The Secret of The Chinese Method of Tran

XGxd+Sâ tic'** r / ^/'^
THE SECRET OF THE CHINESE METHOD
OF TRANSCRIBING FOREIGN SOUNDS
BY
Dr. G. SCHLEGEL,
<i
Professor of Chinese Language and Literature
at the
University of Lcydcn.
Reprinted from the T'oung-Pao, Scries 11, Vol. I.
ORIENTAL PRINTING-OFFICE
FORHERLY
E. J. BlifLL.
INTRODUCTION.
As we have said in our Conclusion 72, the present paper is

only to he considered as an Introduction to a comprehensive
dictionary of all the characters used by the Chinese in their
transcription of foreign sounds.
This is not a superfluous work, for the indications of sources
given for Sanskrit words by Julien in his otherwise excellent "Methode"
are in several instances so incorrect, that I had the greatest pains
in hunting them up, and sometimes failed entirely in doing.
Julien had two copies of the same edition of the Fan-yih
Ming-i-tsih *), the well-known Buddhistic Dictionary, of which he
presented one to his friend, the late Professor Hoffmann in Leiden,
and which copy is now in the Library of tie Leiden university.
In order to justify my judgment, I may mention that Julien
refers the word Pouroushasya (no. 23) to Chap. II, fol. 19 of the
Fan-yih, whilst it stands in Chap. XI; Oupanishad is referred by
him to Chap. XVI, fol. 16, whilst it is mentioned in Chap. VIII,
fol. 15 recto.
He refers Kdugdmbi (no. 55) to Chap. VII, fol. 21, whilst it
is mentioned in fol. 12 verso.
He refers Bahoudjana (no. 135) to Chap. VI, fol. 6, whilst it
stands in Chap. V.
G. L.n33 1
IV INTRODUCTION.
He refers Gavampati (uo. 277) to Chap. XI, fol. 15, whilst it

stands in Chap. II.
For Mahoraga he refers to Chap. XXII, fol. 8, which is im
possible, as the Fan-yih has only twenty chapters. It is found in
Chap. IV, fol. 22 verso.
He seems to have confused the Fan-yih with the In-i as he
refers in no. 337 for the word MnMrt.a to In-i, Chap. XXII, fol. 2.
Kambala (no. 360 a) is referred by him to Chap. XVII, fol. 13,
instead of to Chap. XVIII, fol. 13 verso, where it is found in his
edition.
The word Qankha (no. 468) is referred to Chap. IV, fol. 14 recto,
whilst it stands in Chap. V.
He refers Dakchina (no. 510) to Chap. II, fol. 29, whilst it
stands in Chap. XI, fol. 29 verso.
Likewise Magailha (uo. 611) to Chap. IV, fol. 19 redo, whilst
it stands in Chap. VII, fol. 9 verso.
For goudha (no. 667) he refers to Chap. IV, fol, 25, where it
is not to be found; the indication refers to the word Goupta.
Godhauya (no. 684) is referred by him to Chap. V, fol. 6 verso,
whilst it stands in Chap. VII.
The hell Niraya (no. 840 a) is referred by him to Chap. I, fol.
19 verso, where it is not mentioned. It is to be found in Chap. VII,
fol. 2 recto.
Arhan |$flf. ^ (no. 1049) is referred by Julien to Chap. I,
fol. 9 recto, whilst the word stands in Chap. II, fol. 3 verso. Iu
Chap. I, fol. 9 recto stauds the word |$flf ^ gfil Arya.
For Khankhtha (no. 1079) he refers to Chap. XIV, fol. 8 recto,
whilst it stands in fol. 18 recto.
In no. 1188 and 1189, he gives for Mail the character ,
INTRODUCTION. V
whilst the Fau-yih has in both places jjjjj, without the radical iQ.
The word Sarpduchadhi (uo. 1345) is referred by him to Chap. II,
fol. 2 verso; it stands in Chap. IX.
Piptlikd J|i |H ^{j, (uo. 1354) is referred by him to Chap.
IV, fol. 20 verso, whilst it stands in Chap. VI, as he says himself
in no. 1388.
Bahoii (no. 1442, p. 172) is referred by him to Chap. VI, fol. 6,
whilst it stands in Chap. V.
In no. 1622 S jg is referred to as pouchpa and mentioned
as standing in Chap. XV, fol. 19 of the Fan-yih; but this chapter
has only 18 folios. Julien probably meaut the Tn-i. The word
Pouchpa (flower) occurs in the Fan-yih, VIII, fol. 4 recto; but is
there written ^ ^.
Sub 1623 and following numbers he gives instead of
which character is employed in the authorities he quotes.
Nisthdna (no. 1703) is referred to Chap. XIX, fol. 10 verso, whilst
it stands in Chap. XVIII.
The word Achddha (no. 1762) is referred to Chap. IV, fol. 24 verso;
it stands in Chap. V.
Sub nos. 1871 1873 he gives for tchi, whilst his authorities
have .
Madhu (JfSfjjjj, wine, no. 1938) is referred to Chap. V, fol.
13 recto, where it is not to be found; it stands in Chap. IX, fol.
1 recto.
Indu ^ j (no. 1994 a) is referred to Chap. IlI, fol. 1 recto
instead of to Chap. VII, fol. 8 recto.
The city of Djdlandhara (no. 216 recto) is referred by Julien to
Ma Touan-lin, Book CCCXLVIII, fol. 22 verso, instead of to Book
CCCXXXVIII.
The word Tchakchous ffi j$ (no. 2183) is referred to Chap. XVIII,
fol. 3, whilst it stands in his edition in Chap. XVII, fol. 3 verso.
VI 1NTROUUCTION.
Iu no. 2208 he gives jjja wei for vi in Vichtioii and refers to

the Fan-yili, Chap. IV, fol. 16 recto. The uatne is fouud on fol.
17 verso, but is there transcribed ^ 2p ^ fet-suh (= vich) -nou.
The word Koumouda tyf\ ^ (no. 1942) which he refers to
Chap. VIII, fol. 4 verso, is referred by him, in uo. 2215, to fol. 14.
A
The words Arya and Kachdya are not transcribed in the Fan-yih
by PrT W- H aml ^ lit as Julien sys in no- 2238 ', but by
In no. 2240 Julien says that the character ye ^ has been

employed erroneously for gva in Vigvabhou 3| in the Lalita
vistdra. and he forgets that he has given himself the sound che to
the character ^ in the same word, sub no. 86. He refers to the
Fan-yih, Chap. I, fol. 18 verso, where the word is transcribed
^(5 pi-chou-p'o. Besides Julien is wrong in saying that the
characters ^ ^ and ^ ^ represent the syllable gva, whilst he
ought to have said ^ pi-che and %Q pi-chou for w'f in
In no. 2274 he says that [H| is erroneously employed for dja

iu Rddjagriha and refers to Chap. VIII, fol. 10 recto of the Fan-yih.
It stands in Chap. VII, where the character is simply a clerical
error for ^ dja.
- In no. 624 Julien says, " Men pour few dans Skandha"
and refers to the Fan-yih, VII. fol. 13 ra7o; but there we find
the name of the kingdom of Gdndhdra $|| JS^ ^ kien-t'o-lo.
Page 22 he refers the word skandha AS |J to the Fan-yih,
XVIII, fol. 1, where it is not to be found; it stands in Chap. XVII,
fol. 1 recto.
The consequence of this slovenliness in the above indications
has been that most of these words have not been incorporated in
Eitel's Skt.-Chin. Dictionary, as the author has not given himself
the pains to correct Julien's false indications.
INTRODUCTION. VII
Besides Julien often only gives the value of one syllable of a

word, without giving the others, as in no. 1045, where he gives
|H J^S lo-mo for the fiual syllable lam in Vahoulam, but does not
give the values for va and hou. Sub no. 1424, Julien gives ^
for va in Virvaram aud sub no. 769 for ram in Hevaram; but
he gives nowhere the value for vir; sub no. 1462 he only gives
the first character ^ of the word valgu, without giving anywhere
else the values for l and gu, and so in numerous other cases; so
that he has not always fulfilled his promise given on page 14 of
bis Milhode, that he had made a special card or ticket for every
chinese character in the transcription of a word.
In no. 254 Julien says that jjfjjjj chou-po stands for gva in
Vigvabhadra, whilst he says in no. 1488 that po stands for hha in
Bhadra ( |J ) ; not giving the value of va to this character ,
whose old sound was pat, and can thus stand for bhad, but not for
va. We suppose the Fav-hoa-wen-kin, which he quotes as his
authority, has i$|J )|JJf |J , according to Julien's transcription
pi-chou-po-to, pi-chou standing for Vig and po-to for Bhadra *).
Compare the Fan-yih, II, fol. 5 verso: jjj|5 ^ jj^ [Jfj pi-chou-
po-to (according to Julien's spelling) *) for Vigvabhadra (Cf. Julien,
no. 1400), in which transcription the syllable va is equally omitted.
In number 1771 Julien says that represents kcha (read kchan)
in Tallakchana, referring as his authority to the Fan-yih, VIII, f. 13;
but there we find this character employed for transcribing the word
Baldksha fffi jj| pat-lo (= bal-ld) -Wam, which is represented
in the "Lalita vistara" by Tallakchana.
In no. 1036 he says that jtffr stands for la (read lak) in
Tallakchana, and in no. 475 that ^ stands for na in the same
word. But nowhere he gives the values for the syllables tal and
ksha in his M.ithode.
*) Old sound pit-tu (for vu-su vif) -pat (= bhad) -Co (= dra).
VIII INTRODUCTION.
Only 011 p. 52, XVIII a, he gives as transcription Ta-lo-la-

