[ back]

[© Wolfgang Schulze 2003.

No parts of the present text may be copied, reproduced or distributed or without prior permission by the author.]


Wolfgang Schulze

01.08.2001 / Revised Version 07.09.2003




The Language of the Caucasian Albanian (Aluan) Palimpsest from Mt. Sinai

and of the ‘Caucasian Albanian’ inscriptions

 A tentative interpretation of 2 Cor 11,25-27 (specimen of the Caucasian Albanian (Aluan) Lectionary) and of the Aluan inscriptions

[Based on the transliteration by Zaza Aleksidze, re-read and corrected by © Wolfgang Schulze and © Jost Gippert 2003]

[Select UNICODE!] Comments: Please mail to W. Schulze

See http://titus.fkidg1.uni-frankfurt.de/armazi/sinai/2kor.htm#start for the documentation of the original text, Zaza Aleksidze’s interpretation, and for background information.

See /Uog.html for a description of Udi as the descendant of the language of the Caucasian Albanians (or: Aluans).

http://www.lrz.uni-muenchen.de/FGU.htm for details on the Udi


Udi represents an endangered language of the Southeast Caucasian (Lezgian) language family. Currently, it is spoken by some 4.000 people in the village of Nizh (ni%z^) in Northern Azerbaijan as well as by some 50 people in the neighboring village of Oguz (formerly Vartashen). In addition, a significant number of Udi speakers (~ 200) dwell in the village of Okt’omberi in Eastern Georgia, a settlement founded by Vartashen emigrants in 1922. Since long, Udi has met the interest of both linguists and historians. On the one hand, the language is marked for a number of typologically salient features (see Schulze 1982, 2000, Harris 2002, Schulze (forthcoming)). Historians usually consider the speakers of Udi as the descendants of one of the peoples of Caucasian Albania, a ‘kingdom’ located in the northern and western regions of now Azerbaijan (100 BC – 700 AD).


A famous passage in the Armenian patmowt`iwn (ašxarhi) ałowanic  (History of the Albanians) by Movsēs Kałankatuac`i (or Dasxowranc`i; 7th century (?)) tells us that the Armenian scribe, monk and (later) missionary Mesrob Mašt`oc` (362-440) has “created with the help [of the bischop Ananian and the translator Benjamin] an alphabet for the guttural, harsh, barbarious, and rough language of the Gargarac`ik` (Pat.Ał. Book II, 3, compare Dowsett 1961:69). The Gargarac`ik` represented one of the peoples of the kingdom of Albania the name of which is already attested in Strabo XI,5,1 and which can be associated to the Armenian toponym daštn Gargarac`owc`, a region southeast of the central part of the Kura river (compare the contemporary river name Gargar, a tributary to the Araxes). Most likely, the Gargarac`ik` whose habitat was located to the east of the Aluan province Utik` played a crucial role in the state’s administration at least by the time of conversion to the Christian faith. Although the ‘Albanian’ state started to disintegrate soon after 705, the Aluan script seemed to have been in continuous use until at least the 12th century. For instance, the Kilikean historian Haython (Hethum), a nephew of the Kilikean king Hethum I (1226-1269), reported in 1307: “Literas habent Armenicas, et alias etiam, quae dicuntur Haloën” (Haythoni Armenii historia orientalis, quae eadem et De Tartaris inscribitur, Coloniae Brand. 1671:9). The existence of an Aluan alphabet has been confirmed by two (re-copied, in parts corrupt) alphabet lists that have survived in medieval manuscripts (now kept in the Matenadaran museum, Erevan; M 7117, f 142 and M 3124, see Abuladze 1938:70, Kurdian 1956). In addition, a small number of inscriptions on candleholders, roofing tiles and on a pedestal found since 1947 in Central and Northern Azerbaijan (see below) illustrate that the Aluan alphabet had in fact been in use.


