BIBLIOTHECA SACRA 158 (January-March 2001): 52-74

[Copyright © 2001 Dallas Theological Seminary; cited with permission;

digitally prepared for use at Gordon College] 

 

 

 

                DEUTERONOMY 32:8 AND

                        THE SONS OF GOD

 

                                                Michael S. Heiser

 

MOSES' FAREWELL SONG IN DEUTERONOMY 32:1-43 is one of

the more intriguing portions of Deuteronomy and has re-

ceived much attention from scholars, primarily for its po-

etic features, archaic orthography and morphology, and text-

critical problems.1 Among the textual variants in the Song of

Moses, one in verse 8 stands out as particularly fascinating. The

New American Standard Bible renders the verse this way: "When

the Most High gave the nations their inheritance, when He sepa-

rated the sons of man, He set the boundaries of the peoples ac-

cording to the number of the sons of Israel."

The last phrase, "according to the number of the sons of Is-

rael," reflects the reading of the Masoretic text lxerAW;yi yneB;, a reading

also reflected in some later revisions of the Septuagint: a manu-

script of Aquila (Codex X), Symmachus (also Codex X), and

Theodotion.2 Most witnesses to the Septuagint in verse 8, however,

read, a@ggelw?n qeou? ("angels of God"), which is interpretive,3 and

 

Michael S. Heiser is a Ph.D. candidate in Hebrew and Semitic Studies at the Uni-

versity of Wisconsin-Madison.

1 For a recent overview of the scholarship on the Song of Moses, see Paul Sand-

ers's thorough treatment in The Provenance of Deuteronomy 32 (Leiden: Brill, 1996).

See also Frank M. Cross and David Noel Freedman, Studies in Ancient Yahwistic

Poetry (Grand Rapids: Eerdmans, 1997); William F. Albright, "Some Remarks on

the Song of Moses in Deuteronomy XXXII," Vetus Testamentum 9 (1959): 339-46;

and D. A. Robertson, Linguistic Evidence in Dating Early Hebrew Poetry (Missoula,

MT: Scholars, 1972). .

2 Fridericus Field, ed., Origenis Hexaplorum, Tomus I: Prolegomena, Genesis-

Esther (Hildesheim: Georg Olms, 1964), 320, n. 12.

3 This is the predominant reading in the Septuagint manuscripts and is nearly

unanimous. See John William Wevers, ed., Septuaginta: Vetus Testamentum

Graecum, Auctoritate Academiae Scientiarum Gottingensis Editum, vol. 3.2: Deuter-

onomium (Gottingen: Vandenhoeck & Ruprecht, 1977), 347; and idem, Notes on the

Greek Text of Deuteronomy (Atlanta: Scholars, 1995), 513. Wevers refers to this

majority reading as "clearly a later attempt to avoid any notion of lesser deities in

favor of God's messengers" (ibid.).



Deuteronomy 32:8 and the Sons of God                            53

 

several others read ui[w?n qeou? ("sons of God").4 Both of these Greek

renderings presuppose a Hebrew text of either Myhlx ynb or Mylx ynb.

These Hebrew phrases underlying a@ggelw?n qeou?  and ui[w?n qeou? are

attested in two Hebrew manuscripts from Qumran,5 and by one

(conflated) manuscript of Aquila.6

Should the verse be rendered "sons of Israel" or "sons of God"?

The debate over which is preferable is more than a fraternal spat

among textual critics. The notion that the nations of the world

were geographically partitioned and owe their terrestrial identity

to the sovereign God takes the reader back to the Table of Nations

in Genesis 10-11. Two details there regarding God's apportionment

of the earth are important for understanding Deuteronomy 32:8.

 First, the Table of Nations catalogs seventy nations, but Israel is

not included.7 Second, the use of the same Hebrew root (draPA) in

both Genesis 10 and Deuteronomy 32 to describe the "separation"

of the human race and the nations substantiates the long-

recognized observation that Genesis 10-11 is the backdrop to the

statement in Deuteronomy 32:8.8 Because Israel alone is Yahweh's

portion, she was not numbered among the seventy other nations.

The reference to seventy "sons of Israel" (in the Masoretic

text), initially seemed understandable enough, for both Genesis

46:27 and Exodus 1:5 state that seventy members of Jacob's family.

 

4 Wevers, ed., Septuaginta, 347. The Gottingen Septuagint has adopted ui[w?n qeou?

as the best reading, despite its having fewer attestations.