tvh'a-jo of which only the characters of the third and last syllable
are given, so that we have the choice between the nine ta's he
enumerates in nos. 1668 to 1699, and between the twelve /o's,
enumerated by him in nos. 1022 to 1060.
I sincerely hope to escape in my dictionary such mistakes and
errors in the indication of my sources, for I have experienced
myself the disagreable loss of time occasioned by Julien's false
indications, having often had to read through the whole of the 20
chapters of the Fan-yih in order to find the word quoted by him.
For this reason I give after each typical character the chinese
transcription of the whole word, in order to spare the student the
trouble to look up for these characters in the dictionary.
As for the orthography of the chinese transcriptions, I have
throughout followed the english spelling except for the sound j,
for which I have used the continental dj; for though the sound d
is heard in the english it is not visible to the eye, and therefore
the chinese transcriptions in which the sound dj is rendered, would
seem unintelligible. Besides, the letter is pronounced as y by the
Germans, Italians and Dutch and would lead them into error. I
have retained the letter v in the transcription of the skt. letter
of, because the english and french sanskritists have generally
adopted it. But, in reality, the letter 5f sounds like tea, and the
german and dutch sanskritists always transcribe it by a w.
In our modern maps the name of the river Irawaddy, a name
derived from the Skt. Trdvati (abounding in refreshing water), is
correctly written with a te, for the skt. name sounded equally
Irdwati.
Finally I remark tha< my references to the Fan-yih are those
in the new Nauking edition of 1878, which is conform to Julien's
INTRODUCTION. IX
old edition of 1603, with a slight difference in the number of the

folios, as in the new edition the chief words are elevated one type
higher than the others.
Except for the chapters I and II, where J-ulien's edition is
some two or five folios in advance, it is two to five folios in
arrear with the new edition. The order of the chapters is the
same in both editions.
Leyden, June, 1900. G. SCHLKGFX.

THE SECRET OF THE CHINESE METHOD
OF TRANSCRIBING FOREIGN SOUNDS
BY
G. SCHLEGEL.
1. I have already, on different occasious, called the attention

of sinologues to the insufficiency of the "Methode pour dechiflrer
et transcrire les notns sanscrits qui se rencontrent dans les livres
chinois" by the late professor Stanislas Julien 1).
Dr. Hirth has equally pointed out its defects and illustrated
them by some well-chosen examples 2). This insufficiency is due to
two principal causes. The first is that Julien thought that the
buddhist monks, who taught the Chinese how to transcribe Indian
words, spoke classical Sanscrit, which was not the case; and the
second that the chinese characters were pronounced at that remote
period exactly as in the present day. St. Julien had not the faintest
idea of the ancient sounds of the chinese characters, nor of any
of the southern dialects of China, which have, for the most part,
preserved these ancient sounds even so far that I confidently
believe that a Chinaman from the IT'an^-dynasty if he could
1) Toung-pao V, p. 173; VII, p. 184 icq, 568 scq ; VIII, p. 490.

2) Chinese equivalents of the letter K in foreign names (Journal China Branch Roy.
Asiat. Sec , Vol. XXI, 1886, p. 214 teg.).
1
2 G. SCHLEGEt.
arise from his grave could converse, without much difficulty,

with a modern Chinaman from Amoy or Chang-chow.
A few examples will be sufficient for demonstration. Iu the
exposition of his Methode, p. 10, St. Julien wonders "that such a
considerable number of chinese phonetic signs do not in the least
correspond to the sauscrit articulation which they are intended to
represent" and quotes as an example (p. 11) that the Sanscrit
sound ft is rendered in Chinese transcription by chi ^9J ^| ^|
^ iSH i by c^l tffe $k an<^ by tseh ^ . Now the seven
first characters are all pronounced ti in Amoy-dialect 3) ; the two
characters chui are pronounced t'w, and tseh tik 4).
How to recognize the sound da in 3| cA'a? asks St. Julien.
If he had known the Amoy-dialect, he would have found that this
character is pronounced there te (as the well-known cheering cup
is still called on the continent and by the Irish tay; in English
it has been written tea and is now pronounced tee).
The ancient vowel a has been preserved in the Cantonese ch'a
and the ancient initial d in the Shanghai-dialect dzo. But the
chinese authors have probably made the common confusion between
ch'a 3j> and t'u ^ (Amoy <o, Shanghai du); the character 3|
being of later artificial formation.
Armed with this science, it is not any longer difficult to
recognize Sanscrit da in Chinese 3j or ^ whose ancient souuds
were also da.
As a natural consequence of this ignorance of the southern
dialects we get queer indications like the following: "^| Fou pour
3) = = JJjjig "an owl", pronounced tsi in Amoy, would seem to make an

ejception j but the character is also pronounced liet ( jj^fc -jjjFj ) when
applied to another bird.
4) According to the Jjt? the character ^jj waa pronounced <'(a -)- \)ik m t'ik
daring the T'ang-dynasty ( ;fgj ^ ^ ty] ^ JJ|J ).
THE SECRET OF THE CHINESE METHOD ETC. 3
pou dans Pouroueha; <|j Fou-leou pour pour dans pourqa".

(Julien, Nos. 317, 320).
Now in Amoy-coUoquial 1| (rich) is pronounced pii (poo),
which was the old sound of the character. The l of leou naturally
represents the r, so that poo-l(ao) represents quite distinctly pou-r.
To the present day the characters *^ ^ ^ pu-tut-sin are em
ployed by the Java-Chinese to transcribe the malay word putusin
(pr. pootooseen) to decide, give a verdict.
So we find under N. 298: Fo pour bou dans Bouddha".
Now the old sound for ^ was Bud; and thus the definition ought
to have been : ^ for Bud in Buddha ( ^ |J ) ; for, as we have
already observed in the T'oung-pao (Vol. V, p. 173), each syllable
of a foreign word ends with the consonant with which the following
syllable begins. The Chinese read Bud-dha and transcribed it there
fore correctly by Put-tho.
At the present day Buddha is still called in Amoy Put, for
the older form Bud.
By not observing this rule, Julien also gives: "-tt Fo pour
pou dans Vaxpoulya % $f} I3^f " (also transcribed Bft *g ^ pipoolo
in Amoy). He ought to have said: ^ put for pul in Vaipulya.
For the last character is pronounced in Chang-chow, near Amoy,
lidk = lya. The l in pulya is repeated in the chinese transcription
in order to show distinctly that the initial of lya is an /.
In N. 290 we read; Fen pour powt dans poundartka
(^ ^'J ^Jfl ) '- The character pronounced in Amoy book-
language hurt, is pronounced pun in Colloquial. The syllables punda
are also transcribed ^tj*3j, the first of which, meaning to run
away, is pronounced p'un in Amoy, whilst the second te represents
an old td or da.
Again we find (N. 294-295): Feou pour bou dans
Bouddha; pour M<m dans Adbhouta ( ^ ^1 )". The Amoy
4 G. SCHLEGEL.
pronunciation of ^ (to float) is p'u and was probably also