Until 1996, little had been known about the language used in connection with the Aluan alphabet. The earliest word said to be ‘Albanian’ or Aluan documented so far stems from the fragment of a lexical list ascribed to a certain Heracleides. This list is included in the so-called Oxyrhynchis Papyri (100-200 AD). The relevant passage reads: μιληχ γενειον υπο Αλβανιων των ομορουντω[ν] (‚milēkh – beard according to the neighboring Albanians’, Pap. Oxy. 180265 (Grenfell & Hunt 1922:158), I thank Bill Judge for this reference). Although the word at issue has a Lezgian look’ (compare Archi muz^ur, Tabasaran (Dübek) midz^ri, Rutul me^c^ri, Lezgi c^iri (> Kryts dz^iri), Tsakhur muc ri, Budukh mic^er, Khinalug mic^:äs^), it is difficult to relate it to any of the candidate languages (in Udi, the Lezgian term has been replaced by k’adz^ux). A list of so-called Albanian month names surviving in a number of medieval manuscripts gave the first clue to the language of the Aluan. Basically, we have to deal with the manuscript ‘Paris Arm 114’ (Brosset 1832), a list of month names compiled by par Anania Širakac`i, variants which occur for instance in manuscripts by Hovhannēs Imastaser (~ 12th century, Armenian) et Sulxan Saba Orbeliani (18th century, Georgian), see Schulze 1982:284-5 and (more importantly) Gippert 1987 for details. Obviously, at least parts of the month names are clearly related to Udi. As a result, the long-standing hypothesis has emerged according to which the language of the Aluan people represents an older variant of Udi.


This hypothesis has been supported by a number of co-arguments. For instance, the Udi are the only Christian group in Azerbaijan. According to their own tradition, they once had been part of the Albanian Church which had been abolished by Tsarist authorities in 1836 (re-established in 2003). In addition, names obviously related to the ethnonym udi had been constantly referred to by ancient sources when speaking of the Caucasian Albanian region. This region had been known in Classical times under the name ’Αλβανία or ’Αλβανίς, in the Armenian tradition the term Աղուանք (ałowank`) had been used (Georgian რანი (rani), probably derived from Arabic الران (ar-rānu), which again had been borrowed from the Armenian toponym Առամ (ar̄an)). Caucasian Albania represented a rather heterogenous ‘state’ that had been christianized as early as the 2nd  or 3rd century (according to the tradition by Ełišē (Eleusius), said to be ordained by James, the brother of Jesus, see Mämmädova 2003). One of the provinces of Aluan had been Uti, the population of which is referred to by the name Udini (or Utidorsi) in Latin sources, and by the name Οὐίτιοι in Greek sources. In Armenian, the terms Ուտիք (owtik`) or Ուտիացիք (owtiac¢ik`)) had been used. The province of Owtik` was located between the middle course of the river Kura and the Mountain Qarabakh region, thus south of the actual habitat of the contemporary Udi speakers. Most likely, the inhabitants of Owtik` at least in parts spoke a language related to or equal to that of the Gargarac`ik`, mentioned above.


It should be noted, however, that none of the three names (Udini ~ Οὐίτιοι ~ Owtik`; ’Αλβανία ~ Ałowank`; Γαργαροί ~ Gargar(ac`i)k`) can be safely etymologized with the help of contemporary Udi. There is a slight chance to relate the term udi (also used as a self-denomination of the contemporary Udis) to the ethnonym qūtīm which labels a gentile group having ruled over Central and Southern Mesopotamia (2200-2100) and said to stem from the northern regions of the Zagros mountains. Urartian sources mention a river Uduri said to be located at the border of the land Etiu (e.g. Meščaninov 1978:319), and it may well be that one of the two ethnonyms can be equaled to the term udi. The term ’Αλβανία ~ Ałowank` probably reflects a form *aluan which is sometimes paralleled to both the name of a village in the Shah-Dagh mountains (Alpan) and to the name of a pre-Islamic deity in Lezgistan (Alpan). However, this proposal neglects important historical facts and should be taken with great caution.


In sum, both direct and indirect evidence suggest that the ‘major’ language of Aluan (i.e. the language of the Gargar(ac`i)k`) must have been an early variant of Udi. This assumption saw confirmation in 1987, when the Georgian scientist Zaza Aleksidze discovered a palimspest stored in the Mt. Sinai monastery.


