The Barberini Greek Version of Habakkuk III

Author(s): Edwin M. Good

Source: Vetus Testamentum, Vol. 9, Fasc. 1 (Jan., 1959), pp. 11-30
Published by: Brill
Stable URL: https://www.jstor.org/stable/1516167
Accessed: 17-04-2020 15:59 UTC

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/1516167?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Brill is collaborating with JSTOR to digitize, preserve and extend access to Vetus
Testamentum

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
 THE BARBERINI GREEK VERSION
OF HABAKKUK III

BY

EDWIN M. GOOD
Stanford, Calif.

The Barberini version (hereafter abbreviated Barb.) is

because of its occurrence in the Barberini Codex, MS Barberinus
Gr. 549 in the Vatican Library (HOLMES and PARSONS no. 86).
MARGOLIS called it "Anon." for Anonymous 1), and THACKERAY
called it "Oxford", because it occurs also in H-P 62 and 147, which
are at Oxford 2).
Barb. is a non-LXX, Greek translation of Habakkuk iii alone.
It shows no outstanding affinities with any other known Greek
version of this or any other OT passage. The scribe of the Barberini
Codex, knowing that it was not LXX, Aquila, Symmachus, or
Theodotion, wondered if it might be Quinta or Sexta 3). Examination
of extant fragments of these versions proves that there is no contact.
DE MONTFAUCON was of the opinion that this version is Septima 4).
Since no fragments of Septima for our chapter exist, the conjecture
cannot be tested.
The version is contained in five MSS, four of which also present
the LXX text. The exception is V (Codex Venetus, H-P 23), which
gives only Barb., though in a form which shows considerable adap-
tation to LXX. The other four are minuscules, H-P 62, 86, 147, and
407. All of the MSS are of medieval date, V being 8th centuly, 407
9th century, 86 9th or 10th century, 62 11th century, and 147 12th
century. The four minuscules give the Barb. text first and follow it
with the LXX text.

1) "The Character of the Anonymous Greek Version of Habakkuk 3", Old

Testament and Semitic Studies in Memory of William Rainey Harper (Chicago, 1908),
Vol. I, pp. 131-142.
2) The Septuagint and Jewish Worship (London, 1921), p. 48.
3) Cf. the colophon to the Barberini Codex, quoted by MARGOLIS, op. cit.,
p. 133, and by KLOSTERMANN, Analecta zur Septuaginta, Hexapla und Patristik
(Leipzig, 1895), p. 60: Trv 8031v TOU 'Alopaoxou[ ouX eupov autcpovouSoav OUTsr
To0Z O OUT? axoah O&UT? o CaX OuUT?i S aE080CV TLjVL- aTS oUv, ?L T( _ E T)j] 5
eix8o6ao ?aTLv.
4) Noted by MARGOLIS, op. Cit., p. 133.

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
12 E. M. GOOD

1. The Text of the Barberini Version

By extended comment on the Greek text, I will attempt to show
in detail the Hebrew Vorlage, the translation methods, and the
contacts of Barb. with other versions. Concluding sections will
organize these materials systematically.

1 nIpoazsuX 'A 43 oxou.1') |I?T' ()8 I 2).

2 Kupt?, taaxqxooa -?v axoyv aou xoat ?Xa306v 3).
xupL?, xacTv7co?a 4) 'T ?pyor aou xoc ??crTnv 5).
?v EIo> aUIo 4Xov 6) yvoxC9Na6 7)
&V lT ?YyL[?r.V 8) TX ?T7] 7TUyvo MV0j09'y
SV T') 7aCp?tvaOt TOY xaOLpOV otva?sLXZiOan 9)
ev T(O) Tap'oXy7va ( To V 47Xuv {ou 10) ?V 4pyfn ?x?ou 11) pvyrOe<6n.
3 o6 8%?0;S 07& X lpC 12) ^?1
xao ay0 ocL S opouq c(potv. S?TCoXrA LiaUoLTo0 13C).

1) 86 and 407 add the missing rou TCpoypxTou in agreement with MT and LXX.
2) This rendering (= LXX) is probably a quess at MT 1n11"lJ S,
3) This is Barb.'s sole departure from LXX in v. 2, the two versions having
been conflated in this verse. LXX reads 6(popB0-6v. It presupposes n'1i'l rather
than MT MTnK'
4) Both LXX and Barb. imply Wn"1 for MT n'ln, and the word is a doublet
to eutjap30Yv as well.
5) A doublet of xodL euXcaQ0vv.
6) The Heb. Vorlage may have been T:11l?
7) Whether yvwoo0acTl presupposes the addition of P'T11ln in this line, or
whether it is only an anticipation of the verb in the next line is not certain. The
latter seems more likely.
8) This phrase, together with &v zijo) 7ratpzv in the next line, presupposes
a repointing of MT :1j3 to ij-. Aquila has the same words.
9) This line is certainly a doublet of the preceding, and it may have been the
original Barb. rendering of S7ln P13f ::1p::
10) These words presuppose the non-MT Vorlage "rl l'ln. Note the doublet
following.
11) Here we have the equivalent of MT 3nl1 t-:l
12) Barb. has rendered MT J?Vtn in its general meaning, "south", though V reads
0pet&v. Cf. Vulg. and a Sorbonne Old Latin Psalter: ab austro; Theodotion:
&oO v6OT(t)ou; Targum: Rm1''f}?.
13) The very curious expression was explained by THACKERAY (op. cit., p. 50,
and "Primitive Lectionary Notes in the Psalm of Habakkuk", JTS 12 [1910/11],
p. 206) as a lectionary note. He observes that txropoXyo] in Isa. xxx 32 seems to
correspond to fHD1lf, the technical term of "wave-offering", mentioned in Lev.
xxiii 15, which was a primitive Pentecost lection. Thus THACKERAY would
identify the Vorlage of Tocr0ao?i 8tLoc4&X'CLocroq as t10D lb1lIn. Another possibility
seems open, since THACKERAY'S lectionary-note theory is doubtful on other
grounds. LXX reads xazTxaoxou 8ao0ao for MT ]'}l . The two words form a
doublet, and the most natural Vorlage of both would be something like "17ft,

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
 BARBERINI VERSION OF HAB. III 13

EX004zuEv ' ou'prAv6 'r-c Zperrtv rlq 30~-1q oc&

xca` t~s otiv6arwS ajo~ o h5 -1 2) .j OIXOUp ly 3).
4 8L&ocycocLoC 4) cp 0 6 C 5) Ta'UXL co o -6)
x~pot?ro 7) sX ZSLP04\ ot&rO5 IrC1pZesX 5t&C.
SXS'i Cyr U'V oC 8) t v q 7.w q 75 ocutOi5 9).
5 7rp0 70poa)7ou ocuTo5 7tpozXaeQCT L V (7aLq
xxi xcx7& 7C6~xS oc' T05 10) &XO?XOU0asL &O ElkylTrOt 7G)

"leafy, hairy." BEVENOT, among others, proposes that the

LXX was '1A7.V ("Le cantique d'Habacuc", RB 42 [1933], P.
v.ercxPoX' translates 11wO, and its cognate tLz-cxP6Xoq, rend
Isa. xxiii 3. The corruption of 1'P11X to ?110 by paracousis is
titrapoX' might go back to a corrupted Heb. doublet of
is contained in W (the Freer Codex) and Achmimic.
1) The phrase r-r'v zUepS7rLO&0CV -T- oct&ao5 is typi
dency to paraphrase. It is also contained in the Targum
LA-%,j 1090 o.), Achmimic (AtIcii [I]FICVv'), and

2) Complutensian Polyglot also reads LrXh~oft.