pronounced pu; cf. p^ which is as well read pu as p'u, so that
it represented perfectly well the Bu in Buddha and the bhu in
Adbhuta.
To the present day the character is used by the Java-
Chinese to transcribe the Malay syllable pu as f. i. in sapulu (ten),
transcribed J? ^ j|| sa-pu-lu s).
2. But the worst confusion Julien made was due to his
ignorance that most of the chinese syllables ending with w had
formerly m as a final, exactly as in the present southern dialects.
Page 45, XIII, he says: that final n changes to m before p
or /, corresponding in Sanscrit with p ox b; and he quotes as a
demonstration of this rule the words San-po-tchi ^ jj^ ^ for
Sampatti, Tchen-po-lo tf= ^J? jj^ or ^ ^ for Djambala,
Tchen-po | for Tchampa, Tchen-feou H ^ or ^ ^ for
Djambu, Lan-po for Lampa, Kien-po-lo jjjjj ||| for
Eambala, etc.
Now these characters sound in Amoy-dialect sam-poat-ti, tam-
po-lo or tsiam-po-lo, tsiam-pu or iam-pu, lam-po and gam-poat-lo,
which all have the old final m, now lost in mandarin and changed
to n. At the same time these words show that the actual Amoy-
pronunciation does not differ much from the old chinese pronun
ciation at the time when these sanscrit words were transcribed.
M. Hirth was therefore quite right in saying: "in order to explain
why the Chinese Lan-po corresponds to Sanscrit Lampa, it was
not necessary to say that n is changed into m before p: for the
writer who transcribed this term into Chinese, did not read lan,
but lam".
5) A Chinese Malay Manual, Chapt. ^ >|j (Numerals). I shall preaently return to

this valuable Manual.
In XIII A, Julien says that final n is very often left out

before syllables having an initial m as e. g. in Rama, transcribed
[H* J|| , which he reads lan-mo ; in Samaya, transcribed ^ |p JfP ,
which he reads san-mo-ye; in Yamuna JH] ^ ^JJ$ , which he reads
yen-meou-na, etc.
In Amoy-dialect all these characters retain the old final m and
read Lam-mo, Sam-mo-ia and Iam-baw-na.
Besides, the rule given above 1 of repeating the initial con
sonant of the next syllable, is applied here. The Chinese transcribed
lam-ma and not la-ma because he wanted to point out distinctly
to his countrymen that the second syllable ma began with an ni.
The same law is applicable to the examples mentioned by Julien
in XIII B, as in Panasa S< pan-na-sa, Kapphina
Kia-p'in-na, pronounced at the time Kap-pin-na (Amoy
Kiap-pin-na) where the character hap is chosen in order to
indicate that pin has an initial p, and pin was chosen to show
that na has an initial n, etc. This law is still rigorously observed
by the Java-Chinese in their transcription of malay words, as e. g.
^Jj ^ but-tut for betul (true), ||| ^ ft sang-kut-tan for sang-
kutan (souvenir), ^ kit-tsit for ketjil (little), ^ tin-lun
(= tin-nun) for f.enun (to weave), Pjflf but-lun (= bul-lun) for
belum (not yet), etc.
In XVII A, p. 51, Julien pretends that the Sanscrit fa and
tha are rendered in Chinese by characters pronounced ch' (chh), an
aspirated ch, as e.g. \% for Rfc, P<, J%, f for ,
pronounced in Amoy-dialect ts'a with the exception of the first,
which is pronounced t'a and the last which is pronounced t'ok. All
having the same phonetic ^ t'ok or ^ . Comp. the characters
!j -ft Y an<^ f ' a" pronounced t'ok in Amoy. The said
characters will thus have been pronounced t'a at the period of
G. SCHI.KGKI..
transcription, and were thus very appropriate to transcribe the

Sanscrit ta and tha.
3. 1 scarcely need to strengthen the proofs adduced by M.
Hirth for the rendering of the letter r by t in Chinese e), as I
have given abundant examples of this feature in the transliteration

of Turkish and Uigur names in my "Stele du Teghin Giogh" and
"Inschriften von Kara Balgassun", and I shall ODly add a few
more examples of this law.
2|5 , old sound sut, for sur in Suri ^ %\ .
k'at kar in Kama |
k'at kar in Kavpura | ^ .

j k'at kar in Karma j ^ .
hat > har in harikega 1|I
sat *or in Sarvadj/la
)it ti ' o^ sound <<-/o, for c/ra in Tchandra J$f | | .
-î- piWi pri in Dharma priya.
j> pt<-/t pre in prekchini.
^5 , old sound pa<, for par in parama J ||| par-ra-ma and
in parvata J par-ra vat-ta.
These examples could be multiplied by thousands. The isolated

instance mentioned by Hirth of J3 muh-luh for Mwru or J/erv 7)
is to be explained by assimilation of muh with luh, giving mulluh
for murru, if, at least, ^ iuA does not stand erroneously for ^jc i
old sound mat - mar.
6) Journal China Branch Royal Asiatic Society, Vol. XXI, 1886.

7) China and the Roman Orient, p. 141.
THE SECKKT OF THIS CHINESE MKTHOD ETC. 7
4. But there always remained to solve the riddle why, in

chinese transcriptions, words with a final n are used to represent
foreign words with final r, as e. g. jHfc ^ Timen for Timor, also
transcribed jRfl , in Amoy Tibun, fj^ ti-bun, fft ^ ti-hm
(but also Jjg tibut with final t for r). Further "j^ k'un-tun
for fcuncfer in Pulo Condor. We find WiM^LM.^ M pan-li-
fah-lo-tcho-kia for Parivrajaka; ^ jjs^ pan-lo-sah-mi for
Parasmaipada; ^ '/!=! Ijj5| pan-nieh-poan for parinirvana, ^ ^
(for ^^5^) pan-jO (pan-lo-jd) for pradjfia.
5. This question had preoccupied me since many years when,
by chance, a small Malay-Chinese Mauual, published in 1889 in
Batavia, fell into my hands.
This Manual bears the title of Jj| jgf ^ ^ , or "Ferry and
Bridge for Conversation", and was compiled by a certain jjjjjj ^
and . The copy was written out for the press by the
latter's eldest son ffi j ^jlj.
The authors show considerable talent, and the son, who probably
had enjoyed a enropean education and seems to have been very versed
in the low malay lingo spoken at Singapore and Batavia, wrote
the whole text.
The Manual consists of two volumes comprising 48 -}- 41 or 89
pages, each page being divided into six columns and three partitions,
so that 18 words are meutioned upon each page, giving an average
of 1500 to 1600 words.
As a sample I copy:
0. SOHLEGBL.
As is seen, the chinese word ^ , meaning I,

is placed to the right. To the left is its chinese
pronunciation, and lower the malay signification
goewa, whose transcription in Chinese J
gil-od is given to the right. The note [in smaller
chinese type] says that these two sounds (gu aud
od) ought to be pronounced as one single sound
(= gwa).
The second partition has to the right ^ you,

with the chinese pronunciation Li to the left and
the malay equivalent loe (pronounce lu or loo,
which is transcribed by JjB lu).
The third partition has to the right ffi he.

To the left Ie (= E in English); lower on the
malay equivalent Dia with its transcription in
Chinese li-ia in which initial l represents initial
d which the Amoy-dialect does not possess.
The two volumes contain XVIII Chapters treating respectively of:
Chapt. I. P Jp| i jf^f Parts of speech.

II. 0 ffl & Wi Phrases in daily use.
HI. ffc g ^ Numerals.
IV. U^f ^ ^ Times and seasons.
V. ^ Astronomy.
VI. Mb Jl ill ?fC Physical geography.
VIL If fl ^ Hi Parts of body and motions.
VIII. ^j: ^ ||| Diseases and Medicaments.
IX. ^J' rff Jj| Buildings and Dwellings.
X- H H Implements.
XL fit lit A ^ Official and civil employs.
XII. jHfc ^ Political geography and topography.
XIII. ^ Stuffs and Silks.
XIV. ^ MM Precious things.
XV- ^ P*l l& ^ Tribunals and procedures.
XVI. 3l ^ jjj^ Grains and Greens.
XVII. 1 1=L % ?fC Flowers and Fruits.
XVIII. ^| Jjj| Birds, Beasts, Fishes and Insects.
Ths author's preface tells us that he was born in the village Min-k'i
near Thao-yuen in Fuh-kien ( 3& fH ijg A & )-
That he had studied literature in his youth, and came later, in the
year Ting-mao (1867), with a friend to Batavia, where he commenced
to jot down every thing he heard, until at last "the gussets had
formed a fur-robe" ( ^ jg ^ fifo jfc ^ ). That he had studied
Chinese as well as Dutch, and had compiled his book, that Chinese
born in Batavia could learn Chinese from it, and Chinese coming
from China could learn the Batavian language (Malay) from it;
winding up with long excuses for the insufficiency and shortcomings
of his feeble efforts.
10 G. SCHI.EGBL.
This preface is dated from the year Kia-shin of Kwang-su

which corresponds to the year 1884, whilst the Malay titlepage
bears the date 1878 and the reverse of it the Chinese date: Year
Ki-ch'ow of Kwang-su ( jfc 2i ft ^ ) which corresponds to
1889. The titlepage states that the printingblocks are hoarded up
in Singapore (#f^#3||?&J8t).
I suppose, however, the copy in my hands to be a reprint; for
another preface, printed in red, states that it was compiled in the
year Wu-yin of Kwang-su ( jjfc ffi ^ ), corresponding to
1878, by a certain Tan Bong-Hong in an hotel in Batavia ( (Jjj( <^
This seems to be the same man, mentioned as Tan T'ien-pit