In 1996, the Georgian scientist Zaza Aleksidze – while doing documentary work in the St. Catherine monastery on Mt. Sinai – discovered two Georgian palimpsest manuscripts (conventionally labelled N/Sin-13 or M13 and N/Sin-55 or M55) that contain in their lower, heavily washed layer texts in Albanian script (see Aleksidze & Mahé 1997, 2002 for a detailed presentation of the manuscripts and a preliminary discussion of the language of the lower layers, http://armazi.uni-frankfurt.de/armaz3.htm for a presentation of the Sinai project). Meanwhile, the pioneering work of Aleksidze has been continued by Jost Gippert (Frankfurt) and Wolfgang Schulze (Munich). For the time being, nearly the totality of the readable folios of both manuscripts has been deciphered and interpreted. Aleksidze’s assumption that we have to deal with a rather old lectionary used in the Holy Service turned out to be correct. For copyright reasons, I cannot go into the details of the whole corpus (see the projected publication in Aleksidze & Gippert & Mahé & Schulze (forthcoming)). Hence, I have to restrict myself to more general remarks.


In sum, the two manuscripts consist of roughly 180 folios (recto/verso), in parts heavily distorted and only fragmentary. They show the Aluan text in horizontal lines crossed by the upper layer of Georgian text in vertical lines (see http://armazi.uni-frankfurt.de/sinai/albanica/m13.htm for images). The Aluan text is strongly washed out. Its characters have (in major parts) merged with the Georgian letters of the upper layer. The original Albanian text was written in two columns (22 to 23 lines per page) which 15 to 20 characters per line. In addition, smaller characters were used to add commentaries relevant for the use of the lectionary in the Holy Service. At the end of M13 n63, the scribe seems to have added a ‘personal note’.


The bulk of the lectionary is preserved in M13, whereas M55 is much smaller and more fragmentary in nature. It is not quite clear whether both manuscripts had been written at the same time. Perhaps, M13 is older stemming from the 5th or 6th century, whereas M55 has been written in the 8th century (see Aleksidze 2002). Nevertheless, it comes clear that both manuscripts originally represented a single ‘book’ which contained passages from the New Testament as well as at least one passage from the Old Testament.



2. The Specimen


The following passage from Sinai M13 n75 (Folio 76r-77v, column B) helps to illustrate the language of the Palimpsest (original reading of lines 6-22 by Zaza Aleksidze; re-read and corrected by W. Schulze and J. Gippert):


2.1 Linear version




owq’abiyayza(x) : Xib












Laq’mox avelomc^ar












zak’owg^oxoc : mar





















sownxc’ayax : mar






is^ebaxoc : marak’e












c^ar : bowsinown(i)g^i


2.2 Glossed version


This passage contains a translation of 2 Cor 11,[2]5-2[7]. A linguistic interpretation of (16) is given below. Note that the glosses are derived from the system applied to Udi by Schulze (forthcoming). A preliminary translation of 2 Cor 11,26-27 had been prepared by Zaza Aleksidze. Here, a revised interpretation is given on the basis of the corrections and additions proposed by W. Schulze and J. Gippert):



[…1]       xib-[2]om   n%az^iz^-ac-E               h-E 

              three-coll   shipwreck-lv:pass-perf  lv-perf

              Thrice I suffered shipwreck’


[3]          zow  g^i    own   s^ow   bAwg^a      [4]  (y)~i-g^-ox       marg^(i)-zow-h-E

              I       day    and   night    middle-dat        depth(?):dat2   suffering-1sg-lv-perf2 

              A night and a day I have been in the deep.’



[5]          Laq’-m-ox      avel-om    c^ar

              way-pl-dat2   much-coll fold

              Often on the roads’


[6]          marak’esown-owx  t’(=k ?)owr-[7]m-oxoc

              danger-dat2            river-pl-abl

              in danger of rivers’


[8]          marak’esown-owx    aba[9]zak’-owg^-oxoc

              danger-dat2               thief-pl-abl

              in danger of thieves’

              mar[10]ak’esown-owx   C’inowx-[11]oc

              danger-dat2                   compatriot-abl

              in danger of the compatriots’


              ma(r)ak’esown-owx   [12] het’anos-owg^-oxoc

              danger-dat2                       gentile-pl-abl

              in danger of the gentils’


[13]        marak’esown-owx   kala[14]k-a

              danger-dat2             town-dat

              in danger in the town’


              marak’esown-owx  [15]  k’%aban-a

              danger-dat2                    desert-dat

              ‘in danger in the desert’


              marak’e[16]sown-owx  c’ayax

              danger-dat2                   sea:dat2

              In danger in the sea’


              mar[17]ak’esownowx a[c/]pE   [18] is^eb-axoc

              danger-dat2                   false           brethren-abl

              in danger of false brethren’


<27>      marak’e[19]sown-owg^-on  own   borz[20]own-owg^[-on]

              danger-pl-erg                       and     labor-pl-erg

              with dangers and labors’


              nowg^owr  [21] bowr-es[own-en    av]el-om    [22] c^ar

              wake                 stand-masd-erg     much-coll         fold

              in watches often’


              bowsin   own   ig^(e)

              hungry   and   thirs[ty]

              in hunger and thirst…’



3. Analysis




depth, deep’, dat2

Udi dat2 -ox




Udi apc^i liar’



thief (pl., abl.)