3) These two lines of Barb. are erroneously cited as belon
Theodotion by Codices Colbertinus and Coislinianus.
4) MT's copulative is omitted. Complutensian Polyglot reads
5) MT reads rbn~. Augustine and Cyprian in the Old La
with Barb., reading ut where the rest of Old Latin has eithe
mimic also gives the Barb. reading for this line.
6) MT T''fll is hardly rendered by C`aro otirCp. Cf. the next lin
ocuCrijC renders *. It seems likely that the Vorlage of the line w
7) Barb., like LXX, has mistaken D"I'11P for a plural. The p
and MT's form can only be dual.
8) The two words form a doublet. MT reads 132h, and ixEZ
But e'7rozprx-7r.r certainly renders t3t MARGOLIS (op. cit.,
the verb is "an amplification by the translator".
9) The word order is strange. We would expect 86v0voLL to
866cx is used for tI) in Ps. lxvii (Ixviii) 34; Isa. xii 2; xlv 25
GOLIS (op. cit., p. 141; and cf. ZAFW' 25 [1905], p. 321) believ
presuppose 7'1'1= (the change from ri to 2 is possible, he
script). In Isa. xxviii 1 8uvoc[L renders "=S. It is possible ei
translation presupposes the corrupted NtI) 7JV= or that 8
at the hapax legomenon 71l=. Old Latin texts (except Augu
four Old Latin Psalters), along with H-P 239, interpolate
v. 4c.
10) The reversed order of does not indicate a non-MT Vorlage.
11) The identification of the Phoenician pestilence deity, Resheph, as "the gre
winged ones" is very similar to the translations of Aquila (7r-rvov), Symmachu
and Theodotion (05pveov), Quinta (reported by Jerome as volucer), Peshitt
(?..4), Achmimic (5sz I and Sahidic (w7; u\\ATi, and in th
citation of the text by Shenoute: 9nciui Pvi P AT). Cf. also We: 7err[votq?
THACKERAY (LXX and Jewish Worship, p. 53) quotes CLERMONT-GANNEAU
(Horus et St. Georges [Paris, 1877]) to the effect that this text identifies Reshep
with Perseus of the winged sandals (cf. the various LXX readings, one of whi

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
14 E. M. GOOD

6 arokq1) eLPYa~A) t~V yiYv

xaz~vo~ac I xas ~ 3) t& iXOViy3
aLepop 3_ xadI 4) 7&r ISp_ OpOCuaO IaeTL 6)
Lcrv&~aovtrL a v&~totr. 6) Ix To- adIvo ,
a oL oadt OCL e&op(S &doL&ON aov-ccXL 7)
OCUTOU EVZXX C) aeLa I CL IoIxouLvy 9).
7 xaxcrovsv6=xa -okq arxrV&c 'ALOL67oV,
Tatpac/f]aoVT0tL CL X oroLxo0vz8g 0 ) ra&q a6ppstL MascL&
8 Eu?~ eV 7to'0aq.toZL Op Y '-11), x PLE;
sv 7oUToC~L 4 8u~t6o aou;

is ev 7regLXOL4, contained in A, Q, and some minuscules). Res

fied in some Cypriote inscriptions (cf. COOKE, Nor/h Semiti
55 ff.) with Apollo. When we recall that Apollo was the char
and that the sun was portrayed in Egypt as a winged disc, we
Barb.'s Egyptian provenance. The Jewish translator was perhap
pagan deity and preferred this circumlocution.
1) The participle (like xcrcxvo0iaoaCq in the next line) is used
good Greek style and does not indicate a non-MT IVorlage.
2) MT's Po cel is presumed here to be Pi cel, which is a better
verb.
3) This is a hapax legomenon in the Greek versions, though it also occurs here in
Complutensian Polyglot as Zx'oorv, and is presumed by to-wll in the Pa-
lestine Syriac (though there the word stands in the place of xx-tavo aoc4). It may
translate ?11N, a form of '1I , "to spy out, explore", interpreted by zieLx&cIco,
'to liken, compare."
4) Barb. has divided words differently. MT St12=11'11 has been read

5) Another of Barb.'s doublets, this probably renders rV12V (cf. Deut. xxviii 33;
Isa. xlii 4; lviii 6), a corruption of frI?'. Old Latin Speculum seems to presup-
pose this reading with quassati, and cf. Palestinian Syriac. MT 7W has been omitted.
6) As a translation of 21=L7, oc' vn7rocL looks strange, suggesting rather 1'NX?
But cf. Isa. xl 12, where LXX has rendered '1=11 by vck7ny.
7) Still another doublet, this one of ca7reLvco0'a0V-L, v. 6d. 'AuJoL6
almost always renders i2?ZT, and this is probably a corrupted doublet of MT
W1tri. At the same time, the future passives of Barb. may presuppose Heb. imper-
fects.
8) This could render MT 1j, though 1117?2 would be more likely.
9) YLeto, aSrXL -i) otxou[ivzl seems to be a doublet of v. 6a, perhaps pre-
supposing ?31?Z' for T77?2 (as does LXX in v. 6a). BiWVENOT (op. cit., p. 506, and cf.
MARGOLIS, op. cit., p. 138) thinks that the whole line rests on Heb. r' r)f 1f ,
which would be related to flK 7 1nn, v. 7a, otherwise omitted by Barb.
10) Either MT YIN was not in the Vor/age; or xcxroLxo53vr-e presupposes Z
before 11174V1; or xcx oLxo5v-eq renders r'I in a revised word order, not unus-
ual for Barb. (cf. xa-ro&xo5v-rocq == Y' in Jer. xxvii (1) 45); or o'L X0CoLLXoUVTeq ra
86ppecq Moc&&C4L is a free translation of 7'r? 1 V1?P'V. The last alternative
seems most likely.
11) The form may presuppose an imperfect rather than MT's perfect. It may,
however, be paracousis for the probable LXX original, W'pyL'aoOy (though
most LXX MSS agree with Barb.).

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
 BARBERINI VERSION OF HAB. III 15

sv OoctXaaoa ' opy' cou;

1) &vrpi 2) E 7ao o"lpCCTx 3) COU,
=7 l7m Oa aou 4) CKxaQpLao 8 Cpo0s3P1 5).
9 ' p)Y?Op6o 6) To T6'ov aou
CXopTi:acac 7)) 3ora8a 8) T cppxppc xoU 9). at2a
7oToc|Aout 8ta<axer.aro?cl xoal yiv JioGcI^ 10).