( |5jj[ ^ ) on the titlepage and whose assistance is gratefully
acknowledged by the author in his Introduction ($tj ^j); a man,
he says, thoroughly versed in the Chinese and Dutch languages ( ^
He then explains the method of his Manual as we have indicated

above, stating that he had purposely compiled it for the benefit
of his countrymen born and grown up in the southern seas,
knowing only Dutch aud Malay and not Chinese. In these southern
seas, he continues, are numerous islands and settlements, where
many different languages are spoken; the Malay language being,
however, the lingua franca for all merchants, so that it ought to
be attentively studied (^^j&^^gg^. W fg" #
With his nice chinese ear, the author had distinguished that in
Malay tones exist as in Chinese, so that he gives tonal marks to
the Chinese characters used for the transcription of the malay words,
THK SECttbT OP THE CHINESE METHOD ETC. 11
adding to some of them conventional marks as J^, O Q-

The author, being a Yung-ch'un man (Yung-ch'un is the modern
name for ^ ), this Manual is throughout written in the Tsiuen-
chow dialect, the Chang-chow words being marked by a ; whilst
the change of sound is marked by a O (~f' Wi ^fP ifr
Experience has taught me that this O *s placed next to a

chinese character to indicate that it ought to be pronounced in
Colloquial and not in Booklanguage; for instance is to be
read pu and not Am, is to be read sang and not song, J) o
is to be read ldl and not Uk, is to be read tit and not tik.
A Sinologue, ignorant of the Amoy-dialect, would never under
stand transcriptions as ^ o for the syllable lang, ijjjfc o for bang,
ffi o for bang, jfjô for to, for in Booklanguage these characters
sound dzm, bun, bong and mo. The little circlet to the right shows
that we have to read them as they are spoken in Colloquial. He
would not recognize in the transcription libit the Malay
libar (broad) if the knew not that the character ^ is pronounced
in Colloquial bat (= bar). Neither would he understand that
represents the malay word tulis, if the author had not taken care
to add : Jjjjj ^|] fffi jjf] , gjf| is to be pronounced l(i -f- p)it - lit (= lis).
The syllable sang, having no characters for the ^ i the author
indicates the pronunciation of the word saksi (witness) by the
characters 3$ v*z- san9i with the lower short tone = sdk-si.
T
A
He then describes the Chinese tones, upper and lower series,
of the 2p J- ^ and ^ P* , remarking that in the Chang-chow-
and Tsiuen-chow-dialects only seven tones are to be heard, some
making the and ^ of the lower series the same, others the
Y. J** of the upper and lower series the same, to which the
12 0. 80HLKOBL.
attention of the scholar is drawn. He indicates these tones in the

ordinary way by a half-circle at the four corners of the chinese
characters, as P for the ^ p , Q for the _fc. p , Q for
the ^ p and Q, for the A F -
His further indications agree with those given in our enropean
Manuals and Dictionaries of the Amoy-dialect, so that we need not
repeat them here.
6. Now, though in theory any syllable may be toned in the
7 tones of the Amoy-dialect, there are not always characters for
these hypothetical tones. If we take for example the syllable gan
we shall scand this hypothetically:
f _t * A T* TA
1. gan. 2. gdn. 3. gan. 4. gat. 5. gan. 6. gan. 7. gat.
But if we look up for this syllable in the -j- 3l Dictionary,

we will only find characters for the second HJ^ gdn; for the fifth
and gdn ; for the sixth j^J , ^ff , jJH i 5ran' an<^ ôr
the seventh and gra<. With other words there are no
K , _t an<l _t A sounds for syllable gan.
Again for the syllable Sang there are no characters, except
vulgar ones, for the 2p pij. ||| and ^1 , for the :
}J|, for the Jh i'- j. for the A: There are none
for the other tones.
For the syllable tsang no authorized characters exist in the
-f-* 3L "f=f" i DUt one in the ~J\ A : -HI) isa'^
So the author transcribes the malay word Tjetjak, lizard,
$C ^5 A tsit-tsang with the upper short tone = tsit-tsak.
The syllable fan has no lower short tone character (Ml); so
when transcribing the malay word Djengkrik (cricket) he writes
HI 19 >ft Dziang kin (with the A = kit) ^ kit-lik standing
A
for kir-rik = re'.
The syllable gang has only characters for the 2p v{z, -j'rjj
and f% and for the ^ A w*. fe> UK Mld $' So

when the author of the Manual wants to transcribe the malay word
Dagang, "trade", he writes lagdk; but the tonal mark to
the left of the latter character indicates that gdk is to be pronounced
in the 2| Jx} f ^ e- gang.
Again the syllable he (ha) has no characters for the A ,
(hh); and thus we find the malay loanword kaher = Dutch kar,
English car, transcribed "jffl |p! T^ ; with the addition A placed
under he, indicating that he is to be pronounced with the
A : AÂ, giving kahdh as the nearest approach to kaher.
The syllable Atn has no characters for the ~]\ A * ; so the
author, wishing to transcribe the malay word leher (throat), employs
the characters J|L } ^ lehin with the A = Ai<. Adding in
smaller type: ]L ^ 50? *s to 06 rea^ A(an + p)& = hit.

Thereby the chinese reader is made observant that he has to read
the word lehit (= lehir), the nearest approach to the Malay leher.
The sound lin having no representative in the lower short tone
lit, the malay word Mr (north) is transcribed in this Manual by
-ffe T^ , with the addition in smaller type ~J\ A : Hit for Mr.
Such are the very ingenious means employed by our author to
indicate the pronunciation of the characters used for transcription.
If the supercilious chinese literati who invented the system of
transcription of Sanscrit words had given the same, or similar,
indications, all difficulties for restoring these Sanscrit words to their
original form, would have been obviated 8).
8) The imperial Diet- of K'ang-hi sometimes does so. E. g. it says that the character
HJJ nih is sometimes to be pronounced as the character ning with the short tone ( fjl^
14 B. SCHLKGEI..
7. We shall now give some further illustrations of the reason

why an n is used for a final t (= r or /).
We find the malay word Angat (properly hdngat, heat) trans
cribed o HJ^ 5 ang-gan ; the ) to the right of fl^. tells us that
it has here the upper short tone and is to be pronounced gat, a
syllable for which no chinese characters exist, as we have said above.
Longar (loose, spacious) is transcribed longgan with
the 7^ F*5 = longgat - longgar.
Ingat (to remember) is transcribed ^ |^ ^ ing-gan with the
^ p5 = ing-gat.
Keringat (sweat) is transcribed J|jp ^fit ke-ling-gan with the
^ ^5 = kelinggat.
Dengar (to hear) is transcribed HJ^ ling-gan with the y\
= linggat (linggar for dengar).
Sengal-sengal (to pant) is transcribed HH j ^ft" > sin-gan
sin-gctn with the ^ p*5 = sin-gat sin-gat (= singal-singal).
Anak tungal (an only child) is transcribed >JJt 2$ '/Ir] ana
tonggan with the ^ = tonggat (tong-gal).
Agar-agar is transcribed i|Jl ||J^ } ^ , agan-agan with the
^ = agat-agat - agar-agar.
Tinggal (to dwell) is transcribed "J" } tinggan with the ^
jfS = tinggat: t for
2'anaA #u#r (slide of earth) is transcribed ^ 05 j ^ ,
with the Tana gukut; Gelijat (to stretch ones self) is trans
cribed jj QTu/tn with the short tone = gulit; and so in all cases
where the malay syllable gal, gar or gat is to be transcribed.
Final n before r represents also an r; e. g. in 4|

ban-lien-ti, used to transcribe the malay word berhenti (to cease,
stop), generally contracted in low-malay to brenti. Properly this
word ought to be transcribed by bat-lien-ti; but as the syllable ban
TUB SKCRET OF THE CHINESE HETHOD ETC. 15
has no y\. characters, the author used ban, which is to be