Armenian abazak, Udi pl. -ux, abl. -oxo



much’ (ordinal form)

Udi ordinal -un < *-um



in the middle’

Udi be^%g^, dat. -a



load (pl., erg.)

Udi plural -ux, erg. on



hunger’ (instr.)

Udi busa ‘hungry’, erg./instr. -in



stand (masd., erg.)

Udi masd. -esun, erg. en



sea’ (dat2)

Arm. cov sea’ (?), Udi dat2 -ax








Udi abl. -uxo




Udi g^i day’



be’ (perf2)

Udi perf2 ey



gentile (pl., abl.)

(Greek >) Arm. hetanos gentile’, Udi pl. -ux, abl. oxo







brethren (pl. tant., abl.)

Udi abl. axo



open field, desert’ (dat.)

Udi k’%ava%n wilderness, open field’, dat. -a



city’ (dat.)

Arm. k`alak` , Udi dat. a



way’ (pl, dat2)

Udi yaq’, pl. -m-, dat2 ox



suffer-see’ (masd., dat2)

Udi ak’sun to see’, dat2 ux



suffer-see (masd., pl. erg.)

Udi ak’sun to see’, pl. -ug^-, erg. -on



suffering’ (1sg, perf2)

Udi 1sg -zu, perf2 -ey



shipwreck’ (mp, perf2)

Udi mp:past -ac-, perf2 ey




Udi mog^or awake’




Udi q’a-n and’




Udi s^u



river’ (pl, abl)

Udi kur, pl. -m-, abl. –oxoc



three (coll., ord.)

Udi xib, ord. -un < *-um




Udi zu I’





The Caucasian Albanian’ (Aluan) Inscriptions

(© Wolfgang Schulze 2003)


There exists a small corpus of so-called Caucasian Albanian or Aluan inscriptions the most famous of which is the Mingečaur inscription found in 1949 during excavations in the Mingečaur region in Central Azerbaijan (see . Although we cannot exclude the possibility that one or two of the (often fragmentary) inscriptions are fakes, we can still maintain that the major part of this corpus is related to and stems from the Old Udi period. Bascially, we have to deal with three types of inscriptions: a) a longer, running text on a pedestal; b) short texts on candleholders and roofing tiles, c) parts of Aluan alphabet lists. None of the texts has been safely read and interpreted so far. Nevertheless, those parts that are open to a linguistic interpretation clearly show that the underlying language is a variant of Old Udi. The following documentation of the inscriptions does not aim at a full interpretation. Rather, I will refer to those parts that evince an Udi origin (see Murav’ev 1981 for a description of the corpus).


T 1 (= Mingec^aur Pedestal, serving to carry a cross (Schulze) or throne (Gippert)) [ca. 60 x 60 cm]; Probably 7th century AD.

See Gippert (in press) for the most recent and most detailed analysis.


1          (q’)iyas   BE            be(s)(i)(n)?o(l)o     arah/c^Ene             ei/n

            ?:dat3     God:gen   ?:gen           LOC   verb:lv:perf2:3sg     ?

            For the X of God LOC  X  placed(?)’


2          h/c^Al   yE  owsena    xosroo(w)_

            ?           27   year:dat  Khosrow[:gen]

            […] in the year 27 of Khosrow’


3          _________serb[aun]_______


            [……] firs[t……….]’


