Iranica

By H. W. BAILKY
1. Khazar

IN aIranian Studies [I], BSOS vi (1933), 945 ff., I commented briefly-upon

text of the Batman Yast (Vahman yast or yasn), in which certain ethnical
names occurred. It was desired then chiefly to point out the name of the
Karmir yxfm " Red Chionitae " beside the Spet y%fdn " White Chionitae ",
and the forms of the name of Sogdiana. I did not, however, succeed in inter-
preting the whole passage satisfactorily, and cafter ten years I venture once
more to call attention to it. To the bibliography given in BSOS vi 9451 should
add here the edition of the Bahman Yast by Anklesaria, The Zand-i Vohuman
Yasn 1919 (which came into my hands after I had written the'paper); B. N.
Dhabhar, The Persian Rivayats of Hormazyar Framarz, 1932, which contains
on p. 465 a translation of the Parsi-Persian text to be found in M. R. Unvala's
. Ddrdb Hormazydr's Eivdyat ii 91 (1932); Markwart, Provincial Cities of
Eranshahr, p. 69.
Here I wish to reconsider one phrase of this passage. It is preserved in
three recensions as follows :—
1. Pahlavi text.

2. Pazand text (Pdzend Texts p. 344.7).

6un . hayun . turk . azarat . afS
3. Parsi-Persian text.

with a variant «i5 j [}y>- for the first two words.

As to the relationship of these three texts, although all three go back to
the one original, evidently the Parsi-Persian does not come directly from the
Pazand we have. This Pazand in turn is not from the extant Pahlavi text
which is longer by a whole phrase. I propose now to offer an explanation of
the two Pahlavi words J)tfy and y$y» different from my former interpreta-
tion. I would read now the Pahlavi as follows :—
cegon xyon turk ut xazar ut tujrit cegon averak<ik ut> kofddr
That would mean " who are the Chionitae, Turks, Khazars, Tibetans, who
•are desert-dwellers and mountaineers". To justify this rendering it may be
noted:—
1. J)f£y xazar> I na(
i previously read this by an emendation to 3
VOL. XI. PAST 1. 1
 H. W. BAILEY—

(p. 946) as hftl heftar " Hephthalite". But an equally simple emendation
would be to 3y» xazar " Khazar ". This reading is evidently supported by the
mechanically transliterated Pazand azar, from which can be reconstructed
a Pahlavi variant reading Jy» or Jgxy- A similar Pazand reading seems
to be reflected in the Parsi-Persian j\*»\>- ^asa^r f° r xazar-
2. <x»^yi averakik " of the desert". The Pazand recension omits
this whole phrase, but in the Parsi-Persian iSjljl reflects a Pahlavi £>y*ty
with p (p = v) misread as z, and with -1 < -Ik which provides the desired
adjective. My earlier reading andarak taken as a variant of i)iy» andarg
" among" was never convincing. It may be added that Anklesaria and
Dhabhar read Audrak without explanation. A case of averak " uncultivated
desert" was quoted by Bartholomae, ZAIW 110, from the Pahlavi version
of Videvdat 14.17. In Armenian occur the loanwords auer and auerak both
in use as substantive and adjective. In Psalm 101.8 [ = 102.7] we find ele
es orpes bou y-aueraki " I am become like an owl in the desert" (LXX iv
otKOTre'So)). In the Mekhitharist Dictionary of 1838 is quoted bnakic'k' auerakac'
" dwellers in deserts ". Evidently the desert suits well beside the mountain,
as we find in modern Persia the kuhistan and the dast.
This will also provide, as it seems, the only reference to the Khazars in
Pahlavi.

2. Barmak
It has long been known from Arabic sources (see the Encyclopaedia of Islam,
s.v. Barmak) that the name of the famous Barmecide family of Baghdad, in
Arabic written *ilv», Barmak, was so called from the title of their ancestor,
which he held as head of the Buddhist monastery Naubahar, the Sanskrit
Navavihara, in Balkh.
What this title was can now, I think, be shown. In Khotanese texts the
Sanskrit pramukha " chief" is used in various dialectal forms as the title of
the head of a Buddhist monastery (vihdra). In BSOAS x 921 I quoted the
Khotanese forms of this word prramuha, prramauha, prraumahg, prramdha.
How the word sounded we know from the Tibetan transliteration par-mog
which would indicate *parmok < *paramok. In later Khotanese intervocalic
-h- and -h- seem to have been used to express a fricative guttural y or %, for
which the nearest Tibetan was final -g. It is clear that the use of this Sanskrit
pramukha corresponds to the use attested for barmak in Arabic. We shall
therefore assume that the Bactrians, too, had used the same Sanskrit-title and
that from Bactrian it has passed into Arabic with the usual replacement of
a foreign p by Arabic b. The word seems not yet to have been found in Sogdian
texts. Arabic etymologies are given in Rev. du monde musul. 20.122.
 IRAOTCA 6