prououuced as having the upper short tone: bat = bar; lien - ren
according to the generally accepted rule of rendering initial r by l.
Mdhal (dear, expensive) becomes in a Chinese mouth mdhat.
As, however, the syllable han has no ~J\ y\ characters, hat is
transcribed by Jji han ( Jj| J|L mdhan) which is thus to be read
as hat. But M. may also have had the sound hat, which sound
it sometimes has when used as a phonetic, e. g. JjH^ which is
pronounced k'at ( ^ Jfa , || , |f ffi,).
There seem to be, however, some exceptions where final l
is rendered by n, without any apparent reason. These are the
words panggil (to call) which ought to have been transcribed
pang-git or pang-it, as it is also pronounced by the Java-Chinese,
bub is transcribed in the Manual by ^pL pang In, which last
syllable has a great quantity of upper and lower short tone characters.
It is true that the final / in panggil is scarcely articulated by the
Malays, the accent falling on pang, so that the first Chinaman who
transcribed the word may have indeed heard and understood pang-in,
instead of pang-il. In ^ ang-in (Malay angin, "wind") the
character 23 has its natural sound in.
The same is the case with the transcription of the word tidur,
"to sleep", by yj^ ti tun, where the transcriber had a quantity
of characters with short tones at his disposition, as in the upper
series tut: ffl , |j{Jj , ^}tj, tyft and |j{Jj , and in the lower series tut:
#t *&> 7fc> fR. f. SK. Bffi aud ^. and which
he employs in the transcription for other words ending in l, as
e. g. betnl (true) ^Jj ^ 6< titt, so that he could as well have
transcribed the word tidur by $\ ^ ti-tut or ^jfj |J{J{ i tut. Like
wise in 'p}' |H kut-loan for keluar, to come out.
But we must also bear in mind that the Malays themselves
often pronounce w imtead of ?\ exactly as the Siamese do. La Lou
16 6. SCHLKOEL.
bere, in his Description of Siam (Vol. II, p. 77), says: "The Siamese
often pronounce an n for an r and an / at the end of a word,
because they do not detach the tongue from the palate at the end
of a word, as it ought to be done in pronouncing r or /; for in
the pronunciation of the l, the sides of the tongue do not cleave
to the palate. The Siamese write Tahar and Mar, but pronounce
Tahan and Man" 9).
This laziness of the tongue is very common among Malays and
even among the dutch halfcaste or the creoles born in India. I
could never teach such a man to pronounce words like kept, left,
far, etc.; he always pronounced them kep, lef and fah because
he was too lazy to pronounce the final t and r.
This explains why final ri is often rendered in chinese trans
cription by gi as in bediri (to stand) transcribed ^c|J W
Ba-li-gi; sometimes even 5lt <7t xs used to render the sound bih
as in the transcription of Malay lebih (more) by [HJ 5it lap-gi
for l&b-bih.
In general syllables having a lower or upper short tone are
transcribed by such characters. So e. g. in the word sepdt (harsh,
sourish), transcribed jj? 't-Pî where the little mouth fj before
/\ pat indicates that the syllable is to be read in the lower short
tone ~f\ ^ , as there are no characters for that tone.
This explains why the Tamil word kappal (ship), which the
Malays pronounce and write kapal, is transcribed in the Manual by
jjjfl /\ kapat in the word for the docks in Onrust Q S
jjjP /V , in Malay palabuhan kapal, which term is translated by
jfjKj kap-pan b, or dock (b) for double-planked (kap-pan)
ships, i. e. european ships (Cf. T'onng-pao, Vol. VIII, p. 527).
9) The change of r into n is not solely confined to Siamese. The Skt. ttdan (water)
sounds in Greek illup, in old High German wazar and in English water. The Hindoos
probably pronounced the old final r of udar as an , like the Siamese.
THE SKCKET OF THE CHINESE METHOD ETC. 17
8. If we now turn to other chinese words wherein final ?i

represents an r, we must suppose an analogous or similar rule as
that employed by the author of our Manual.
So e. g. if Dr. Hirth is right in identifyiug ^ an-sik with
Arsak (Parthia) or an-kuk with Orchoe, we must suppose a
clerical error, viz. ^ for ^ an, which, in the meaning of "to
press down" is read at (in modern Mandarin ngbh) and is a synonyme
of at (Mandarin ngdh). ^ Jg, at-sik, according to the old sounds,
would then perfectly well express Arsak, as chinese t stands for an
r, a letter which does not exist in Chinese. It is the same case
with the transcription in the malay Manual of the name Sidhoardjo
jf^p ^ tJ^ J^j , according to the Amoy-dialect Si-lo-an-dzio, where
is = ^ and is to be pronounced at-ar; the sound an having
no ~J* y\ characters.
In the transcription pH ^ for Termed or Tirmez, quoted in
Eitel's Handbook for the Student of Chinese Buddhism, p. 147 a, we
must remark that Dr. Hirth reads these characters erroneously tan-mih
(Cauton tan-mat); for the character flB. is pronounced tdh ( fjH ^
'|) ty] K'ang-hi) and in Cantonese and Amoy-dialect tat (/(aug + k)a<
= tar). Dr. Hirth must have been led astray by the remark in
K'aug-hi that fjH had the sound of 'jfl., which, indeed, is generally
pronounced tan ; but also tdh and tah ( ijg. |j |*|J jj/J ^ ~f îj
#J ^ 7"n is als0 reatl Tah 1) (ia Amoy and Canton T'at).
Consequently the characters jfl. (or flJJ) ^Jf , used to transcribe
the name of the city of Tharaz, ought to be read, according to the
old sounds, T'at-la-sz = T'ar-ra-z.
The city of Ahikshetra is transcribed |$flj" ^ l|j ^4-Ai-
tse-tat-la, where <<-/a stauds for tra. The name of the town
Charitra, S.E. of CWa, is transcribed ffi (0 5p| tsiat-li-t'at-lo,
10) In the name of the celebrated concubine Tat-ki T"** {TarkiV).

i
18 G. SCHLEGEL.
wherein t'at-lo again represents tra. Djina puttra is transcribed

J|5f jjfa PH jjsfl Tsin-na put-tat-la, where tat-la represents tra.
In Buddha ksh&tra the last word is transcribed ^ ^fi j|fl

kit-che-tat-la, where jjj <a<-/a represents tra ").
We may add also that the phonetic tan J|. is read t&h in ^
and fa ")-
Julien (no. 1710) has committed the same mistake in reading the
characters Jjpj (Sumatra) Sou-mo-tan-lo, instead of Sou-
mo-f&h-lo, old sound Su-ma-tat-la, wherein tat-la is = tar-ra = tra.
If the maker of this transcription had made use of the diacritical
mark to the right of the character 1fflj) indicating that it was to
be pronounced in the y\ |S [tat), this confusion would not have
been made by Julien.
The character JJ must have had two sounds tan and tat; for
else the Chinese would not have transcribed the Sanskrit word
Antarava by 4^ J3 j|| (Julien, no. 1713), of which the old sounds
must have been an-tat (for tar) la (for ra) pak (for vtfA); nor the
island Ternate by J3 $5 'fljj Tan-na-te (Chin.-Malay Manual, Chap.
XII), and which evidently shows that we have to read Tat-na-te
for Tarnate; otherwise Tanah-abang in Batavia would not have been
transcribed in the same Manual hy ft #R >JJt JJ Tan-na a6an^.
Instead of Jf. sometimes qH To, ancient sound Vat, is used
11) Eitel's //ty in both transcriptions is of course a misprint.
&Kffl!&itf*iE#B<**> IB B #J W # ft <**>-
Dr. Hirth (o. c, p. 222) erroneously reads the last character in f0 /an, which
ought to be read tat-tat and not tat-tan = Tatar and not Tartar; the find I of
<o< only indicating that the second syllable begins with a / ( ^>F1 13 /or). It is the
same case as with the transcription of the Skt. word Tathdgata, by
which we have to read Tat-t'a-gal-ta, and of the word Vivadatta, transcribed
jj^j , and which is to be read Di-va-dat-ta, and not tarthdgata or deeadarta.
as in ^ ||i ttU (ft Sam-mo-t'at-t'a for Samotata, name of an

ancient kingdom close to the sea at the mouth of the Brahmaputra;
or in the name of the 1st month of spring Chditra, transcribed
$Hj nM ?H Tri (Canton Chai) Vat (for tar) la. T'at-la represent
ing tra.
The word Sukhitara ^ |p| 1|lJ3 j|f , has been correctly trans
cribed Sou-k'i-ta-lo by Julien (p. 77 and p. 191, N. 1681; also
in Nos- 1680, 1682, 1683, 1684, 1685 and 1686). But on page
78 he makes again the same mistake in reading the characters
?H iM ^f|] (Agvayoudja) cAn-chi-fo-yu-ch4 instead of Oh-shi-
fo-yu-shi (ancient sound Atsippakjuja). For the character ^ is not

pronounced an or ngan, but Oh (at, as in Amoy): ^[|
1=f" Vide J|j in K'ang-hi. At-sip, contracted, represents op,
for, as the Chinese have no words ending in s (sh or p), they are
obliged to join to the first syllable another beginning with an ;
and so suchlike words in our Manual are transcribed t + f, as
f. i. in kurus (lean) ^ ^ ku-lut-si, bagus (beautiful) ^ ^
Jj^ ba-gut-si, bdlas (to return, reciprocate) gjfj ba-lat-si, etc.,
wherein < -(- represent final s. An-shi never could represent ash (ap),
but a<-At does.
In the chinese transcription of the word Arbnda !$f| ^ |J ,
we have to read these characters at-pu-tha, at standing for ar
(Cf. Eitel, Skt.-Chin. Dict., p. 12 6).
As had two sounds tan and tat, so jpf. tan must have equally
had two sounds, as appears from the characters in which tan enters
as a phonetic. So the character had two sounds : tan ( ~J*
and to (TUSK ^f!#J, T Witf. * Witf)
The character |p is equally pronounced o(~J"pfj>^ :5ft 1^/ )-
Consequently the character ^ to is employed in the fljj j) gj| j|
syllabary to represent the Sanskrit syllable tha (Julien, N. 1727), which
proves that the ancient sound of ^ftl was ta or t'a.
20 G. SCHLKGKL.
When (as in N. 1707), j|[ j|0 represents tram in SUtrarn,