4a        __Aw/s. h/c^os/b/%  (i)(n=p’?)isk’ap’osen    bi

             PN                                      bischop:erg          make:past


4b                                yayn



The present reading deviates in minor parts from the up to now most comprehensive and most reliable interpretation of the Mingec^aur inscription (Gippert (in press)) which also aims at situating the contents of the inscription into the clerical history of Albania. Here, I cannot discuss in details Gippert’s highly promissing and methodologically well-founded approach. Nevertheless, the reader should note that Gippert’s analysis for the first suggests an interpretation that seems to be coherent with both historical data and the findings related to the language of the Palimpsest. The following segments of the Mingec^aur inscription can be safely related to Udi or to the language of the Palimpsest:


            q’iyas (Gippert: miyas)              Obviously the now lost Old Udi -s-Dative (‘dat3)

            BE                                           Abbreviation of ‘god’ or ‘lord’ (= Palimpsest)

            -hEne (?)                                  = Palimpsest h-E-ne (be-perf2-3sg:foc)

            owsena                                     = Udi usen-a in the year’

            [s]er[b]-                                    = Palimpsest serbaown ‘first’

            -en                                           = Udi ergative -en

            biyay                                       = Palimpsest biyay (do:past)


T 2 (Candleholder, Mingečaur) [8 x 5 x 5 cm]

(Trever 1959:Tabl.35, new reading © W. Schulze 2003)





















b~E et’

















The four sides of the candleholder are not fully aligned. Hence, the restoration of the original lines is somewhat problematic. Nevertheless, a possible reading is:

            za     yog     gokar(e)XE             naibow   b~E          et’owX          be(c)e

I:dat   ?        be=ungodly:perf2   servant  God:gen   this(?):dat2   beg:perf(?)

            g^ahak’  hAwk’e  q’a(g^).(x)   biyay 

?             joy:3sg  ?                  make:part:past


The following elements can be identified:

            za                                = Udi za I:dat

gokar(e)XE                   = Palimpsest gokarXE (Perf2), derived from Pal. karXesown

to save’, meaning of the preverbially

marked form probably ‘ungodly’ or ‘humble’ (< ‘not saved’ ?).

            naibow                         = Palimpsest naibow servant, slave’

            b~E                             Abbreviation meaning god’ or ‘lord’ (genitive or ergative)

            hAwk’                           = Palimpsest hAwk’ joy’

            biyay                            Palimpsest biyay (do:past)


T 3 (Fragment of candleholder (?), Mingečaur) [16 x 4(,5) cm]

(Murav’ev 1981:275, new reading © W. Schulze 2003)


1(?)      zow   va            ba(l)a        oa[_____?]

I          you:sg    do:part:fut   ?

2(?)      ……biya(y) [_____?]


3(?)      ?iye   bowq’ana [____?]

.       love

 Old Udi segments are:

            zow                  = Udi zu I’

            va                    = Udi va you:sg:dat

bala                  = Udi/Palimpsest b-ala do-fut2

biyay                = Palimpsest biyay (do:past)

bowq’ana          = Palimpsest bowq’ana beloved’


 T 4 (Candleholder, Mingečaur) [18 x 11 x 10 cm]

(Murav’ev 1981:279, new reading © W. Schulze 2003)

zow  ki(W)pe                

I         burn(?):lv:perf


The meaning of ki(W)pe [phonetically ki(dz)pe] is obscure. Obviously, we have to deal with a simple’ perfect (-e) added to the light verb p-. The initial form zow corresponds to Udi zu I’.


T 5 (Roofing tile (?), Mingečaur) [10 x 10,5 cm]

(Murav’ev 1981:273, new reading © W. Schulze 2003)


1          zow m[_________]

I …

 2         bAwg^a[_______ ]

in=midth …

 3         h~k’e zow[______]

because I…

 4         (b). hel(i)[_______]

[do] soul:gen …

 (x)       [_______________]


This fragment shows the following correspondences with (Old) Udi:

            zow                  = Udi zu I’

            bAwg^a             Palimpsest bawg^a ‘in midth’ (Udi be^%g^  ‘middle’)

            h~k’e                = Palimpsest h~k’e, an abbreviation meaning because’.

            hel                   = Palimpsest hel (> Udi (pl.tant.) el-mux) ‘soul, spirit’


T 6 (Roofing tile (?), Mingečaur) [16 x 4 cm]

(Murav’ev 1981:281, new reading © W. Schulze 2003)


mana (k’)?[____]

[Personal name?]


The interpretation of this passage remains unclear.