5 The words lx ynb are not an option for what was behind the Septuagint reading,

as demonstrated by the Qumran support for the Hebrew text underlying the unre-

vised Septuagint. First, manuscript 4QDtq has spaces for additional letters follow-

ing the l of its [ ] lx ynb. Second, 4QDtJ clearly reads Myhvlx ynb (Sanders, The Prove-

nance of Deuteronomy 32, 156). See also Emanuel Tov, Textual Criticism of the He-

brew Bible (Minneapolis: Fortress, 1992), 269.

6 Wevers, ed., Septuaginta, 347; and Field, Origenis Hexaplorum, Tomus I: Prole-

gomena, Genesis-Esther, 320. The manuscript of Aquila is Codex 85.

7 As Allen P. Ross notes, "On investigation the reader is struck by a deliberate

pattern in the selection of names for the Table. For example, of the sons of Japheth,

who number seven, two are selected for further listing. From those two sons come

seven grandsons, completing a selective list of fourteen names under Japheth. With

Ham's thirty descendants and Shem's twenty-six, the grand total is seventy"

("Studies in the Book of Genesis; Part 2: The Table of Nations in Genesis 10--Its

Structure," Bibliotheca Sacra 137 [October-December 1980]: 342). Some scholars,

Ross observes, arrive at the number of seventy-one for the names, depending on how

the counting is done (ibid., 352, n. 18). Ross and Cassuto agree that the accurate

count is seventy (cf. Umberto Cassuto, A Commentary on the Book of Genesis: From

Noah to Abraham [Jerusalem: Magnes, 1964],177-80).

8 Cassuto, A Commentary on the Book of Genesis, 174-78; Albright, "Some Re-

marks on the Song of Moses in Deuteronomy XXXII," 343-44. A Niphal form of drp

is used in Genesis 10:5 (Udr;p;ni), and the Hiphil occurs in Deuteronomy 32:8 (Odyrip;haB;).



54        BIBLIOTHECA SACRA / January-March 2001

 

went to Egypt in the days of Joseph.9 Little thought was given,

however, to the logic of the correlation: How is it that the number

of the pagan nations was determined in relation to an entity (Is-

rael) or individuals (Jacob and his household) that did not yet ex-

ist? Even if one contends that the correlation was in the mind of

God before Israel's existence and only recorded much later, what

possible point would there be behind connecting the pagan Gentile

nations numerically with the Israelites? On the other hand what

could possibly be meant by the notion that a correspondence ex-

isted between the number of the nations in Genesis 10-11 and

heavenly beings?

Literary and conceptual parallels discovered in the literature

of Ugarit, however, have provided a more coherent explanation for

the number seventy in Deuteronomy 32:8 and have furnished sup-

port for textual scholars who argue against the "sons of Israel"

reading. Ugaritic mythology plainly states that the head of its pan-

theon, El (who, like the God of the Bible, is also referred to as El

Elyon, the "Most High") fathered seventy sons,10 thereby specifying

the number of the "sons of El" (Ugaritic, bn il). An unmistakable

linguistic parallel with the Hebrew text underlying the Septuagint

reading was thus discovered, one that prompted many scholars to

accept the Septuagintal reading on logical and philological

grounds--God (El Elyon in Deut. 32:8) divided the earth according

to the number of heavenly beings who existed from before the time

of creation.11 The coherence of this explanation notwithstanding,

some commentators resist the reading of the Septuagint, at least in

part because they fear that an acceptance of the Myhlx ynb or Mylx ynb

readings (both of which may be translated "sons of gods") somehow

 

9 There is a textual debate on this passage in Exodus as well. Although space

prohibits a thorough discussion of Genesis 46:27 and Exodus 1:5, they do provide

examples, in conjunction with Deuteronomy 32:8, of the primary guiding principle

in textual criticism: The reading that best explains the rise of the others is most

likely the original. In the case of Genesis 46:27 and Exodus 1:5, the Septuagint and

Qumran literature disagree with the Masoretic text together when they read that

seventy-five people went to Egypt with Jacob. The number seventy-five incorporates

five additional descendants from Ephraim and Manasseh. This example from these

verses features the same textual alignment as with Deuteronomy 32:8 (the Septua-

gint and Qumran agree together against the Masoretic text), but in Exodus 1:5 the

Masoretic reading is to be preferred. The point is that one cannot be biased in favor

of either the Masoretic or the Septuagintal readings; instead, the reading that best

explains the rise of the others is the preferred reading, regardless of the text-type.

10 Manfried Dietrich, Oswald Loretz, and Joaquin Sanmartin, eds., The Cuneiform

Alphabetic Texts from Ugarit, Ras Ibn Hani and Other Places, KTU, 2d ed. (Mun-

ster: Ugarit, 1995), 18. The reading in the article is from KTU 1.4:VI.46.