Ip- must have sounded for tat-lan becomes tal-lan or tar-
ran = (raw. In the A.moy-dialect ^ pa< ldng (another man)
is always pronounced as pdl-ldng, a general feature, as in Dutch
where the old word potlepel (ladle) is not only pronounced, but now
even written, pollepel ").
9. The same feature is observed with regard to the syllable

pan ($) in 5ftJ j| ^ pan-li-fah-lo-choh-Ua (pari-
vrajaka); ^[ yj| pan-lo-sah-mi (parasmaipada) ; ^ ^ f>an-
;oA (pradjfla); ^ ^|J f jjj4 pan-lah-mih-ti M) (paraiuiti). (Fan-yih
Ming-i, Chap. IIl, fol. 18 recto).
Page 47, XIII u, Julien tries to explain this feature by saying
"la finale w s'elide devant un l qui doit se chauger en r".
This explanation is wrong and ought to run; "final n before l
becomes l"; pan-li becomes pal-li (= par-ri), the consonant being
doubled according to the constant rule, for pari; pan-lo becomes
pal-lo (- par-ra = para), or, as in pradjfla, pal-lo for pla - pra;
pan-lah becomes pal-lah = pdr-ra = para and so forth.
Now this final n or l must represent an older t, and this is
indeed the case. The j ^ jg| says that ^ pan is also read pat IS)
13) Pott, Etymolog. Forschimgen, I, 193, II, 276, adduces as such examples of as
similation: Latin polltre from poT -\- valere ; II, 47 parricida from palricida ; French pourrir
from lat. putrere; II, 223 Greek a<rrip,'da-Tpo, Latin jiW/a for tterra, Goth- ttairnd; I, 245
Greek /imAit, Latin meW, Sanskrit madhu. I may add pellectio (reading through) from
perlec/io, pollentia and potentia (power). Professor Kern tells me that first changes to rf/
and then to W. It is a feature occurring in more or less degree in all languages, among
others in English, as Porridge and Porringer from French potage, what is boiled in the
pot and poltenger, a vessel for holding porridge.
14) This is evidently the name meant by St. Julien under N. 1340, but where he has
transcribed it, by a slip of the pen, pan-jo-mrti. Page 47, $ XlII D he transcribes correctly
pan-lami-to, where to ought to be read ti. The word pftramita is transcribed in the Fan-yih
(Chap. Ill, fol. 14 verso) ^jj? ^ | po-lonoh-to (old sound Pa-la-mat-ta).
15) >f$iC stands for pat in patera (almsbowl), in Pali /m//0, in Singh pitara, trans
THE SECRET OF THB CHINESE METHOD ETC. 21
as in the Sanskrit word pradjfia, which means in Chinese "wisdom";

whilst % is to be read like 5g jia ( J$ ^ ^ fr jgg ^ ,^
W it? f(S o iS1 ^ ^ )- According to St. Julien (p. 47), we have
to read pan-la-jo ( jijjij y^*)> which was read pat-lat-jia at the
time of transcription. This pat-lat became pal-lat in pronunciation
(as pdt-ldng in Amoy becomes pdl-ldng); and / representing in
Chinese transcription r, we get regularly plat = prai = prac?.
The character ^ stands always for ^ in transcription of
Sanskrit djna. During the T'ang-dynasty this character was pro
nounced jia ( ^\ "tjfj ^ pj ), a pronunciation it has retained
to the present day in Amoy. The author of the Malay-Chinese
Manual uses it exclusively to represent the Malay letter dj (English ;),
as in Madjalengka R,^ ^ jffe- jjfl Ma-jia-littg-ka; in Gadja l6) (an
elephant) j/ ^jr Ga-jia; in Djamur (mould) l^j? ^ jia-mS; in
Djagung (Mais) ^jjr jia-gong ; in Wddja (steel) Jfci ^ oa-jia, etc.
Consequently we must admit that the ^ in the other words
was pronounced pat, and not pan, so that we have to read ^ ^HJ as
pat-li = pal-li = par-ri = part; Jj j^lj 38 pat-lat = pal-lat - par-
rat = para, and ^ ^ as pat-la = pal-la = par-ra = para, and
so in all cases where corresponds to par.
10. Dr. Hirth's identification of ^ with Parthuva ) is

quite probable. But he has made the mistake in reading these characters
p'an-tou, and has overlooked the fact, that, in names of places,
is pronounced po and not p'an (^jf yj jjf/ according to the
^ gft; $r ft W according to the jE ^ . ^ # ^). So in
^flf* 1^ > which is to be read Po-yang and not P'an-yang. ( Vide
cribed ^ ^J, pat-tala ; *||| ;W-Js becomes /ira by passing through pal-la
= parra = pra in pratenadjit (Julien, MeHhode, p. 75).
16) From Sanscrit Gaya, which is transcribed in the Fan-yih, VI, 11 recto ffll J^JJ ffa-ya.
17) China and the Roman Orient, p. 139.
22 G. SCHLEOEL.
Po-wu ^f) was a place in Chao (j|j|), Vide 1 ^

The dictionary M ##^<.X#^,^#^:
This Po-wu was the name of the present town of P'ing-shan,

department of Chin-ting, during the Han-dynasty ( ^ ^ jfrp ^
T Uj jr in & # ^ . # 9 # iff , ra ^ *te * ). Biot,
Dict, des villes de la Chine, p. 161, writes consequently the homo-
phonous characters ^jff ^ o
As a familyname 9^ is equally pronounced Po (Vide J
jM) ^ ^ p'o-p'o means "valiant". It has the sound p'o ac
cording to the JE g| (# f$ #K # When we now
remember that jjjlj ffi p'o-lat-sz, generally contracted to
p'o-ss, represents Parsa (Persia, Eitel, Sanskrit-Chinese Dictionary,
p. 91), there is not the least doubt that ^ , which we ought
to read Po-tow (ancient sound Pa-tao), represents Par-thuv-a.
The result of our enquiry is thus that n is not used for r in
the transcription of foreign sounds by the Chinese in China itself
and that all these characters sounded originally at (- ar) and not an.
11. As for the character j^j , whose old sound at the time
of the transcription was at I8) (in Amoy hat, but as phonetic in
-jj^, at), I have already shown (T'ourtg-pao, VII, p. 189) that it is
simply the equivalent of the aspirated letter R (at - ar) as in
Manorhita jfc Jfo %\\ fi mat-nu-r-ri-ta, where represents an
22 and is introduced in order to show that jjilj ought to be pronounced
with an aspirated r : rhi. In fact J^j at (ar) represents a vibrating
r or double r, as e. g. in ||| ^ R-rd-jia (Radja), Jfjxj J| ffi
R-ra-hu-la (Rdhula); for the other transcriptions: j|j Jjj^ ^
18) Seduced by Mr. Hirth's authority, I have formerly accepted that n could be a
substitute for r {Toung-pao, VII, p. 190). I have now proved that the characters
and )|>JjJ were also pronounced pat and tat = par and tar.
i9> % m i wm mx x m i w ^.
la-hou-la and ^ tifa Jjp la-hou-la represent the second form of

rahula Ldghula. The Chinese, having no R, used the trick to place
the character J^j at = ar, before la-hu-la, to show that the word
was to be pronounced Rdhula. The same is the case with Jf^j jjJjJ
Ifi M (Wj ^) (Fan-yih Ming-i VIII, fol. 15 verso) which we read
at-lat-tan-na - R-rat-tna. Elsewhere (Chap. I, fol. 13 verso) this
word is transcribed simply jjjjj |;J3 lat-tan-na = rat + tna (Comp.
Julien, N. 1711).
The characters ^ ^ represent garbha, meaning | ; they
also occur in the transcription of A gmagarbha ( |$flj" ||| ^
"emerald". From this we see that the old sound of Jfft$j was gat = gar.
The characters for Sarvarthasiddha ^ ^ # |J
ought to be read sat(sar)-pa(vd)r-rat-t'a (= rtha) ,0) siM'a (= siddha),
the chinese transcriber having read Sar-vd-r-rtha-sid-dha; which
shows at once that Julien is wrong in taking to represent an a
(N. 368) instead of taking jfxj po-at as representing vd-r (vdr).
The ancient sound of was hat, as in Amoy-dialect. It was used
for har in harali, a kind of fine texture, transcribed <jjj] jj^jj
hat-lat-K - har-ral-li, correctly given according to the law of making
the preceding syllable end with the initial consonant of the next.
The second, beginning with r, the first ought to end with r and
the third beginning with l, the second ought to end with /. Now
final t in Chinese answers as well to r as to /, as I have abundantly
shown in my "Chinesische Inschriften in Kara Balgassun".
The same is the case in the transcription of the name of king
Harsha (Varddhaua) ^ 5fl|J which we ought to read hat (har)
li (/) so (sha), and so on in other cases.
12. Many quibbles could have been avoided respecting the
80) Jalien, p. 134, N. 744 ^jj ^ for rtha.

