SOME REMARKS O N T H E SONG OF IN DEUTERONOMY XXXII

BY

MOSES

W. F. ALBRIGHT
Baltimore (Md., U.S.A.)

The Song of Moses is one of the most impressive religious poems in the entire Old Testament, but it differs so strikingly from other poems in genre x) that it has been exceedingly hard to date. The views of serious scholars have in the past ranged over nearly a millennium, but there is lately a strong tendency to date the Song earlier. So, for instance, in the first edition of my book, From the Stone Age to Christianity (1940) I dated it about the seventh century B.C. (p. 227). In the latest edition, on the other hand, published seventeen years later (1957), I dated it "apparently" in the tenth century (p. 296). Meanwhile Otto EISSFELDT has gone much farther. In the second edition of his famous Einleitung in das Alte Testament (1956) he refers the Song to the middle of the eleventh century B.C. (pp. 271 f.). This very early date is defended at length in a monograph on Deut. xxxii and Psalm lxxviii, published two years later, where he proposes as termini post and ante quern 1070 and 1020 B.C. 2) My own first reaction was a somewhat dazed admiration for Eissfeldt's daring, but after rereading the chapter over and over again at intervals I have come around to his eleventh-century datingthough a little later with quite different interpretations of some key passages. In this short essay I shall limit myself to observations on a few passages. Here again the Dead Sea Scrolls have come to our rescue, with Mgr. Patrick W. SKEHAN'S publication of a large fragment of a stichometric text of the Song, in typical book-hand from about the Christian
x ) S. R. DRIVER'S comments (An Introduction to the Literature of the Old Testament, 6th ed., 1897, p . 96) vividly illustrate the distance traversed by literary criticism during the past half-century. He comments on the " m a t u r i t y " of thought and style, comparing Deut. xxxii with H o s . ii, Jer. ii, Ezek xx, Psalm evi, as a historical retrospect. Needless to say, there is little in c o m m o n when we turn to compare form and language. 2 ) Das Lied Moses Deuteronomium 32 1-43 und das Lehrgedicht Asaphs Psalm 78 samt einer Analyse der Umgehung des Mose-Liedes {Berichte ber die Verb. d. Sachs, Akad. d. Wiss. %u Leipzig, Phil.-hist. Kl.y 104: 5, Berlin, 1958).

340

W. F. ALBRIGHT

era or a little earlier. x ) Verse 43 is preserved in six cola, as against the four of M T and the eight of LXX. The question is : Have we secondary expansion or contraction of the original text? SKEHAN, following an earlier reconstruction, made before anyone dreamed of the Qumran discovery, proposed a four-colon modification of the MT obtained with the aid of LXX 2 ) CROSS, in a very careful analysis of the recensional variants, agrees substantially with SKEHAN 3 ) . EISSFELDT proposes a six-colon reconstruction. Interesting suggestions were also made by H. L. GINSBERG 4 ), . H. TUR-SINAI 5 ) and others, before the Qumran text appeared, and similar efforts have been made by various scholars subsequently. 6 ) In 1955 I proposed an eight-colon reconstruction 7 ), which I should now modify in detail but which still seems to me more likely than the curtailments offered by other scholars 8 ) : Harnn smayim 'imm Harnn gymy et-camm K dam banw yiqqom 10 )
2

we-histahw-l 9) bene ^Elhm we-hi^q(J)-l kol mal^ak *El we-nqm ysb le-sdraw we-kipper *admat ""ammo n )