11 Job 38:7 states that the heavenly host was present at creation.



Deuteronomy 32:8 and the Sons of God                            55

 

means that Yahweh is the author of polytheism. This apprehension

has prompted some text-critical defenses of the Masoretic text in

Deuteronomy 32:812 based on a misunderstanding of both the tex-

tual history of the Hebrew Bible and text-critical methodology, a

prejudiced evaluation of non-Masoretic texts, and an unfounded

concern that departure from, the Masoretic reading results in "Isra-

elite polytheism." The goal of this article is to show that viewing

"sons of God" as the correct reading in Deuteronomy 32:8 in no way

requires one to view Israelite religion as polytheistic.

 

TEXTUAL CRITICISM AND THE "SONS OF GOD"

IN DEUTERONOMY 32:8

 

A WORD ABOUT TEXT-CRITICAL METHOD AND PREJUDICES

The textual evidence cited above presents a situation in which one

reading (that of the Septuagint) is supported by very ancient

manuscript evidence (notably Qumran), while the other (the Ma-

soretic reading) has a preponderance of the support, thereby cre-

ating an "oldest-versus-most" predicament. As in similar New Tes-

tament cases the correct reading can be verified not by counting

manuscripts but by weighing them. Hence it matters little that the

Septuagint reading is "outnumbered," especially since the more

numerous sources are much later, and in fact are interdependent,

not independent, witnesses. When considering the evidence, it is

wrong to assume that the Masoretic text is superior at every point

to other texts of the Old Testament. It is equally fallacious to pre-

suppose the priority of the Septuagint. Simply stated, no text

should automatically be assumed superior in a text-critical investi-

gation. Determining the best reading must be based on internal

considerations, not uncritical, external presumptions about the

"correct" text.

Unfortunately the notion of the presumed sanctity of the Ma-

soretic text still persists. The dictum that the Masoretic text is to

be preferred over all other traditions whenever it cannot be faulted

linguistically or for its content (unless in isolated cases there is

good reason for favoring another tradition) is all too enthusiasti-

cally echoed.13 The idea seems to be that whenever a Masoretic

 

12 For example David E. Stevens, "Does Deuteronomy 32:8 Refer to 'Sons of God' or

'Sons of Israel'?" Bibliotheca Sacra 154 (April-June 1997): 139. However, since

writing his article Stevens has repudiated this view and has accepted the reading

"sons of God" (David E. Stevens, "Daniel 10 and the Notion of Territorial Spirits,"

Bibliotheca Sacra 157 (October-December 2000): 412, n. 9.

13 Ernst Wurthwein, The Text of the Old Testament, trans. Peter R. Ackroyd (New



56                    BIBLIOTHECA SACRA / January-March 2001

 

reading could be accepted it should be accepted. Such an approach,

however, hardly does justice to non-Masoretic readings that also

could be acceptable on their own linguistic and contextual terms.

Put another way, the above view seldom addresses why the Ma-

soretic text should be held in such esteem. Where there are wide

and significant textual divergencies between the Masoretic text

and the Septuagint, many textual studies have shown that the

Qumran witnesses demonstrate the reliability of the transmission

of the Hebrew text underlying the Septuagint. For example it is

well known that the Masoretic text of 1 and 2 Samuel is in poor

condition in a number of places and includes instances of signifi-

cant haplography.14 First and 2 Kings are riddled with both short

and lengthy pluses and minuses, transpositions, and chronological

differences.15 Also portions of the Masoretic text of Ezekiel, espe-

cially chapters 1 and 10, could serve as a veritable digest of textual

corruptions.16

Judging by the survival in Old Testament textual criticism of a

"textus receptus" approach like the one that once held sway in New

Testament textual criticism, more consideration is needed as to

how the Masoretic text came to be considered the "received text."

Just because the Masoretic text was the received text of the medie-

val Masoretes does not mean that it merits textual priority among

today's extant witnesses, or even that it had textual priority in

biblical times. The Masoretic text rose to prominence only after

centuries of textual diversity and not, as noted above, by "intrinsic

factors related to the textual transmission, but by political and so-

cioreligious events and developments."17

The evidence from Qumran unquestionably testifies to a certi-

 

York: Macmillan, 1957), 76-82.

14 P. Kyle McCarter, I Samuel (New York: Doubleday, 1980); and idem, Textual

Criticism: Recovering the Text of the Hebrew Bible (Philadelphia: Fortress, 1986), 38.