1) MT "' is omitted.
2) This would seem to render a Heb. perfect, n:r, rather tha
3) "Ap[L0o is never elsewhere used for 010. It accords better w
appears in the next line. (Cf. note 4).
4) The terms may have been reversed. 'H tWrartolc renders
and LXX of this passage.
5) The basis of this expression is all but impossible to dete
(op. cit., p. 142) calls it "an unsolved problem of identific
usually renders MN1, ;lH, or T'1'. Given Barb.'s propensity
the ends of lines or groups of lines, we might expect 6 npoP3c
&vPrsq. But no likely Heb. equivalent of irpooclvc can be id
ruption of MT F1n. Furthermore, the relative pronoun, 6, c
a translation of 3, omitted by Barb. in the previous line, or a
(LXX for '"), but would more probably render 'lttR. It is b
6 rcpo6i5 is a witness to a divergent Heb. form of the po
:l'nn rather than :3n111 ". Certainty is lacking.
6) It is unlike Barb. to condense, but that seems to be the
was probably taken as an absolute infinitive, acting to streng
eiF,y?pOh renders the whole expression, '11ln S'71 . On the
possible 1) that ;1t"V had dropped out of the Vorlage, or 2) th
something like it, perhaps a form of TI1. 'E3y?pO0h seem
perfect form, nfl7, for MT's imperfect.
7) This line in Barb. has attracted a good deal of attention
its rendering of this most baffling Heb. sentence is a helpful sou
(DUHM [Das Buch Habakuk (Ttibingen, 1905), p. 86] took a d
said, "Jene griechische Obersetzung ist eher ein interessante B
Konjektur aus alter Zeit.") Clearly iX6pTr:cacS renders nlV
n1S?tll. Peshitta has read the same Heb. with - L =Jo. The number of scholars
who have accepted this or a similar emendation is legion. They include TORREY
(n217'), EHRLICH, GUNKEL, GUTHE, MARTI, NICOLARDOT (nI2 '), G. A. SMITH,
WARD (?it), ALBRIGHT (nYit), CASSUTO, RUBEN (nlslt.), ELLIGER, HORST,
HUMBERT, IRWIN, NOTSCHER, NOWACK, PROCKSCH, SELLIN, TRINQUET, and
WADE (157a).
T - -

8) Though 6?tXL is not an accurate tra

LXX), it is probably intended to transla
9) Codex V reads zqT cpocyappac; aou
BEVENOT, op. cit., p. 515; LACHMANN, D
or In" (so THACKERAY, JTS 12 [1910/11
10) Another case of duplication. The cur
line seem to have non-MT Vorlagen. Ne
renders Ij'?. The possibility of determi
mote. '1fl may lie behind &Socaxesaarz

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
16 E. M. GOOD

10 Ev '6 &Vr0O?0CXp,eZV as r0pOaCZ0oUV'VCt 1) 71 se_?

eV Th 'QV e5aOCELOV aou ' 4)
[L3V3) OLSAXISLV at' _T .4

- p &f3uaaoq 'vyc 'v_s 5) ELSZov 6).

11 ?p&O -r ?o CXP,7Zp0v 05 )LOU Z'7reCX7Z 7)
?r o cp6yyoS r~5 ~Ehrasx 6 a- Orj 8)
y- cpEyyoq tr v 3oX(~&v aou 9) n
k pyy N 10) 4&alpo= voqxocYoCza[ aou.
12 ~E~h ~CL~O~ ~Yeej~~' l 7CI X V yY,V 12),
~.ts- opy~q &XXO-~cas CM-~
13 &vZ?p&v l 13) f c 7oi~ ?coo aou?
27; xxiv 5, etc.), and tZ9111 may be presupposed by aetLart (cf. Hag. ii 7 (6);
Isa. xiv 16; Ezek. xxxi 16, etc.), but these are only guesses. 1jfll does not seem
to have been in the Vorlage.
1) Barb. has rendered the asyndetic *'7'n 7JIX by a good idiomatic Greek
construction, using the prepositional infinitive phrase as the participles in v. 6
were used.
2) Where LXX is haggadic, using Xocot for P11f, Barb. is literal.
3) This rather complex phrase is doubtless intended as an interpretative trans-
lation of M"M t3'1.
4) AL' ot,cx) seems to imply the addition of T1 after '1Y.
5) As in v. 9a, Barb. has condensed. This clearly represents 1'i1j In1.
6) Where MT 1311 is the first word of v. 10d, Barb. takes it (reading 31) with
v. 10c. The rest of v. 10d is taken with v. h a.
7) This is a very difficult line, as is indicated by the fact that MARGoLIs gave
up on its solution (op. cit., p. 142). The equivalent in MT is clearly Htvl' 1,77,11
Tint. As we see elsewhere in the version, the translator was quite free in the
rearrangement of words. Here 1V77' has been figuratively rendered by cpcJd -o'
Xoct7pov, this being the meaning of the sun's "hands." Since the "hands" be-
long to the sun, the translator has rendered the line as if It'T'7 and ?tintf were in
construct. And XtVz, "lift up", is translated "to spread out", emphasizing its
contrast with 1M7 in the next line.
8) The same phenomenon as in the previous line can be observed here. Barb.
translates as if the line read '7nv ?VV 'T. This does not imply that this was the
Vorlage. The translator was trying to give an adequate translation.
9) Palestinian Syriac, with r \O*4o4LS, seems to presuppose Barb.'s
reading, as do Achmimic and Vulgate.
10) It is strange that qg6yyoq occurs three times in this verse, each time as the
equivalent of a different Heb. word ('=T, lIX, and ; One wonders whether
it was a terminus technicus among Hellenistic cults in Egypt or elsewhere, but
evidence is lacking.
11) Since ZyepO'afl is not a good translation of MT 71f211. we may suggest
the corrupted j1flfly as the Vorlage (cf. Cs'yepo = '111, v. 9a; 'yeLpo = '11,
Ps. cvii [cviii] 2; Prov. x 12; Cant. ii 7; iii 5; viii 4; Joel iii [iv] 12, etc.). Palestinian
Syriac seems to have the same reading with 's&E.t?.
12) Although it is not impossible that Barb. read rlX '37, MT is probably pre-
supposed.
13) This word is cited by MARGoLIs (op. cit., p. 136) as evidence of Barb.'s
Targumic character, since Targum to the passage is XZV'1'UX. 'Avecpm&vrq could,

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
 BARBERINI VERSION OF HAB. III 17

puba6oot T0OU; SxXx~:o6u; aou' 1)

xcaTsTcoSuaas x?cpaaS p av9vOp7ov 2) u7.?p_y Ip vo V
?0 c B7ou a T66 O oaXaCiS xaTaSuav ovTc 4).
14 ?i?Sx7oca 5) ptl?'& Suva[xS( CTOU 6) TOU apXYO
oacapTCx)X v 8),
To 7?ToL06'oTac; 9) ?7r TN aO6aS OC aUTCOV 10)