^ BASOR, 136, 12-15. ) Ibid., p . 15. 3 ) The Ancient Library of Qumran and Modem Biblical Studies (New York, 1958), pp. 135 ff. For another loss in vss. 14 f. see SKEHAN, JBL, 78, p . 22. 4 ) TarbiZ, 24 (1954/55), pp. 1-3. 5 ) Tarbiz, 24, p . 232. 6 ) See EISSFELDT'S convenient survey, op. cit., p . 14, n. 1. 7 ) BASOR, 140, p p . 32 f., . 27. 8 ) T h e metrical structure appears as a rule t o consist of bicola with 3 + 3 beats, but in the extant text 3 + 2 structure is often found, and our knowledge of H e b r e w metrics remains in its infancy. My transcription of H e b r e w is as simple as practicable. Since there was as yet not spirantization of BGDKPT I have not indicated it except in the case of p-f, where consistency would have flouted all pronunciations employed today. 9 ) SKEHAN and CROSS (doubtfully) suggest we-hab c% for the unique Greek , but this seems rather arbitrary. 10 ) It seems likely that the colon -l-mesanne^wyesallm, which appears both in Q and G, but which is missing in M T , comes from verse 4 1 , as thought by S K E H A N and CROSS. However, the person is different in vss. 41 and 43, there are other partial repetitions in the chapter, and we may have a tricolon here (as certainly in verse 39b). Since it does not fit the eighth colon at all, I omit it and suppose that the original seventh colon has been lost. It should be observed that the sixth colon is not close enough to the parallel text in 41 to warrant cutting it o u t t o o . u ) Cf. the suggestions made by GINSBERG and T U R - S I N A I to read some word for " b l o o d " here. With reference t o the latter's comparison with Accadian adammatu, " d a r k b l o o d " , cf. my o w n view that the antonym of the word in question, sarqu, " r e d b l o o d , " appears in Ugantic with the same meaning (BASOR, 83, 41)

SOME REMARKS ON THE SONG OF MOSES

341

This may be rendered: Rejoice with Him, O heavens, and bow before him, O sons of God! Rejoice with His people, O nations, and work hard(?) x) for it, O angels of God! Truly He will avenge the blood of His sons, and He will visit retribution on His foes; and He will cleanse the land of His people! This restoration is based primarily on G, which is put back into Hebrew in accord with the Qumran fragment, whenever possible. The first colon has been lost in M T by obvious vertical dittography, but G and Q both have it. The second colon has become partially conflated with the fourth; /o/may have come into the Vorlage of G from the fourth colon, which is preserved only in G. For bene yElhm cf. Deut. xxxii 8 in Q. The third colon follows M T (with an inserted V ) and G; it has been lost in Q just as the first colon was lost in MT. The fourth colon follows the Greek literally, but since it seems to make good sense, it may be approximately correct. The fifth colon follows G and Q ; cabdn> in M T was presumably substituted for banw because of its occurrence in verse 36. The sixth colon appears in all three recensions, but there has been dittography in G. The seventh colon seems to have vanished from all our recensions, leaving the eighth colon a torso which may be variously interpreted. Of course, I should not insist on the correctness of my point of view, but it does accord with the increasing evidence from the Qumran Scrolls that our Hebrew originals, once edited in antiquity, suffered far more from omissions by copyists than from additions. In other words, glosses and conflate readings can seldom be detected on the basis of recensional differences alone, though we are often justified in assuming from the state of a text that they may have been incorporated into it by the original editor. The textual condition of M T elsewhere in the Song of Moses is much better than might be inferred from the above discussion, but there are numerous examples of similar phenomena elsewhere in it. Verse 17, for example, is awkward and strange as it stands in M T ; G repeated jedcum in the third colon, disregarding the otherwise
x ) Cf. Deut. xii 23, Josh, xxiii 6, etc., for similar idioms. This rendering is only a guess, since we do not seem to have any good parallel.

342

W. F. ALBRIGHT

wholly unattested vcrbScR (secrum). ) H. WINCKLER seems to have been on the right track in vocalizing seHrm^ "demons", which he thought stood in parallelism to sdm. 2) However, the latter is an Accadian loanword through Aramaic, 3) and in this passage it is more likely to have been itself a corruption of an original secrm (preserved in consonantal form by a scribal error, for which there are several possible explanations). Read perhaps as follows: Yi^beh lisHrm ly-yElah ^Elhm lo* jed^um hresm mq-qrob < yalhm > (we-)l* jed cm ^abthm 4)