3. Notes on the Rama Text

(1) padamda P 2801.4.
This word is not, as the ending -dmda suggested, 3 pi. preterite of a verb
but a comparative of an adjective *paddna-, that is, padamda < paddmdara,
which occurs in an official document Or 11344, 17.5 mara ajumyara paddmdara
ttd parau tsve " here bring first, so the command went o u t " . Similarly Or 11344,
11 a 2 iau kusi paddmdare ttd parau haude " one kusa-measure first, so he gave
command". The shorter form again in Domoko A 4.7 padamda ttd parau
haudem " first so I gave command ". The word occurs also a second time in
the Rama poem P 2783.5 padamda jabvl myg,na -- rre std ye " formerly there
was a king in Jambudvipa ". The shorter form of the comparative suffix
-tara (older -taru) occurs also in.hasta = hastara " b e t t e r " , tvada, utvada
= utvadaru " more ", buda — budaru " m o r e ' ' ; note also for final -ra also
tcau, tcahau — tcaura, tcahaura " four ", ysa = ysdra " 1000 ", duska =
. duskara " difficult", hada, hamda = hamdara " among ".
This explanation of padamda "before, first" makes it possible to group
four of the short lines marked off by punctuation in the MSS. into two verses
as follows:—
suka ttye sasam kimna -* khvam esta buysye badna ~-
padamda Mmdva ksira ~* mahaisvara bahyq,: tta gauttrrai ~-
and so throughout the poem. The previous translation made this impossible.
(2) std " tired " in P 2801.18, as explained p. 598, occurs in a hendiadys
with khajauttd " tired ". The same meaning of std- is found in Ossetic stad
" tired " < *stdta-.
(3) A difficult passage was presented by P 2801.21 ama ra stdm dmdyara n\.
I had hesitatingly read ranj, = ramnd " jewel". But it is clearly to be read
with n\ as a separate word, dmdyara is the usual 2 plur. imperative ( < -dta ra)
and n\ is a form of the infinitive of nds-: ndta- " to take ". As passete infinitive,
beside pa&sdta- ptc. pass, shows, the infinitive to ndta- " taken " would be *nete
in older Khotanese. This has not been found, but we find four forms of the
infinitive to nds-: (1) note, P 5538 A 39, 40 (official document) paryami ndte
" we condescend to take " ; (2) nati, Vajracchedika 11 b 4 and nata ibid. 12 a 1 ;
(3) nita Vajr. 12 a 4 (all three translate Skt. udgraMtum in the Vajracchedika);
and here (4) n\ with the usual spelling without intervocalic -t- (as in the familiar
dam,, dye, da " appearance ", sata, sse " 100 " and the like). The passage is then
to be rendered : " You must indeed think how to get that thing such as I do
not possess."
(4) haste.
This is not " s e n t " but " reported ", and hence to be kept distinct from
hesta, haiste, older hataiste " he sent ". The word is found also in Mazar Tagh
c. 0019.5gvdrihastddi " they reported the affair " ; Ch ii 002,103 r 4 (Siddhasara)
Itauda haste " h e explained its being h o t " = Tibetan 187 v 2 cha-ba yin-par
 4 H. W. BAILEY—