15 Tov, Textual Criticism of the Hebrew Bible, 142.

16 Daniel Block, "Text and Emotion: A Study in the 'Corruptions' in Ezekiel's Inau-

gural Vision (Ezekiel 1:4-28)," Catholic Biblical Quarterly 50 (July 1988): 418-42.

17 Emanuel Tov, "Textual Criticism (OT)," in Anchor Bible Dictionary, ed. D. N.

Freedman (New York: Doubleday, 1992),6:395,407. Tov summarizes the historical

situation as follows: "By the end of the 1st century A.D. the Septuagint had been

accepted by Christianity and abandoned by Jews. Copies of the Samaritan Penta-

teuch were available, but in the meantime that sect had become an independent

religion, so that their texts were considered Samaritan, not Jewish any more. The

Qumran sect, which had preserved a multitude of texts, did not exist after the de-

struction of the temple. Therefore the sole texts that existed in this period were the

ones that were copied and distributed by the central group in Judaism. . . .This

situation gave rise to the wrong conclusion that the MT had 'ousted' the other texts"

(ibid., 407).



Deuteronomy 32:8 and the Sons of God                            57

 

fiable textual plurality among Jews in Palestine for the period be-

tween the third century B.C. and the first century A.D.18 Precursory

forms of the Masoretic text, the Septuagint, and the Samaritan

Pentateuch existed and are attested in the Dead Sea Scrolls. As

further proof of textual diversity the Qumran material also con-

tains "independent" or "unaligned" texts, which exhibit both

agreement and disagreement with the textual traditions of the Ma-

soretic text, the Septuagint, and the Samaritan Pentateuch.19 The

Qumran fragments that support the Septuagintal "sons of God"

reading, 4QDeutj,n, are among the unaligned texts.20

Two points derive from this review of the textual plurality in

the Dead Sea Scrolls. First, no evidence exists in the actual textual

data that the Jews held a negative view of Hebrew texts not

grouped among those that later received the appellation "Ma-

soretic." Second, the undeniable textual diversity at Qumran ar-

gues against any suggestion that the Qumranites altered a text

ultimately used by the Septuagintal translators as their Vorlage.

Besides the chronological and logistical difficulties of such an idea,

this question remains: If the Qumran members were in the habit of

altering texts to reflect allegedly strange angelic views or Gnostic

tendencies, why did they leave so many texts within each of the

major textual strains unaltered? Stated another way, why did the

Qumran inhabitants allow so many passages of the Hebrew Bible

that point to God's uniqueness, omnipotence, and sovereignty to

stay in the texts they deposited in the nearby caves? It hardly

makes sense to sneak one alteration into Deuteronomy 32:8 while

letting hundreds of other "nondualistic" texts remain.

 

EVALUATING THE INTERNAL TEXT-CRITICAL EVIDENCE

FOR DEUTERONOMY 32:8

Those who assume the priority of the Masoretic text might offer

two explanations as to why Deuteronomy 32:8 reads "sons of God"

in some manuscripts, including the Qumran material. One option

is that this reading should simply be regarded as an intentional

error reflecting the theological predilections of Qumran and the

Septuagintal translators. However, this theory has already been

called into question. The other explanation suggests that the vari-

ant arose unintentionally; that is, the consonants rWy were acciden-

 

18 Tov, Textual Criticism of the Hebrew Bible, 116-17. See also S. Talmon, "The

Old Testament Text," in Cambridge History of the Bible, ed. Peter R. Ackroyd and C.

F. Evans (Cambridge: Cambridge University Press, 1963); 1:159-99.

19 Tov, "Textual Criticism (OT)," 395, 402, 404, 406.

20 Ibid., 402.



58                    BIBLIOTHECA SACRA / January-March 2001

 

tally omitted (by parablepsis) from the word lxrWy leaving lx ynb in

the text in the place of lxrWy ynb.

This second explanation is less than satisfactory for at least

two reasons. First, one could just as well argue that rWy was added

to the text. This is hardly a satisfying response, however, for it is

as much of a speculation as the competing proposition. The real

problem with the parablepsis proposal is that, while it accounts for

the consonants lx in the text, it fails to explain adequately how the

consonants Myhv would have come to be added after lx to the text

underlying the Septuagint reading. It is particularly significant in

this regard that the texts from Qumran that support the Septua-

gint do not read the consonants lx ynb as this explanation would

postulate, for in one text, 4QDeutq, there are spaces for additional

consonants after the l of the word lx. The other Dead Sea text that

supports the Septuagintal reading, 4QDeutj, unambiguously reads

Myhlx ynb.21

Second, and perhaps even more damaging to the proposed

parablepsis explanation that an original "sons of Israel" was unin-

tentionally corrupted to "sons of God" in Deuteronomy 32:8, is that

there exists another text-critical problem in Deuteronomy 32 in

which heavenly beings--"sons of Myhlx / Mylx"--are the focus (v.

