the New Testament, written in Greek, hearing them the Greek way.

Conclusion: what manuscript evidence can and cannot do

Three conclusions follow from the six cases, and they are deliberately modest. First, the Septuagint is a witness to a Hebrew text, not merely a translation of the Masoretic Text: wherever it agrees with a Qumran manuscript against the Masoretic Text — Deuteronomy 32:8, 32:43, 1 Samuel 17:4 — the agreement proves a Hebrew Vorlage behind the Greek. Second, the Masoretic Text is not thereby demoted to a corruption: it is one edition among editions, and in Jeremiah the longer edition it preserves may itself contain later expansions, just as the shorter edition may contain losses; each passage is argued, not assumed. Third, the evidence has edges: a single scroll (Nahash), a mixed Qumran picture (Deuteronomy 32:43), and cases where the difference is vocalisation rather than text (Genesis 47:31) all resist a single slogan. The Dead Sea Scrolls did not give scholars a master key. They gave them, instead, the ordinary and difficult situation of ancient books: multiple editions, overlapping witnesses, and judgements that have to be made passage by passage, with the witnesses named.

Bibliography

Emanuel Tov, Textual Criticism of the Hebrew Bible, 3rd edition (Minneapolis: Fortress Press, 2012) — print; the standard handbook, with full discussion of 4QDeut-j, 4QSam-a, and the Jeremiah manuscripts.
Eugene Ulrich, The Dead Sea Scrolls and the Origins of the Bible (Grand Rapids: Eerdmans, 1999) — print; the argument that the scrolls reveal a pluriform Hebrew text before standardisation.
Ernst Würthwein, The Text of the Old Testament, 2nd edition (Grand Rapids: Eerdmans, 1995) — print; concise introduction to the Masoretic Text, Septuagint, and scrolls, with the Deuteronomy 32 and Samuel cases.
H. B. Swete, An Introduction to the Old Testament in Greek (Cambridge: Cambridge University Press, 1914) — print and open access (public domain); the history of the Septuagint text behind the edition quoted here.
H. B. Swete (ed.), The Old Testament in Greek according to the Septuagint (Cambridge, 1887–1894) — open access (public domain); the edition from which the Greek quotations in this essay are taken, as printed in the library’s LXX reader.
Leon Levy Dead Sea Scrolls Digital Library, Israel Antiquities Authority — open access: photographs and descriptions of the Qumran manuscripts named here (4Q37, 4Q44, 4Q51, 4Q70, 4Q72, 4Q73), for checking any siglum in this essay against the artefact itself.

Feeds the Day 19 bibliography: the print items above are carried into the scholar’s resources bibliography, with the open-access editions linked from their work pages.