24 B. SCHLEGEL.
phonetic value of |$Fj" , which Julien takes to represent a in Arya,

ha in Harina, and in contraction |$flf for ar in ÎrAcrw, and
Hirth (op. cit. p. 215) for R in |Sflf jj^ (Ruben?), if attention
had been paid to the ^ |jtl ^jjj , quoted in K'ang-hi, that the
character |$flf is also pronounced a ( |Sflf J)^ |Sflf ^ jj^J , j^:
o is also pronounced a(-\- k)at = a<).
This is, besides, proved by the transcription of the word Adbhuta
1% ffi M (Julien N. 295) where (ftf. represent ad, ^ Mm and
3^ ta. Properly ^ <aA belougs to the next word (Adbhuta) dharrna
(N" J?) M = !Ta- (for Dhar)-ma (Eitel, Sanskrit-
Chinese Dictionary, p. 3 b; Fan-yih Ming-i, Chap. IX, fol. 17 recto).
It is curious that Julien has not given this value to in his
"Methode" s. v. Ngo. We further find it for the value at in the
transcription of Atdli flflf H ^ At-t'a-li; Atata ffi fl R At-t'a-t'a,
Atchdra |$flf ^ff ^ At-tsiat (for tsiar) lo (for ra); Atcharya |$flf
T^lJ 1$ At-tsia-li-ya\ Atharvana I$Bj" fH |H f^l At-Vat (for M
p'o-na; Atyanvakela |$flf ^ ^ ijgj ||| , ^Miam-p'o (for a) <'t (for
ke)-lo (for /a), in which latter word ^ is to be pronounced t,
which was its old sound, as appears from the Amoy-Colloquial,
wherein the characters , , jffe and ^ , pronounced tsi in
Reading, are all pronounced ki. In jW jfet, "a harlot", the value
of the phonetic j reappears. Also in ^|fj ^ ffjt H{jj
At-tsia-to-siet-tut (for tfur) (for rn), for -4d/<f<apa<rn; in |$flf
$2. At-tian-ti-ka for Atyantika; in |$flf i& ^ ^ -^Mt-zhi-to
for Atigupta (pt becomes by assimilation <<). The sound at for
is proved also by the transcription of Alni or Ami, part of ancient
Tukhara, by |$flf ^|J ^ J</t (for ^4/ or Ar) and nt. In all these
cases, the law of repetition of the end-consonant with which the
next syllable begins is observed, which fact has been overlooked by
Julien. (See T'oung-Pao, Vol. V, p. 173).
|SJ is thus unisonant with J^j and represents * = Rr; and,
consequently, Aruna is correctly transcribed by |$flf jjfr At-lu-no

for ^r-ru-wa; hdriti or .dri<t by |Sjf. ^HJ Jgj At-li-ti for Jr-W-<t
and .drAan by |$flf, ^ j?H at-lo-han for Ar-r'-han. According to
tbis rule, the name mentioned in tbe Syriac inscription at Si-ngan
foo cannot be Ruben, as Dr. Hirth advocates, but must be Ar-r pun
which, already 22 years ago 11), I have stated to represent Orban
or some such name.
In the transcription of the word Rupya (silver) |$flf. Q
(Fan-yih Ming-i, Chap. VIII, fol. 13 recto), we must read it as
R-ru-pa = Rhupa. By modern chinese authors it is transcribed
P/H P$ lu-p'*eh and j| 2 lo-p'i - English rupee.
The example of n for r in the chinese name for Pulo Condore
Jj&j- k'un-lun (10th century) and J| k'un-fun (A.D. 1292)
is out of place.
Father Gaubil, who visited the larger island of this group in
1721, states that the natives called it Conon (Gaubil, Lettre de
Pulo Condore in P. E. Soc. Obs. Math. Astron. 1729). Crawfurd
writes it Kohnaong, stating this to be an Annamese name JJ).
Bretschneider (On the knowledge possessed by the Ancient Chinese

of the Arabs etc., p. 15, note) rightly says that the chinese name
seems to be derived from the former name Conon (Con-non) K'un-
lun = Kun-nun "), exactly as ^& fls||5 Pinlang becomes Pinnang
(island of Penang).
21) Inaugural lecture at my nomination as professor of Chinese language at the

University of Leiden. B. J. Brill 1877.
22) On the map of Annam, joined to the Vocabulaire Latin, Francaia, Anglais et
Cochiuchinois, from 1838, Pulau Condore is called Hon C6n nSn (Hon means island in
Cochinchinese).
23) Condor is the Malay name Pulau Kundvr (pumpkin island). Now the word
Knndur (pumpkin) is transcribed in the Malay-Chinese Manual
If the Chinese did not tell us distinctly that ""J- |J ^C|J Kan-to-li, also written
jjjf jjjjj Kun-to-li, waa the old name of Sumatra, these characters would much better
render the name Kundur.
26 G. SCHLEGEL.
With respect to the other transcription of the name of the

island of Sumatra (see above p. 18) ^ |^|J Su-men-tah-lah,
final n in men does not stand for r, because these chinese characters
are not the transcription of the word Sumatra, but of (Dttr) Samund
a corrupted form of Dvara Samudra, "Gate of the sea", the name
of the capital of the Balalas, a mediaeval dynasty in S. India. This
name was corrupted by the Arabs to Dtirii Samundur, Bur Samun
and Bhur Samund, and even in Anglo-Indian to Doorsummund
(Cf. Yule and Burnell, Glossary of Anglo-Indian words) M). The
Chinese, who probably heard at Sumatra the corrupted name Sa

mundur from the Arabs, correctly transcribed it Samundara or
Sumundara (Su-mun-tah-lah).
We thus see that final n never was used in Chinese transcription
for final r and we may safely dismiss the question.
13. The example quoted by Dr. Hirth of ^ J|| lam-mo for

Dharma (in Dharma nandi) is no proof that final m stands for r,
for Je^ |p is not the transcription of the sanskrit Dharma, which
24) If Mannert (Geographic der Griechen unci Romer, V, 288) ia right in identifying
the name nA, Zifimiu (Palay Simundu) of Ceylan (Tapobrane) with Polu Simundu,
"the island of Simundu", 1 am inclined to read Pulu Samundur; but then it would refer
to Sumatra, and not to Ceylan, because pulu is Malay and not Sanskrit; for, as Mr. ran
der Tuulc has rightly observed, this word cannot be derived from the Sanskrit Pulina, as
Von de Wall did; this word meaning an alluvial island, or the bank of a river.
If Tapobrano was Sumatra, the distance of "20 days sailing south from India" which
the greek ambassadors learnt at 1'alimbothra would be more reasonable.
At all events it is remarkable that none of the ancients, nor Ptolemy himself, speak
of the cinnamon in Tapobrane for which it has always been renowned; whilst they called
a whole tract of land upon the cast-const of Africa Regie cinnamomonifera. If Ptolemy's
Palay Simundu is Sumatra, and not Ceylon, their silence would be justified. We now all
know that the name Taprobane was applied during the Middle Ages to Sumatra. It is
transcribed in Chinese ^jjf P^ to-p'o-lo-bun for Ta-pro-ian, which Porter Smith,
p. 58, erroneously identifies with Turkoman. The char. p^ are also used to
transcribe the word brdhman (Cf. Eitel, Skt.-Chin. Diet., p. 27).
the Chinese transcribed ^ ^ or ^ jpjj, Tat-ma = Dhar-ma 15),