The first colon agrees with LXX, and so do the third and fourth cola, aside from the transposition of the third colon. The second colon requires only the insertion of a word lost by simple haplography to yield perfect sense, though with different vocalization; the less said of hadsim miq-qrb b^ the better, especially since b' is transparent dittography. Render the verse: They sacrifice to demons, not divine, who are too deaf to approach 5 ), Gods whom they know not, (and) whom their fathers did not know. The two most important passages from the standpoint of religious history are verses 39 and 8 f. The former requires no emendation ; it may be rendered: Behold now, I am I and there is no other God than I; I kill and restore to life, after I have smitten I heal, and none can save from my hand!
) Arab. facara, " t o k n o w " , does not help much, since n o cognates are known. ) Geschichte Israels, II (1900), p . 133. 3 ) This has always been my own assumption, but it cannot be proved that the w o r d is not early in Northwest Semitic. In 1942 (BASOR, 87, p p . 29-32) L E V I DELLA V I D A pointed out that the divine name Sdrp* (Greek Satrapes) in Phoenician, Palmyrene and Greek inscriptions from Syria, also occurs in N e o - P u n i c ; it is, therefore, perhaps very much older than the Persian period (it has n o connexion whatever with Persian satraps!). It could be interpreted as "Healing G e n i u s , " from an earlier *sdu rpPu, in which case the well-known Accadian meaning " g o o d and evil spirit" would be attested for an early stage of Northwest Semitic. 4 ) Read with the Greek. 6 ) Following the identical syntax and idiom in H a b . i 1 3 : te hr Hnyim m-re^t rac, " t o o pure of eyes to look at evil."
2 x

SOME REMARKS ON THE SONG OF MOSES

343

The first colon has been completely misunderstood by most translators. There is no mysterious divine name " H e , " but only a copulative pronoun of a type familiar in Hebrew, Aramaic and Arabic; the unusual order, *an *an h* instead of *an hu* >an is exactly paralleled by the fifth colon, we-*en miy-jad masslinstead o we-en masslmiy-jad. Such free word-order may often be found in Ugaritic, where the preservation of case endings made the rigid word-order characteristic of later Hebrew unnecessary. The other important theological passage is verse 8 f., where the new Qumran fragments published by Mgr. SKEHAN confirm a very important reading of LXX, bene *Elhm (so; see JBL, 78, p. 22) instead of the bene Yisr*el of MT. The scribal error could not be more natural, in view of the fact that both expressions were so common. Here I must differ from EISSFELDT, who thinks that the appellation c Elyon refers to the later-forgotten head of the pantheon, to whom Yahweh was subject like the other bene *EL On the contrary, we have here merely another example of parallelism carried over groups of verses (cf. verses 21 and 30 f. for cases). Render: When the Most High distributed lots, when He separated the children of man, He set the borders of the peoples, like the number of sons of God; But Yahweh's portion is His people, Jacob is his allotted domain! In other words, c Elyon = Yahweh kept Israel for His own special domain. This does not, of course, mean that in some earlier polytheistic form of this ethnogonic myth two gods were not involved. It may be pointed out that the bene *Elhm ( = Ugaritic bn *El) were in pre-Mosaic times also the bnu *Athirat (Asherah being the consort of El), whose number is given in the Canaanite Baal Epic as seventy. In the Table of Nations (Gen. x) we have approximately seventy peoples listed, and in the Haggada we read that seventy angels were appointed by God to rule the seventy nations. x) This is, therefore,
*) See especially the material collected by Louis GINZBERG in his Legends of the Jews, I, p . 181, and V, pp. 204 f. (which gives the references). I was perhaps not entirely w r o n g in From the Stone Age to Christianity (1940), p . 227, in identifying the "sons of G o d " here with the stars, but the emphasis is not on the stars but on the angels as members of the heavenly assembly (see most recently F . M. CROSS, JNES, X I I , 274-277).