however, render f12, from n1S, "to make to blaze up; "the
1n3S' is an easy one.
1) This reading of "1'rPf} is a most unusual one. Like LXX
Heb. to be plural. The interpretation, "the elect", is to
Latin texts (the Mozarabic Breviary, Verecundus, and the
electos tuos), and cf. the Sahidic citation (rirteCXI peCCTl^
doubtless original in Barb., probably reflects both Jew
anti-Christian polemic. The latter point is disputed by BEV
que notre traducteur a evite de se servir du mot 'christs' p
de Dieu en general, probablement par un scrupule sacerd
2) A corruption of n"13 to n1n3 is presupposed; this eme
HUMBERT, Problemes du livre d'Habacuc (Neuchatel, 1944), p.
3) Though u7trpy'cpavo; never translates Ytll elsewher
equivalent of MT here.
4) The notable fact about vv. 13-14 in Barb. is that in the
of the expanded translations so characteristic of the ve
contain more departures from MT than any other com
poem. The present line is a good example. MARGOLIs (op. c
that Scoq a&pu6oou TTrq5 QaocTaan renders '710 11" n17, an
IRI2S '7. This, however, is a much freer translation than even Barb. ever
undertakes. "Eco &O6puCoou must be something like 'I1S 'T1, perhaps H12l? T'7
(cf. Job. xli 22). We have noted above that Barb. in several places makes genitive
constructions of words which are separated in the Heb. text. Thus the genitive
TSq? 0OaxCacT; may represent a corruption of fl137 or '10". It would most na-
turally render VIH, but it might translate tl11f, perhaps a corruption of nll'Y.
KcocTa86ov-ctL would most naturally represent '181 (cf. Exod. xv 5), which might
replace '71'. This is one line in which Barb. so obviously read a non-MT Vorlage
that its determination is a matter of pure guesswork.
5) This presupposes n173 for MT 1n3p (for ixSLx& = 72, cf. Lev. xxvi 25;
Num. xxxi 2; Deut. xxxii 45; I Kings xxiv 13; Nah. i 2; Ezek. xxiv 8; xxv 12).
6) A reading l't:3;1 (MT T1t301?) is indicated, on the basis of which many scho-
lars emend.
7) The plural might suggest 'ILtR for MT t1'}, but it may be only interp
tative.
8) Since most of the versions have guessed at the meaning of 1T0, Barb. may
also have done so. MARGOLIS, however (op. cit., p. 142), notes Aquila's rendering
of r"' by &capTco)o6q in Ezek. xviii 10, and suggests that the text read 1n&.
9) MT v. 14b seems to be almost entirely omitted by Barb., though this phrase
is probably a remnant of it. MARGOLIS thinks that the translator read a Niph. of
'lO for 1YO, an easy corruption (op. cit., p. 142). HIIELt never renders t7O els-
where, nor does the Niph. of '7O occur in the OT. The equivalent of rrsit0 is
usually ntD, but MARGOLIs' identification is possible. Cf. also Peshitta, ON\l.t?.
10) MT l~5";1t is omitted by Barb. This expression presents a word used else-
Vetus Testamentum IX 2

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
18 E. M. GOOD

1) SVeXeV T0U xa?cayzev TOU5 7.TCOZOUg a6pq.

15 avrp iaotc 2) d OaX6CaaOt; T0o L7trou6 aoou'
oCTapaIXO 3) ra IIaLOCa isTara Tf aI uI6aou 4).
16 Ta:octiJ,jv 5) xaOl &vraToCT rEO IT& ITXyXva 6) ~LO,j
aX0 7) T]( Cp0Vmq 8) TOU 6TT0CJJITO5 aou 9).
ars au 10) Tp6eo04 11) e?L T0' oarec AOtu
12) xzr' spLOCUTrV ?1ocapaOv 13).
Tavra 14) VXaC^E 15) ?V 7|L?p4 0X[i?
n0CaycyEZv 16) 1s7 ivo4Q 7troXetLoV r1v AxOcv aou 17).

where in Greek versions only in Isa. xxiv 8, where it renders

show a word division within MT DI1?f37, perhaps DnS 'Y, p
"because of their anguish." Cf. Peshitta, ,oomtl.-ua.
1) MT 1?Z has been omtted.
2) MARGOLIS (op. cit., p. 140) considers the Hiph. of llt
here, since avoc3aptpc3 often renders the Hiph. of ;*1. In vi
method. there seems to be no compelling reason to assume
3) MT has the noun I?n, "heap". Barb. has read the verb
stir up."
4) Another example of circumlocutory translation. The tr
identified tl'l1 Dt with such expressions as 131 tl;ln (Amos
G. MAY, "Some Cosmic Connotations of Mayim Rabbim, 'M
JBL 74 (1955), pp. 9-10.
5) Haplography of v has produced this reading, which m
flnnt (cf. Hab. i 12; ii 9; iii 19 [LXX]; Hag. i 5, etc.).
8) A more genteel translation of 7D1 than LXX's xoXtoa.
7) For MT 71?i, Barb. seems to presuppose 'ipj.
8) Barb. has omitted MT *1'S, as has Peshitta.
9) Where MT has niDt, Barb. has read l'n[i1.
10) Barb. presupposes the perfect :1 for MT IX1'. Cf. LXX
11) LXX also uses this noun, which would accord better with I

12) MT's wvaw has been omitted.

) MT's waw has been omitted.

13) Where MT has imperfect TI3R, Barb. implies perfect nTT31

14) This clearly presupposes Heb. l*X, which might be a cor
tE or MniN.
15) MARGOLIS (op. cit., p. 142) is disposed to think that (pu
rn1n, but 9puXcaacco is not elsewhere used for nli. It would be
form as nftln. The Vorlage of the two lines may have been
following:
inttn inTn innn
mns r1r rftK
This is not good Hebrew, but neither was it the work of Habakkuk.
16) The form could represent either MT's Qal infinitive (cf. Exod. xxxiii 5) or
a Hiph. infinitive (cf. Deut. xxviii 61; Jer. xl [xxxiii] 6).
17) It is possible that 7roXeLouv txv Xoa6v oou is an expanded rendering of
1T11 (MT 1U'IT). On the other hand , Xao6 is the LXX equivalent of MT DS
(Barb. 06voO), and r6v Xo6v may be a doublet inserted, according to Barb.'s
frequent pattern, at the end of the line.

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
 BARBERINI VERSION OF HAB. III 19

17 1) a aux~ oi' E p0jZ 7v xcxpi9v o 2),

XMXL eV 'TOCL;4VOtkXtAQL QUX ESaToXL ypOX
c 3C a rLrOZ ioa ),

&X?lrett 4) iL'X ,tC~Vapo 5) 7p663POTtc

xod B636 O' tr o' pouaM e'T YpotrVULS.
1 OtpkaO LO L 6) tu)O~ la(,YLOC 7)

19 X~P!.O4 O6~ e.tu 108) EIOx& /.tOL ilaX?v 9)

xo8 xc aas 10) 'co x6 occ LOU aepot?oc 11)
xou. xt ot; 7pc3Xooho 12) ?7~jv gxOp(7v ~iou 13) ~xPP E
I 14).
1) MIT po is omitted, as it is in
the Psalm. The Old Latin of Cyp
2) A free translation of fl1bf
Latin agree (omitting a;~)
3) Either Barh. has condensed
or MT 4?Z has been omitted. Sin
we need not assume an omission.
4) The future seems to presuppose Heb. imperfect, its'. 'ExXet7w is nev
used elsewhere for 'lfl.
5) Achmimic seems to agree with Barb. here against LXX: O-T&P~ ric&vST
AT-TCt1 \i?h\& q TAI1J6&'i Pa_ and cf. WC: [&76] ~L&,vpocq 70Lo[Lvto.
6) The only difference hetween LXX and Barb. in this verse is that LXX
inserts rCo here. Both versions represent MT.
7) Complutensian Polyglot reads &yyocXa&ao~tcL, which is clearly an error,
since no such Greek verb exists.
8) Barb. has evidently taken 'f7l7 to have a first singular suffix.
9) Another of Barb.'s free translations, which implies a mild interpretation.
10) The aorist probably implies Heb. perfect, DtVl, rather than MT 13tV11. Augus-
tine, the San Michel and St. Gall Old Latin Psalters agree, reading statudt.
11) MARGoLIs (op. cit., p. 136) thinks that 'aycx)ceZq is "an interpretative para-
phrase" of the Lucianic (or Syrian) LXX: C'arit e'X&cpou. But the notion of
"safety, immovability" is a poor interpretation of "like the hind." Barb. might
hive seen in fl&*XO a form of f1:) (Hiph., "to endure, sustain"), perhaps reading
1?S'~. It is not impossible, however, that Barb. read 11*"HX as the plural of i1'NK,
" oak, terebinth" (which seems to the writer to be an improvement on MT),
and interpreted it as "immovable".
12) This is an unusual translation of Z?l7t , though it is found in Deut. xxxiii 29.
13) Does this expression imply the insertion of "VIN in the text, or is it another
free rendering (so MARGOLIS, op. cit., p. 136)? It is interesting to note that 1"VIN
appears in Deut. xxxiii 29 (cf. note 1), and that in the same verse t11 is trans-
lated hy eim3ocuvov.. We may postulate some influence of that passage on Barb.
The reference to "my enemies" was prohably the attempt to be consistent while
translating 11=1 by rpocy,'Xoq.
14) The rendering of the subscription to the Psalm is very difficult. THACKERAY
(JTS, 12, pp. 202-203). following his lectionary theory, finds in the Barb. text a
- here used for Achmimic hai (printer 's note)