b&ado [This text is already in print but not yet published]. In an official docu-
m e n t P 2 7 9 0 . 5 4 u tti tta vd hve si mistami vd ttdttdhd: haste si . . . " a n d so
he said that the greatest tutuk reported that . . ." ; P 2741.46 u si ha nose
bisd virdsta hvamdi haste si . . . " and this humble servant to the men reported
t h a t . . . " ; P 2741.6 viri vd didye hadai si khum-cd svdm-sl haste " there on the
third day the princess (?) SVamil reported ".
(5) I had not noticed the present bases of the preterites bida and sasta
in P 2801.32, but both do in fact occur : P 3513 17 v 1 na bvtti u ni pdristd
with bitti 3rd sing. < *abi-gundati (?) and Ch ii 003, 45 r 1 sana in the passage
tcamna tta arvi sana u sa agada " therewith this medicament is to be prepared
and this medicine " (also in 91 r 4, 92 r 2). Similarly P 2927.91 maista ted
padimdnd uphviramdai sand -> van ysindhand = P 2025.223 ted padimauna
u spiradai syauna vara ysindhauna " a great pool must be made, a fountain (?)
must be prepared, there she must be bathed" (in the Sudhana-Avadana;
the Sanskrit text has sudhayd praleptavyd). Hence sg- sya- is present base to.
sasta-. This might correspond to Skt. ksad- " prepare ", if there was an Old
Iranian *yzad-.
(6) P 2801.31 va = vaska (see BSOAS x 1022): kdrsa va ca burai Hdka
" for the magic circle whatever was necessary to him ". This va " for " is
frequent. A correction, however, needs to be made in BSOAS x 1022 in respect
to in. Clearly vi is not older Khotanese vdtd but a later form of* vvra. This is
confirmed by P 2958.36 hairtha wra beside Ch ii 002, 144 v 2 hairtha vi
" suddenly " ; and by P 2787.164 disa paiskala vim beside Ch 00267.35 ttye
disg paiskala vi. The older vdtd, however, survives as vya, vyai in Ch c. 001, 940
hamdrri vya, Ch ii 002, 129 r 1 hamdrre vya, 126 v 2 hamdrri vyai " among "
i n E 939 hamdrvdte.
(7) In P 2783.16 samai byamdd jdrnda rr%ne should perhaps be rendered
" but for him they beat the queen into unconsciousness ". That means the
use of byamdd as proleptic adjective fern, from bitandaa-.
(8) P 2783.39 sgisa " &esa ". In comparing Khotanese $ with Skt. s in
this word I had not noticed the similar case of Khot. sola = Skt. sati" curcuma
zedoaria " (Ch ii 002, 155 r 2 sola = Skt. sati; ibid. 125 r 1 said = Tib. li-zur-
ba ; ibid. 129 v 1 said = Tib. ja-ti). Khot. I < Prakrit d = Skt. t is regular.

4. simdbandha
A Skt. word simdbandha is attested in the Divyavadana (15021) simabandhah
krto bhavati but it is of uncertain meaning. The Mahavyutpatti (ed. Sakaki
6825) has the same word : simabandhah •= Tib. mchams bcad-pa " cutting the
boundary ". Recentjy the word has been met by Professor Minorsky (Marwazi
(1942) 124) where simdbanddt corresponds to nairanjdt " spells " in a parallel
text. I t has also recently been published in Sogdian as sym'fintt (Henning,
Sogdica p. 60 and p. 61, note). The same word occurs in a clearer context in
Khotanese : P 2958.145-7 mistye ttikye kirdstq/nd nama jsa hvdstd u him: s\na
 IEANICA 5

hwi gyasty,na ttiranidard iyaistha drahd:hsq s\mSbamdha haurlmde " they give
protective talismans to be put upon the divine body of this great lord with
name Kirastana and his consort". lyaisthaa- is < Skt. adhisthita-ka-; hvu:s\na
is < Chinese ^ \ (Karlgren 41, 930) ju-zm, < pj,u-nzien ; on this title see
Pelliot, JA. 1925 1. 258 (htm = Chin./w <piu). In Khotanese we have also
husaina, husina, hou:sjne. With bandhana we have in Khotanese Ch c. 001, 979
simdbamdharii yanymd = Tib. mchams bead-par bgyiho " I will make for
him a magical boundary ". In Hindi similarly slmdbaddha means " put within
a boundary ".
5. nydnadu
In Vajracchedika, 38 a 3-4 occurs the verse—
Khotanese. Sanskrit.
dharmahe jsa baysd dyyna dharmato bwddho drastavyo
[daft] damne taramdara nydnadu dharmakdyd hi ndyakdh
Konow (in Hoernle, Manuscript Remains found in Eastern Turkestan 1916)
had failed to distinguish nyd from ttyd, a difference which can easily be seen
when attention is awakened. With the reading nydnadu we have either a 3 pi.
imperative to nydn- as a causative base to nay- " lead ", hence a mechanical
translation of the Sanskrit ndyaka-; or perhaps a pres. ptc. *nydnanda- with
the -u which is found also in mdnamdu " like " in place of mdnamda. Certainly
a noun to translate ndyaka- would be better. It has occurred to me that possibly
in the Rama text P 2783.2 naddm should be emended to *nydnaddm " leaders "
if we assume that the initial syllable has been omitted by the scribe.

6. tcamgala
tcamgalai in P 2783.72 is not "chain", but must be compared with Pasto
cangal " elbow ".