344

W. F. ALBRIGHT

a very old motif, which may perfectly well go back into Canaanite times; it has often been suggested that the Table of Nations is modeled after a Phoenician prototype. x) It is, on the other hand, doubtful whether we should attribute much special significance to the interesting passage in verse 21, where we may render: They set up a no-god as my rival, 2) they made me angry by their follies; So I will set up a no-people as their rival, I will make them angry with a foolish nation. The foolish nation (gqy nbl) appears in verse 6 as an appellation of Israel itself (cam nbl we-l* hkm), so it seems clear that the expression l*-cam refers precisely to Israel, against whose follies God is warning its people. There is no need to turn to the Philistines or to any other non-Israelite nation in this particular context, though EISSFELDT must be right in considering the former as the enemy par excellence in this poem. Characteristic of the frequent repetitions in the poem is verse 16, where we find the same two verbs in parallelism: They make Him jealous of strange < gods > , 3 ) They make Him angry with < foreign(?) > abominations. 4 ) The style of the Song of Moses is intermediate between archaic repetitive parallelism, such as we find in the Songs of Miriam 5) and Deborah (as well as in the catalogue of very ancient hymns preserved in Psalm lxviii, and the tenth-century style of the lyric Lament of David in 2 Sam. i, etc. (as well as in such Psalms as xviii = 2 Sam. xxii 6 )
) The most obvious point of contact is the derivation of Canaan from Ham rather than Shem, in agreement with Phoenician ideas but against the linguistic situation. Yet it cannot be denied that Gen. is, as it stands, a product of Israel; I should n o w date the "Sefer hat-Toledot" to the late tenth century B.C. 2 ) I owe this convincing rendering of qine^n to Mr. Shalom PAUL of the Jewish Theological Seminary of America, w h o compares Eccl. iv 4, where qitfh is translated "rivalry" by a number of recent commentators; he also compares Baba Batra (Talm. Bab.), 21, "the rivalry (qin^at) of scribes increases wisdom.' We should presumably replace the hifcil vocalization of the t w o verbs in 21b with picel; n o change in consonants is required. There is certainly no need to identify the foolish nation with any people other than Israel itself; for a list of past suggestions see EISSFELDT, op. cit., pp. 16 f. 3 ) Insert *lm or ^elhm obviously lost by vertical haplography. 4 ) Perhaps insert <%art or rct (as in Ezekiel). 5 ) See the detailed examination of this poem by F . M. CROSS and D . N . F R E E D MAN, JNES X I V (1955), 237-250. 6 ) Well treated by CROSS and FREEDMAN in a recent study. There is no evidence
x

SOME REMARKS ON THE SONG OF MOSES

345

and lxxviii x ), which are not much later). In these later poems repetition of words seems to have disappeared, except in refrains; its place is taken by paronomasia. (It is, of course, too soon to propose cogent hypotheses about the origin of the non-repetitive style, except to say that it belongs with a narrative rather than a lyric category.) In our poem we find only a few examples of repetition according to familiar Canaanite patterns, but we do find much repetition of words and phrases in adjacent bicola. Besides the examples already quoted in verses 17, 21, 43, note especially verse 30 f.: *ekh yirdf *ehd *elef *im-l* k sram mekrm k-l* ke-srm 2) srn u-senyim jnisu rebbh we-Yahwh . . . hisgirm < ha-l* > 3) *qyebenpellm

The foe is described as crediting the God of Israel with deliberate punishment of His own people, an astonished assertion which the poet emphatically accepts: " H o w can one man chase a thousand, and two put a myriad to flight, Unless their "rock" had sold them and Yahweh had handed them over. . ?" Truly our "Rock" is not like (their) rocks < are not > our foes the judges? In these lines we have a very archaic use of sr in the sense of " g o d " 4) as well as an elaborate series of repetitions, with the word taking three different endings. We also have an example of transposition through confusion of two similar words, with which we may compare an apparently unrecognized transposition in verse 19, which should read:
of early orthography in the psalm, but it is unnecessary to suppose that the psalm was at all obscure to the u n k n o w n scribe w h o copied it in a somewhat later preexilic spelling. x ) I accept EISSFELDT'S argument for a Davidic date of Psalm lxxviii (op. cit., p p . 26-43), which is historically important, because of the otherwise unique references to early Israelite history and geography. 2 ) This vocalization is demanded by the context. 3 ) By inserting ha-l' we obtain better assonance, better metreand better sense. 4 ) It must be remembered that sr in the sense " g o d " is derived from the word *%uru (for still older *%uhru, "back, ridge") = Aramaic tura and Ugaritic guru (likegm^a, " b e thirsty," Arab. %amPa, etc., etc., as shown by GOETZE), " m o u n t a i n " . In exactly this sense the w o r d srm appears in parallelism with gebcot, "hills", in the Balaam Oracles from the late thirteenth century; see my remarks inJBL 63 (1944), 212. In second-millennium Syria and Anatolia all important mountains were deities.