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
20 E. M. GOOD

2. Characteristics of the Vorlage

One striking fact emerges from this study: Barb. has
edited in the first section of the poem (w. 1-9), wheth
form or in the Hebrew Vorlage. This is shown by the
of doublets, almost all of which occur in vv. 2-9. In v
been conflated with LXX, but the other doublets h
with LXX, and most of them represent non-MT re
doublets have frequently been added at the end of the l
in which their couterparts occur.
v. 6c Opau7O6 etcrT (= trsnr), doublet of &sOp6fp-
v. 6d-e &aXXocoL0oovcnL ( = ? l3tt), doublet of Talrlv
( ? wn)
v. 6f au'To S"vsxoc aseLaO Toz t ot ouL te'V, doublet of v. 6a
v. 8d-e o Tpo?[37]<, doublet (?) of ovbep-y
v. 9c aEdaLq, doublet of acxE0ast
In three cases this arrangement does not obtain:
v. 3b EsrxaoX4i, doublet (?) of LXX xxzocaxCou itself a doublet of
Barb. opop&v.
v. 4c ?azTz'PLXctXL (- :i), doublet of exe (=- tt )
v. 16f Xa6o, doublet (?) of 0Ovoq
The fact that these doublets correspond with demonstrable Hebrew
words, most of which could have been corruptions of MT, indicates
that they probably go back to a variant Hebrew tradition 1). The
corruption of a non-MT Heb. rubric. He thinks that aotZocxa xcreTrccralzo
renders 1'l"ti1 H1rH, an erroneous reading of nlti rn nlI, "morrow of the Sab-
bath." This phrase occurs only in Lev. xxiii 11, 15, 16, probably the earliest
Pentecost lesson, and the original rubric referred to that passage as the Torah
reading. BEVENOT (op. cit., p. 518) feels that xazTsoc6uoaCT is the equivalent of
Targ. nnWliZnD, but he arrives at this by way of a corruption to nItI'n, the relation
of which to xareTocu6aaczo is not visible to the naked eye (BEVENOT does not
analyze the anomalous form). It would appear most probable that xaOCTsero6aTO
best corresponds to nl3l, an easy corruption from MT rn31?7. TaXZiocxq, a
hapax legomenon in Greek literature, presumably means "swiftly", and might
render Vt~, which bears little or no resemblance to MT S1T1P3:l. One is tempted
with MARGOLIS simply to give up on this obscure rubric. It is probably best to
say either that it renders a liturgical direction totally different from MT, per-
haps resembling mnr1? IrH, or that its origin is in Greek liturgical usage, and it
does not render any Heb. Vorlage. It should be noted that, instead of TocXLiocs, V
reads xca0qoac; over an erasure, which may be paracousis.
1) Contra MARGOLIS (op. cit., p. 135): "Especially characteristic of our version
is the introduction from a parallel clause of a verb wanting in [MT]; a Greek
synonym is naturally chosen." The independence of Barb. indicates that Hab.
iii circulated independently of the Book of the Twelve, and the liturgical notes in
all texts of the chapter show that the circulation was liturgical. The variant. Heb.
text forms probably stem from this separate circulation.

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
 BARBERINI VERSION OF HAB. III 21

fact that LXX and Barb. have been conflated in v

on this problem. Barb. and LXX agree extensi
and 18, but in v. 18 there has been no conflation.
Barb. shows only two additions to MT:
v. 2c yvCO6a0 (= t"nlt; cf. v. 2d)
v. 10b 8t' oCUTC (= 1=)
The first may be doubtful, since yvwOaaj n may
of the verb in the next line.
Barb. has, however, fourteen omissions from MT, of which seven
are particles, prepositions, and copulae. With the remaining seven
omissions, we are dealing with more substantial lacunae. Three are
somwhat doubtful:

v. 7a 71l nnn (BEVENOT thinks that Barb. v. 6f presupposes a

corruption of the phrase 1))
v. 9a ;,'iS (The translators may have taken this as an absolute
infinitive and joined it to the following verb in translation)
v. 17c n~iot0 (Again perhaps the translators joined this to the next
for stylistic reasons, though Barb.'s tendency is rather to
expand than to condense 2))
The remaining omissions are:
v. 1 '=33n (86-407 rectify this omission in agreement with LXX)
v. 6c 177
v. 14b ,ly&,n;
v. 16b ib5s
It appears that at several points, Barb. has read different tenses
from those in MT. Two of these are changes from perfect to imper
fect:

v. 6d Tr7C?tlvwo0jovTxL. ? inP (MT intr)

v. 6e XXotco67OvTocXL = ? - ij' (doublet to the foregoing) 3)
Five times Barb. seems to have read a perfect tense for MT imper-
fect:

v. 8d (v603S = n- n (MT i:nn)

v. 9a S'y6p0y = ns (MT -nvn)
v. 16c etlSu := a (MT xKgl)

1) Cf. note 9, p. 14,; cf. also MARGOLIS, Op. cit., p. 138.

2) But cf. v. lOc, where dv,cpcovtje = ip 11n.
3) Two additional cases are probably only apparent. In v. 8a, 6pylaOt?, which
seems to imply an imperfect, is probably paracousis for LXX tpyjLCaOq;. In v
17e, exXr,Lst seems to presuppose '1t" (MT Y1T), but the whole verse is cast in
the future tense, and the translator may only have been maintaining consistency