43a)! Deuteronomy 32:43 reads differently in the Masoretic text,

the Septuagint, and a Qumran text.

The Masoretic text has one line:

"O nations, rejoice His people."

 

4QDeutq has a bicolon:

"O heavens, rejoice with Him

Bow to Him, all divinities."

 

And the Septuagint has two bicola:

"O heavens, rejoice with Him

Bow to Him, all sons of the divine.22

O nations, rejoice with His people

And let all angels of the divine strengthen themselves in

Him."23

 

21 Tov, Textual Criticism of the Hebrew Bible, 269. Also see note 5 in this article.

22 The translation of the Septuagint provided by Tigay could reflect Mylx instead of

Myhlx since "divine" rather than "God" is chosen as the translation (Jeffrey H. Tigay,

Deuteronomy, JPS Torah Commentary [Philadelphia: Jewish Publication Society,

1996],516).

23 The translations are from Tigay, Deuteronomy, 516.



Deuteronomy 32:8 and the Sons of God                59

 

            It is significant that the Masoretic text lacks a second line in

what should be the first pairing. Even more striking is the fact that

this missing colon is the one in which reference is made to divie

beings in the Qumran and Septuagintal texts. In these latter two

texts each colon has its partner. This argues strongly that the Ma-

soretic text originally had a bicolon, a pairing that was deliberately

eliminated to avoid the reference to other "divine beings."24 While

the other Masoretic omissions can be explained by haplography,

the absence of the line that would have made reference to heavenly

beings cannot be so explained.25

            What does this imply? It suggests, for one thing, that those

who defend the priority of the Masoretic text would have to argue

for accidental changes in Deuteronomy 32:8 (the missing rWy) and

in 32:43--changes that produced false readings in favor of angelic

beings in both cases, while simultaneously accounting for all the

consonants in Myhlx in 4QDeutj. Such a coincidence is possible, but

it stretches credulity to argue that the Masoretic text of Deuteron-

omy 32:8 and 43 best represents the original text when (a) the ex-

clusion of heavenly beings in verse 43 is so obviously a textual mi-

nus and (b) its conceptual parallel in verse 8 cannot coherently ac-

count for how the Septuagintal reading for verse 8 may have

arisen. It is far more likely that both texts were intentionally al-

tered in the Masoretic text for the same reason, namely, to elimi-

nate a reference to heavenly beings in order to avoid allegedly poly-

theistic language. It is inconceivable that a scribe would have done

the reverse, that is, altering an innocuous lxrWy ynb ("sons of Israel")

to a potentially explosive Myhlx ynb ("sons of God"). Therefore the

reading in the Septuagint sufficiently explains how the Masoretic

reading could have arisen, but the alternative does not.

 

DEUTERONOMY 32:8 IN LIGHT OF GOD'S DIVINE COUNCIL

                        IN THE HEBREW BIBLE

Although some may fear that adopting the Septuagintal reading

for Deuteronomy 32:8 amounts to embracing the notion that Yah-

weh is the author of polytheism, this is not the case at all. In fact a

proper understanding of the concept of the "divine council" in the

Old Testament provides a decisive argument in favor of the Sep-

tuagint/Qumran reading.

            The Old Testament often reflects literary and religious contact

 

            24 Ibid.

            25 Ibid., 516-17.



60                    BIBLIOTHECA SACRA / January-March 2001

 

between Israel and her ancient Near Eastern neighbors. One evi-

dence of such contact concerns a "divine council" or "divine assem-

bly" presided over by a chief deity.26 Of particular interest to the

study at hand are the Ugaritic texts, since that language bears a

close linguistic affinity to biblical Hebrew.27

 

            THE DIVINE COUNCIL IN THE OLD TESTAMENT

An example of the divine council assembled for deliberation is in 1

Kings 22:19-23 (cf. 2 Chron. 18:18-22).28 First Kings 22:1-18 in-

troduces the political alliance forged between Jehoshaphat of

Judah and the king of Israel for invading Ramoth Gilead, the ap-

proval of the plan by four hundred prophets of Israel, and Je-

hoshaphat's insistence on hearing from a true prophet of Yahweh

concerning the matter. The king of Israel revealed that there was

indeed a prophet of God, Micaiah ben Imlah, whom they could con-

sult, but that Micaiah never prophesied anything favorable about

him. Micaiah was summoned, and at first he mockingly prophesied

blessing for the inva