346

W. F. ALBRIGHT, SONG OF MOSES

way-jar* Yahwh waj-jikcds mm-na*as1) banw -bnotw Yahweh saw and He became angry at the insults of His sons and daughters. Cases of archaic morphology and vocabulary are common in the Song of Moses ; there are many which have not been mentioned here. It is hard not to see a number of instances of archaic consonantal spelling without the matres lectionis at the end of words, which generally suggest a written original not later than the tenth century B.C. 2) EISSFELDT'S early date for Deut. xxxii thus appears to be justified. He is also clearly right in recognizing the Philistine period as the background of the poem, though I should go farther and point to the intransigent monotheism of the author, which is made all the more vivid by his archaic imagery. Such virile monotheism belongs to a time when Yahwism was fighting for its life again st both external and internal foesin brief, to the period when Samuel rallied Israel against its hereditary enemy as well as against the paganism rampant m its midst. We cannot directly compare the literary genre of Deut. xxxii with the Blessing of Moses in Deut. xxxiu, since the latter belongs to a category of tribal blessings which goes back long before the Song of Deborah and is therefore replete with extreme archaisms While we can probably date Deut xxxiu before the Philistine conquest of Israel in the third quarter of the eleventh century, it may be considerably later than the Song of Deborah 3) The Blessing of Jacob probably dates after the Philistine conquest but before Saul's reign 4) and the solitary incipit of this category which we find m Psalm lxvm may belong to the reign of Saul. 5)
x
2

) With the same sense as ne^asah [Cf JBL, 78, 22 ]

63, 208 ff, CROSS and FREEDMAN, Early Hebrew Ortho

) Cf A L B R I G H T , JBL

graphy (New Haven, 1952) and JNES X I V (1955), 237 ff 3 ) Note the striking absence in b o t h J u d g and D e u t xxxiu of the plays o n the names of tribes which characterize G e n xhx and the slightly later Lament of David T h e bitter criticism of Levi in G e n xhx seems t o point t o the period of lowest Levitic fortunes in the second half of t h e eleventh century, and is in striking contrast to the earlier praise of Levi in Deut xxxiu O n the great archaism of the latter see CROSS and FREEDMAN, JBL 67 (1948), 191 210, I heartily concur with their orthographic argument for the dating of the Blessing of Moses in the eleventh century 4 ) Verse 8b does not fit and is presumably a later insertion As yet n o up-to-date treatment of the Blessing of Jacob has appeared, t h o u g h there are numerous Vorarbeiten Here again we must draw heavily o n Ugaritic parallels and o n recog nition of archaic orthography 5 ) Cf HUCA X X I I I , 1 (1950 51), p p 30 f According t o MOWINCKEL the entire psalm belongs t o the reign of Saul, whereas I should date much of its con tent t o the tenth century

^ s
Copyright and Use: As an ATLAS user, you may print, download, or send articles for individual use according to fair use as defined by U.S. and international copyright law and as otherwise authorized under your respective ATLAS subscriber agreement. No content may be copied or emailed to multiple sites or publicly posted without the copyright holder(s)' express written permission. Any use, decompiling, reproduction, or distribution of this journal in excess of fair use provisions may be a violation of copyright law. This journal is made available to you through the ATLAS collection with permission from the copyright holder(s). The copyright holder for an entire issue of a journal typically is the journal owner, who also may own the copyright in each article. However, for certain articles, the author of the article may maintain the copyright in the article. Please contact the copyright holder(s) to request permission to use an article or specific work for any use not covered by the fair use provisions of the copyright laws or covered by your respective ATLAS subscriber agreement. For information regarding the copyright holder(s), please refer to the copyright information in the journal, if available, or contact ATLA to request contact information for the copyright holder(s). About ATLAS: The ATLA Serials (ATLAS) collection contains electronic versions of previously published religion and theology journals reproduced with permission. The ATLAS collection is owned and managed by the American Theological Library Association (ATLA) and received initial funding from Lilly Endowment Inc. The design and final form of this electronic document is the property of the American Theological Library Association.