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
22 E. M. GOOD

v. 16d arpdaX0r v = mntn' (MT tjr) 1)

v. 19b xctrcr-ae = t3t (MT otr)
Barb.'s Vorlage contained one different prepo
in MT: v. 16b cb7ro ? poxv = 'ipz (MT 5ip5). At
has read a pronominal suffix different from that i
v. 14a svra c 8uvct'isa ou = 7b;3 (MT rlftn)
v. 16b TO aro6LacTo c a =ou = l'ntt (MT ini't)
v. 16f 7toX0toivv 1 Ov Xa6v aou =? -l-" (MT 'nr
Only one Barb. reading presupposes a difference
from MT: v. 14b mtL T oaOoc[X. ocuI&V = mn
Most of the divergences (there are about twent
shown to be corruptions from MT, though severa
something completely different from MT. One su
has been identified above as a doublet to v. 6a.
word of the three presupposed by this line agrees
v. 9b bears witness to a non-Massoretic tradition o
not to a totally variant line. The verbs in v. 9c can
render what is in MT, nor is it likely that their V
rupted from Span. V. 13d is so different from MT
can only be conjectured. Yet it fits the context. A
Xnz7rol06T0a in v. 14b might point to a corruptio
normal equivalent of xs0Qo is nt:, which, if presup
non-Masoretic.
For all these disagreements, Barb. agrees with MT in the ma-
jority of readings, and no convincing pattern can be made of the
disagreements. Many of the discrepancies are added to form doublets,
many are integral to the text. The largest number occur in the very
difficult passage, w. 13c-14; in five lines there are at least twelve
divergences from MT, which range from a changed suffix (ltnan for
MT to:t) and different word division (wnr "y for MT trwl72) to
a completely variant line (v. 13d). On the basis of these observations,
it must be assumed that Barb. had as its Vorlage an early variant
tradition of Hab. iii. In the doublets, it seems to have retained
vestiges of what may have been still a second non-MT tradition.

3. Characteristics and Method of Barb.

The translator knew Hebrew quite well, and he felt entirely

1) It is interesting to note that these four verbs occupy two pairs of conse-
cutive lines. In the former case, the perfect of v. 9b may have influenced the tense
change.

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
 BARBERINT VERSION OF HILA ITT2 23

free to translate the text into good Greek. In doing so, he ha

parted several times from a strict rendering of the Hebrew ten
in the interests of good style (cf. the Greek participles for
perfect in v. 6a-b, the two examples of Cv + infinitive for perf
in v. lOa-b).
In the freedom of translation we find Barb.'s most outstan
characteristic. In several passages, the Heb. word order bas b
altered. In v. 5b, the rendering of "~9~ comes at the beginn
of the line, producing a symmetrical parallelism with v. 5a.
17b and 17f Barb. has also lent a better climax to the lines by
versed word order. A different type of changed word order is
found in v. 11a-b, where genitival constructions have been m
from words separated in the Heb. text. Thus MT has been trans
as if lisrr and ?ttt were in the construct: cpy6 -q ro ?aprp6v 'ro5
zt76aZz. The same phenomenon occurs in v. 1ib, where cply
TrS aek?'jvr would seem to render rrv and 1i'w as if they we
the construct. A more radical change has taken place in v.
where Barb. presupposes a totally variant TVorlage from MT
prepositional phrase at the end of the Heb. line comes first,
lowed by what probably was the first word of the Vorlage in
genitive. Barb. is certainly an attempt to render the sense rather
the letter, much in the manner of a Targum 1).
Another characteristic of Barb. is the use of paraphrase. M
of the paraphrastic translations are prepositional phrases:
v. 3c zU' 7rPeXL6V'C' ~kq a~T~~If;
v. 5b -r& ytX'YL ~GT v 7XeLVGV -= V0 i
v. 7b o' x0oCLxoii0v'req '& B6PPs ? erY'P
v. 10b t6v 'oaaOt6v aou 044pov - trI
v. hla ypcS -T'o )cMELnpov = ;11))?
v. 15b -o'& SxLotot {Uoc-xoc rcq o&cp'aaou -~ V,2I a1n
v. I6f TcoXetoiov -o6v Xoc6v aou -?
v. 17a o U ?vl' cocPoca& rO'V xocpnov aUT-?q S i-pn X';
v. 19a Ciscz to& L'ayjv

v. 19c -rOU' q'poC Xouq T-rcv e'ZOp&OV [IOU =? 'Tn=

At only three points has Barb. condensed:
v. 9a Ekry'pO- =? il,n nr2ny
v. 17c X"cVTco =9 rr n'p 7nm
v. 17c -?' 'Xata =?. n4t ;rfpyn

1) Cf. MARGOLIS, Op. Cit., pp. 135 f., on the Targumic character of Barb.

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
24 E. M. GOOD

Still another peculiarity of Barb. is its vocabulary.

this passage appear nowhere else in Greek translat
and one of them appears nowhere else in Greek literat
v. 4a sLauyao[a
v. 6b 'sixat')
v. 17c ~iCTro ?
v. 19d TracyXCaG (the hapax legomenon in Greek)
None of these words appears in the NT. Three of Bar
appear elsewhere only once in Greek translations of the O
~vTOqp9|xbClt, v. 10a and Wis. xii 14
ix6OaXXO, v. 17d and Sym. Cant. ii 13
cpopoc, v. 17b and Sym. Ps. lxvi 7
Two words appear twice elsewhere in Greek translations o
auO00c?Sa, v. 14b; Isa. xxiv 8 (ABS); Sym. Eccl. ix 3
scrs6uvo, v. 16c; Jer. iv 29; 1 Macc. vi 46
'E[cctatoq, vv. 10 and 15, appears elsewhere only in the LX
where it is used eight times.
Other words are characteristic of later books, such a
Wisdom, and Proverbs:
a&vc8CevuCt, v. 2e (and LXX, v. 2e), twice in LXX
elsewhere only I Esdras, II and III Macc.
XotLrcp6s, v. lla, Tobit, Wisdom, Ben Sirach, Epistle of
Symmachus, Theodotion
iopac=Xq, v. 19b, Tobit, Proverbs, Wisdom
ivaoMtTocx6, v. 16a, LXX Dan. vii 23; Aquila Ps. x 1; Sym
Ps. lviii 12; Isa. xxii 3; xxxvii 13; Jer. xlix 30 (xxx 8).
These vocabulary examples show a certain affinity to Sym
But the fact that Barb. and Symmachus coincide in only on
in this passage rules out their identification; the one agre
the rendering of 5RK?, v. 14c, by xcxaycpaxyv.
The translator thus has mastered his subject extremely
knows classical Hebrew, and never mistakes it for Aram
LXX translators have done several times in this chapter 2).
1) Compare also the similarity of translation methods between Bar
participle and a verbal form to translate two Heb. verbs joined by wa
with Symmachus' use of the same device in Exod. v 7; 4 Kings i
translation methods of Symmachus, see SWETE, Introduction to the
(Cambridge, 1914), pp. 52f. Jerome identified Symmachus' aim as "s
sequi", which was certainly the case with Barb. But the most that ca
that the Barb. translation may have been made in a circle which kne
vocabulary and methods as those of Symmachus.
2) LXX Aramaisms in this chapter are: the rendering of ,i1, v.

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
 BARBERINI VERSION OF HAB. III 25

lates literally when it suits his purpose, and h

freely when he prefers paraphrase. The translatio
exegetical rendering, though there are none of th
which are to be found throughout the Targum to
the comparative restraint in paraphrase marks th
acter of Barb. as a rather early one 1).