T 7 (Candleholder, Mingečaur) [11 x 7 x 7 cm]

(Murav’ev 1981:277, new reading © W. Schulze 2003)














Textfeld: i(s)i



Textfeld:     g^ar


This inscription contains the first 10 letters of the Aluan alphabet. In addition, two words appear in vertical lines. i(s)i remains unclear, whereas g&ar undoubtedly  means ‘son, child’ (= Udi/Palimpsest g^ar).


T 8 [Tablet, Verxnyj Labkomaxi) [10 x 5 cm]

(Alphabet; Murav’ev 1981:283; perhaps a fake)





Abuladze, Il’ja 1938. K otkrytiju alfavita kavkazskix albancev. EIMK moambe 6:69-71.

Aleksidze, Zaza & Jean-Pierre Mahé 1997. Découverte d’un texte albanien: une langue ancienne du Caucase retrouvée. CRAI 1997:517-532.

Aleksidze, Zaza & Jean-Pierre Mahé 2002. Le déchiffrement de l’écriture des albaniens du Caucase. Comptes-rendue de l’Académie des Inscription, Paris XXX.

Aleksidze, Zaza 1998-2000. Remarques sur le déchriffrement de l’écriture albano-caucasienne. RArm 27:423-428.

Aleksidze, Zaza & Jean-Pierre Mahé & Jost Gippert & Wolfgang Schulze (in preparation). The Caucasian-Albanian Palympsest from Mt. Sinai. Edition and interpretation. Monumenta Paleographica Medii Aevi. Turnhout: Brepols [prévue].

Annasian, H.S. 1969. Une mise au point relative à l’albanie caucasienne. REArm VI :299-330.

Dirr, Adolf 1904. Grammatika udinskogo jazyka. SMOMPK XXXIII:1-101.

Dowsett, C.J.F. 1961. The History of the Caucasian Albanians by Movsēs Dasxuranc¢i. London: OUP.

Gippert, Jost 1987. Old Armenian and Caucasian Calendar Systems [III]: The Albanian Month Names. AAL 9:35-46.

Gippert, Jost (in press). Ein iranischer Personenname in kaukasisch-albanischer Nebenüberlieferung. Festschriftbeitrag.

Grenfell, Bernard P. & Arthur S. Hunt 1922. The Oxyrhynchus Papyri, part XV. London: Egypt Exploration Society.

Harris, Alice 2002. Endoclitics and the Origins of Udi Morphosyntax. OXford. OUP.

Hewsen, Robert H. 1964. On the Alphabet of the Caucasian Albanians. REArm I:427-432.

Karamiantz, N. 1886. Einundzwanzig Buchstaben eines verlorenen Alphabets. ZDMG XL:315-319.

Kurdian, H. 1956. The newly discovered alphabet of Caucasian Albanians. JRAS 1956:81-83.

Mämmädova, Färidä 2003. Qafqaz albaniyasında xristianlıq. Bakı: NHE.

Mes^c^aninov, I.I. 1978. Annotirovannyj slovar’ uratskogo (viajskogo) jazyka. Leningrad: Nauka.

Murav’ev, S.N. 1981. Tri ėtjuda o kavkazsko-albanskoj pismennosti. EIKJ VIII:222-325.

Schulze, Wolfgang 1982. Die Sprache der Uden in Nordazerbaidžan. Wiesbaden: Harrasssowitz.

Schulze, Wolfgang 2001. The Udi Gospels. Annotated Text, Etymological index, Lemmatized Concordance. Munich: Lincom.

Schulze, Wolfgang 2002. The loan layers of Udi: Armenian. W. Bublitz et al. (Hrsgg.). Philology, Typology and Language Structure. FS Winfried Boeder, 211-223. Frankfurt a.M.: Lang.

Schulze, Wolfgang (in press). Das Alte im Neuen: Sprachliche Überlebensstrategien im Ostkaukasus. P. Schrijver & P.-A. Mumm (Hrsgg.). Sprachtod und Sprachgeburt. Schriftenreihe des Zentrums für Historische Sprachwissenschaft, Univ. München.

Schulze, Wolfgang (in press). Towards a History of Udi. G. Authier & B. Balci (eds.). Papers of the IFEA Round Table on the languages of the Caucasus (Istanbul 2003).

Schulze, Wolfgang (forthcoming). A Functional Grammar of Udi.

Trever, K.V. 1959. Oc^erki po istorii i kul’ture kavkazskoy Albanii. Moskva & Leningrad: Izd. AN SSSR.