4. Relationships with Other Versions

Relationships with other ancient translations of
pointed out in the notes on the text of Barb.
catalogued.
a. Hexaplaric versions
v. 3a 'X ' XtS; cf. Theodotion, &rTo vo6tou
v. 5b Tr& itytra 0TO -v 7TT-r?LV)V; cf. Aquila, 7Trrvov, Symmachus
and Theodotion, Opvsov, Quinta, volucer (as reported by
Jerome)
b. Septuagintal Versions
1) Old Latin
v. 3a &Jro Xt0B6; cf. the first text of a Sorbonne Psalter with
a dual text, ab austro = Vulg.
v. 4a 8tLayacpa poT-or; cf. Augustine, Cyprian, splendor it lumen
(Cyprian lux)
v. 4c X? C?7CT'rpcXTaC T] SaVpi Tt r 8^ 6-rc &UTOU ; cf. the
insertion of this line before its LXX counterpart in all
Old Latin texts except Augustine, Jerome, and the Psalters
of the Manuale ambrosianum, of St. Germain, the second text
of the above-mentioned Sorbonne Psalter, and the Psalterium
had Aram. third masculine suffix (cpiyyo auToo5); the interpretation of 'n", v.
6b, by &-sToxy, as if it were a form of Aram. '2N, "to dissolve"; the translation
of '7 in the next line by 4, ,,strong", as if it were Talmudic 'T1S (cf. Bib. Heb.
T?); the reading of what was probably '1"D for 'IlO" in v. 13d in the Aram.
sense of "chain" (8sCTaL6); the seeming translation of '157 for MT '1S" in v.
14b as if it were Aram. T1S7, "to shake" (ateio); the rendering of tps in v.
16b by TrpoaouXq, clearly reflecting Aram. *13; the presupposition of an Aram.
infinitive form in 1nM=', v. 19d (-rou Vtx7act).
1) Cf. the insistence of CHURGIN, Targum Jonathan to the Prophets (New Haven,
1907), pp. 35 ff., that the early Targumim were more literal and less paraphrastic,
in opposition to the opinions of such scholars as ZUNZ, GEIGER, FRANKEL,
BACHER, etc.: "Literalness was insisted upon and expository rendering would
only be tolerated in difficult or poetical passages, or where the danger of a mis-
interpretation had to be averted." The Targum to the present passage is on the
whole astonishingly literal, though paraphrase and haggadic interpretation are
not by any means absent. One of the great needs of textual criticism of the OT is
a critical edition of the Targumim.

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
26 E. M. GOOD

vaticanum latinurn (referred to by BAUMSTARK as the Antipho-

narius missae 1))
v. 6c Opauao0eorsT; cf. Speculum, quassati
v. 13b TroUs xxXexxoU aou; cf. Mozarabic Breviary, the Veron
Psalter, and Verecundus, electos tuos
v. 17a ov p Txccpoc rObv xocpT7rv ocuiq; all Old Latin texts have
this reading, except that only Cyprian omits quoniam at
the beginning (= MT ,), and no Old Latin text has an
equivalent for axur-q.
v. 17e sxXstrt; cf. Mozarabic Breviary, Cyprian, deficient
v. 17f 67crapout; cf. Mozarabic Breviary, Cyprian, erunt (LXX
has urnapXoual, with which the rest of Old Latin agrees
with sunt)
v. 19b xaT,T7ro ra cf. Augustine, San Michel and St. Gall Psal-
ters, statuit
2) Palestinian Syriac 2)
v. 6b sitxcacs; this version has the equivalent of Barb. in place of
the first verb of the line, reading ot-.
v. 6c OpauOcas0LTa; Syp seems to presuppose the plural (read by
V), and cf. Wc: LOpuPrc6aov.
v. lc xaocr&a TO ypyYoq tv T 3o;Xov aou; Syp y4.~? 0oo9?or can
quite confidently be called a Barb. reading.
v. 12a zyzp06ajTn;;&jw ll, "thou dost shake" (LXX has obXtyaetS).
3) Coptic 3)
v. 3b ,uroxpoXy; 8L& oa'ckkTo;;; Achmimic, T^eTAO&c0 'AH AlA\-
ax'\.\ a, and cf. W.
v. 3c Tzv ezUnprtzLav TGq a 86oE auTo6; Achmimic, Anc<AIp [n]
rinqcr; Sahidic, x nca r lnte TOOv

1) "Aramaischer Einfluss im altlateinischen Text von Habakkuk 3", Oriens

Christianus, 3te Serie, 6 (1931), pp. 163-181.
2) The Palestinian Syriac version of this passage exists only in the Malkite
horologion, published by Matthew BLACK: A Christian Palestinian Syriac Horo-
logion (Cambridge, 1954). The Syriac text is on pp. 167-170, and Prof. BLACK'S
critical notes are on pp. 45 f.
3) The Coptic versions of the Book of the Twelve have been minutely studied
by Willem GROSSOUW: The Coptic Versions of the Minor Prophets (Rome, 1938),
and some corrections and additions have been made by Joseph ZIEGLER, "Bei-
trage zur koptischen Dodekapropheton-Ubersetzung", Biblica 25 (1944), pp.
105-142. The only difficulty with GRossouw's very useful work is that he too
frequently gives the readings in Latin translation rather than in Coptic. The
Achmimic version, which bears the closest relation to Barb., is extant only in
the Codex Rainerianus, Kopt. 11000 of the Vienna Nationalbibliothek. It was
first published by Carl WESSELY, Duodecim prophetarum minorum versionis Ach-

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
 BARBERINI VERSION OF HAB. III 27

v. 4a Sccayya?o6c (po?To6 TalV aorTjz;; Achmimic, o

oTAC-I HAL[*])]ue req.
v. 5b zro pteyyCor v ?Ttv&v; cf. Achmimic's addition, i51
HNA'&xTC; Sahidic, nHCI nPa'AT?; Sahidicc'i, Hcti
p-It OA *ATC
v. 8a opyLaON0; Achmimic, HAKrtHOTruC
v. l lc Cov PoXL&ov aou; Achmimic, nHu KCAT?; Sahidic
COTe

v. 12b aXo7AoL;; Achmimic, i-rh pi 1)

v. 13b TOU; exXsxTou; oaou; cf. Sahidic"'t, H1rCq
v. 13d xacrc6ovrocL; cf. Achmimic, Alielri
v. 16b TOV 76o[IaToo6; ou; cf. Achmimic, AHA
2*p^ nK*2ppA 2)
v. 17d 3oTvrvyv; cf. Achmimic, 6pe 3)
v. 17e exXzL4st. x ?%pavpoa; tp6opa3a; Ach
A^TcTClq ^&a& *pn TTIrq. e-pc
4) Agreements with LXX MSS 4)
v. 1 omit TOV 7xpoyp'Tou, L-22-711-613 CyrP
v. 2b x6ptl; S L"-49-764-407 C'-68-239 Cyr
omits the word)
v. 3b p.?Tra0oX 8taaktLoaXaTro;, W
v. 4b uoCpZLt accur, L'-86mg Th Tht
v. 4c i?X~ ?7T?TrpL-rpTaL ot vaprE Tt ao6 OC
239 before LXX 4c, with the variant ?6rnpltxr
v. 5b ra ptLyTa TC 'V rvTT?eVzOV; cf. We, 7?Tr[
mimicae: Codex Rainerianus (Leipzig, 1915) as Vol. 16 of
graphie und Papyruskunde." A better edition was produ
achmimische Version der Zwolf kleinen Propheten (Haunia
has been used here. TILL dated the MS, following KRALL
century A.D., and CRUM, in his review of TILL'S work
calls it "the fifth (or? fourth) century parchment volu
which are extant only in fragments, are published in a gr
which see GRossouw.

1) CRUM, op. cit., p. 210, objects mildly to this identification, saying, "riA w
pi- would be strange."
2) GROSSOUW (op. cit., p. 72) wrongly translates this a voce mea, rather tha
a voce tua. This is the only occurrence of the phrase K<i pwoO in the Book of
Twelve, and its equivalent remains doubtful. The line is perhaps not deriv
from Barb., but it is somewhat similar.
3) ZIEGLER would emend to * pp, which is used for PpCouc in Joel i 16 (op. c
p. 118). Cf. also Bohairic 5pp in our passage.
4) The sigla and MS groups are those of ZIEGLER, Duodecim prophetae (G
tingen, 1943), by far the most adequate edition of the Twelve in the LXX.
* (See printer's note on p. 19).

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
28 E. M. GOOD

v. 8a 6pyt.06q, attested by most LXX MSS

v. 17e ix [i&va8paq; cf. W" [xTor] pavSpSac;
c. Targum
v. 3a (&rT Xt4oS; cf. Targ. K?7'l17
v. 3c zTYV ?rUVpkr?7iltv zT( 861or c5UTOU; cf. Targ. n;p' l,tI
d. Peshitta
v. 3c -%v Sr7cp?retlav TV 860q oCU'TOU; cf. Pesh. SJ'L! 0Q'.
v. 5b Tz& ri Ytrra Trcv ,rzLVvCV; cf. Pesh. {;q
v. 9b so6pc0roTr ; cf. Pesh. e mio
v. 14b 'roru 7zoE8oTaS ?1 T-7 a~uOCSdqt auTo' V; cf. Pesh. a_sl.l!

v. 16b .xi6 Tq (; poVi OU aT6OTooi caTO U; cf. Pesh. J\\oo oA.

Lto .a ?, agreeing with MT "n&t against Barb.'s presuppo-
sed r"Tntv
e. Vulgate
v. 4c CxXZ; Vulg. ibi (MT tthl)

These relationships present somewhat ambiguous evidence. The

closest relationships to Barb. are in the Coptic versions, particularly
the Achmimic, which was probably made in about the middle of the
third century A.D. 1) Otherwise, we have several contacts in Old
Latin texts of North African provenance (Cyprian, Augustine; and
Mozarabic Breviary probably represents a somewhat independent
North African text-type), some in the Palestinian Syriac, which may
antedate the fifth century A.D., 2) and a few with Peshitta.

5. Provenance and Date of Barb.

These relationships with other versions suggest that the p

venance of Barb. is Egypt. Egypt would be the most likely ce
from which a textual and exegetical tradition would spread b
west to North African centers and east to Syrian ones. And the
lationship with the Coptic versions are perhaps the earliest attes
Egypt is also suggested by the fact that the translator of B
knew classical Hebrew uncorrupted by Aramaic and by the f
that his Greek is very good. Alexandria, of course, was the site of
most cosmopolitan and Hellenized Jewish community. And

1) GRossoUW, op. cit., p. 122, thinks that it was not later than the second
of the century.
2) So BLACK, op. cit., p. 21, on the basis of a few traces in the version as a wh
of pre-Rabbulan Syriac readings.

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
 BARBERINI VERSION OF HAB. III 29

Alexandria such a Hebrew textual tradition as tha

Barb. could well have been extant 1). The very
dence in Barb. may point to Egypt, but this e
as to bear very little weight.
The problem of date is more complicated. The t
is the first half of the third century A.D., as show
Cyprian and the Achmimic and Sahidic versio
from the second half of that century. But this ter
be carried further back than that in view of the fact that Barb. and
LXX have been conflated in v. 2 of our chapter. The presence of
this conflation in all LXX MSS is indication that it must have been
made quite early, though the precise date cannot be established. It
seems unlikely that it could have been done any later than the early
second century A.D. The terminus a quo is inpossible to determine.
THACKERAY suggested a date early in the second century B.C. 2).
His theory was that Barb. was the first Greek translation of our
passage, made for the purposes of the Pentecost lection, and that
the present LXX displaced the more ancient text. This is a very
difficult position to support in detail. It seems quite likely that Barb.
represents a translation made for liturgical purposes. And it is further-
more probable that Barb. is a very early translation, for the measure of
its divergence from the Massoretic text-type is considerably more
even than LXX.
We might even argue that Barb. presupposes an edition of th
poem in Hebrew which was quite independent of the rest of H
bakkuk and the Book of the Twelve. The fixation of the Hebrew
text was not by any means complete. Of course, it has been main-
tained by many scholars that the Hebrew text was not uniform until
the time of the ben Chayyim text 3). But it seems probable that the
process of establishment of a uniform text began with the initiation

1) Cf. the work of Peter KATZ: Philo's Bible: The Aberrant Text of Bible Quo-
tations in some Philonic Writings and its Place in the Textual History of the Greek
Bible (Cambridge, 1950). The character of Philo's text merely suggests the pos-
sibility of such a non-MT textual tradition in Alexandria. Since Philo does not
quote from Hab. iii, we have no evidence to link his text with that of Barb.
2) JTS 12 (1910/11), p. 213; and cf. GROSSOUW, op. cit., p. 125; BEVENOT, Op. Cit.,
passim; and SCIINEIDER, "Die biblischen Oden im christlichen Altertum",Biblica
30 (1949), p. 31, all of whom agree with THACKERAY on the chronological pri-
ority of Barb. over LXX, though they do not all follow his early date.
3) Notably J. M. P. SMITH, "Studies of the Massoretes", JAOS 44 (1927/28),
pp. 208 f. Cf., for full discussion of the various views, ROBERTS, The Old Testament
Text and I'ersions (Cardiff, 1951), pp. 23-29.

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms
30 GOOD, BARBERINI VERSION HAB. III

of translations, and the process was furthered in the Jewish tra-

dition by the acceptance of LXX as the Christian Bible. Important
aid in the development of a uniform Hebrew text was given by
Aqiba, who simply assumed dogmatically a pure, uniform text. The
kind of close study which Aqiba and his disciples carried on shows
this presumption, and the work of Sopherim and Massoretes not
only was the attempt to bring about textual conformity in the He-
brew Bible but also operated on the presumption of uniformity.
Barb.'s divergence from MT indicates that this process was not well
started-if it was started at all-at the time the version was made. A
date late in the first century A.D. might be suggested on the basis
of this observation 1).
It is not unimportant to notice that the existence of this non-LXX
version lends weight to KAHLE'S "Greek Targum" theory of trans-
lation 2), since it is highly likely that such was its purpose. The Targumic
character of the version and its probable Jewishness give further
support to KAHLE'S argument, and it is curious that KAHLE nowhere
gives any indication that he knows of Barb.'s existence. At the same
time, the writer would disclaim the KAHLE theory insofar as it
applies to the LXX itself in Hab. iii. The clear evidence of variants
there is that there was a "proto-LXX." The conflation of the two
versions in v. 2 and sporadic LXX variants agreeing with Barb.
show only a very limited contact between them, and no pattern of
"recension-family" agreements with Barb. emerges from LXX
variants.

1) Cf. above, note 1, p. (17). The suggestion that Tou ?XxXTOUqs aou in v. 13b
represents an anti-Christian polemic might indicate a time between the two
Jewish revolts when Jewish anti-Christian sentiment was increasing. But this
consideration cannot be decisive.
2) See especially The Cairo Genita (London, 1947) and Masoreten des Westen
II (1930).

This content downloaded from 87.241.14.18 on Fri, 17 Apr 2020 15:59:37 UTC
All use subject to https://about.jstor.org/terms