but that of the Pali-form Dhamma, without medial r ,6) ; and, in
the second place, R is often suppressed in Chinese transcription as
appears from the Malay-Chinese Manual mentioned above.
We find therein:
for Kota ardjo rfrj ^ >5* kota adzio;
Suko ardjo 55 su^ dzio;
> Kartasura "fjfl ypj Jjjjk katasulah;
Martapura J ^ matap'ulah ;
> Modjokerlo ijjff ^ ^ ^ bodziokoto and JlflJ 3H|
Purmerend (Island) is transcribed ^ R,^ pongmalan, and

we know that Parsa is often transcribed p'o-sz, of which
Hiuen-ts'ang given the fuller form ^ p'o-lat-sz = pa-r-s
or iâr = Persia.
Likewise the Java-Chinese transcribe the Dutch word Kantoor
(Malay Kantor, "office") ^/ Kanto, without final r, and the
ward Noordwijk in Batavia, called Norbek by the Malays: j&
Ndn-bik (pr. gnawbick), without medial r, this being replaced by the
nasal sound of non. In sanskrit transcription r is also suppressed
before n, aa in Purna, transcribed ^ pun-na (Julien, N. 291).
In Malay we find sometimes two forms for one word, the one with,
the other without medial r, as in perkakas and pekakas (instruments),
pergi and peggi (to go), etc.
It is further suppressed in the transcription of the name Tanah
Datar ffi ffi JJ^ ^ Tan-nq la-ta; in that of Pin<u Pdgerman
25) So the name Dharma Samnaha is transcribed Jjjjji '^J? |Slf Tat-ma
Hana-ha. When Jjg is used for tama, as in the transcription Ii .VY?- .t'?\
i as? w
tairma-sit-tvit-ti for Tamasthiti, it is because final / before m becomes, by assimilation, hi,
so that we have to read tam-ma and not tat-ma.
20) Eitel, Sanskrit-Chinese Dictionary, pp. 31, 32 and 96a.
28 0. SCHLKGEL.
(ia Batavia) ^ Jpj- PPJ, J!|fl ^ Pintu Pctkdban, where d stands

for ger; in Passer Serpong (western suburbs of Batavia) ^ ^
^ Pasat Sdpong, where sd stands for ser; in Tegelalur j^jj
Pyf P$T^C 7e<ra?a/ without final r; in persegi (square) Q fcHt
pasdgi. R is suppressed in the transcription of dirga ifjj j^B <i-a
or jij^ Ipf ti-k'o, where ti stands for dir; likewise in the trans
cription of Nirdagha |g, |J ^ Ni-Va-ka and Nirgantha $g ^ |J
Ni-kien-t'a or ^ jjf |J Ni-kan-lian (for graw) -<'a, where iVi
represents nir. It is always suppressed in the transcription of Mitra
^ wnta, as in Vasumitra Jff\ ||f ^ Wa--mt-<a and Mitra-
sena ^ ^ ^ Mi-ta sze-na, where ia stands for tra.
Iu the transcription of the name of Samarkand |||
Sahmohkien, the old name of this town Samakan is retained. The
modern name Samarkand is correctly transcribed by ^
Sat-ma-r-kan (Eitel, S.-C. Dict., p. 116 a). It is also transcribed
^ ^-p sih-man-kun, where the author ought to have written
with the mark of the short upper tone mat (=mar), for which
no characters exist.
The river Margus is transcribed J?| fj^ma-hu, without the medial r.
Final r is sometimes expressed in the Manual by a short tone, as
in timur (east) transcribed ^91 ^ ft'-mo with the ~f* ,A J** timdh,
for which there are no characters in the -j" ^-Dictionary.
14. By the law of assimilation p before t becomes t. This

is proved by the name Tdmalipta or Tdmalipti, which becomes in
Pali Tdmalitti, and is therefore transcribed by the Chinese ^[ |||
i? Tam-mo-lit ")-ti or P0 JlH W Fjg tat-mo-lit-ti, where final
27) The character M (chestnut) is pronounced in Reading in Amoy lik, but in

Colloquial lat; in Canton lnt. During the T'ang-dynasty it was pronounced lit (
TJ /(ik+ts)tf).
the secret ov the Chinese method etc. 29
t of tat becomes, by assimilation, m. In the transcription ^> Jj|t ^

to-mo-li-ti, the p is entirely left out.
This is the same with the word gupta in Atigupta, transcribed |$flf
J|| ^> a<-<i-/fc-<o; in Buddhagupta jfc |J ^ ^ Pu<-tfo fo'a-io
aud in Qrigupta, transcribed ^ ffiffi ^> iWt (= pn) kuk-to, where p
and before became equally <. The Chinese read gutta for gupta ie).
15. Sometimes the letter r is transposed, as in the transcription

oiKritya ^ Kit (= kir) tsia, also rendered in full by g 2p
R'it-li (= ri) <a. If one asks how the Chinese came to trauscribe
the Skt. word Qrigala by ^ Jjfl sit-ka-la, we must begin by
reading '< as sir and then transpose ir to ri when we get sri (pn).
It was quite unnecessary for Julien to say (N. 1568 of his Methode):
Si par exception pour pi dans Si-kia-lo, abreviation de Si-/i-
kia-lo. Volpicelli J0) makes the same fault, in saying ^ , old sound
sit, for cri in Qrigala. He ought to have said: sU for sir, transposed
to sW, in Qrigala. The same metathesis is to be observed in N. 1690
of Julien's Methode (Volpicelli, I.e. p. 171). Both authors say ^
(old sound top) for <a in Gautama ^ ^ j|| . Now it would be
stupid if the Chinese had chosen a character sounding top, or rather
tap (as in Amoy and Canton), for rendering ta; but as the next
syllable began with an m, they transcribed Kau-tap-ma, the p of tap
becoming m before ma by assimilation.
In no. 1688 (Volpicelli, p. 171) we find the same sound ^ dop
(or rather dap, Amoy Tap) stated to represent dd in drdvira (^
BftfclS)' This is wrong; tap stands for dav, because the character
flUj represents vi which syllable begins with a v, according to the
coustant chinese rule of transcription that they choose a character
28) In P&li gupta is read gutta; Dharmagupta is pronounced Dhammagutla, wherein
both r's are suppressed. Cf. Eitel, Skt -Chin. Diet., p. 81 B.
29) Actes dn 11""" Congres international des Orientalistes. Paris 1897, 2 Section,
p. 170 (56).
so G. SCHLEGEU
having the same final as the initial consonant of the following

syllable. We thus get dav-vi-ra = ddvira for drdvira. The r is sup
pressed, as is often the case.
When Kshetra (Julien, N. 1674, Eitel, p. 28) is transcribed
jei nM IH K'it-ts'e-t'at-la, we must read P0 as t'at (as in Amoy),
which represents the form t'ar-ra contracted to tra; it is the same
case as in the transcription of Chaitra $j|J P0 ^ chay-t'at-la.
When the Malays render the Sanskrit word mantrl, they write
menteri, and the Chinese ffij Men-tik-li 3o), because neither a
Malay, nor a Chinaman, can pronounce triphthongs like tri, mri, sri, etc.
16. It is impossible for a Chinaman to pronounce Sanskrit

combinations of letters as Kshat, Kshe, Kshan 3I), etc. So he had
to make a shift. He either suppressed the initial ksh and rendered

it by an as in Kshdnti, transcribed )J| ^ San (modern is'ten) -ti,
or in Kshatriya transcribed 5j?|J JfJJ sai-tat (= tar) -li-ya
\sat-tti-ya\; or by a ts' as in the transcription of the same kshat
by 3jf? ts'at or ts'at (Julien, nos 2131, -32, -34). Kshduma is
transcribed ^ J|t ts'u-ma (Amoy iso mo); in Kshunadeva and
Kshunahila, Kshuna is transcribed Tsao-na (Amoy <so-na).
In Kshepu, Kshe is rendered by flfl. fot (Amoy so). Or, when he
wanted to transcribe correctly, he divided the syllable, so that the
initial k was joined to the preceding syllable, as in the transcription
of Nikshepa which we have to read nik-ts'd-pa.
Both Julien (no. 1262) and Volpicelli (p. 165) are thus wrong
in saying that stands for ni in nikshepa. Prom the latter this
is still more astonishing, as he himself says that the old sound of
was nik.
80) Toung-pao, X, p. 41.

31) As less as the Malays can. The Skt. Kshatriya becomes in Malay Kasitriga or
Tjatrya; Kailttri (musk) becomes Kajiluri.
Badakshdn is not transcribed as Bada-kshdn, but as Badak-shan

ls|l jfflj Pat-tok-ts'ong for Bad-dak ts'ang.
In the modern transcription (2, ^ |_L| Pa-ta-k'ik shan, shdn
is rendered by sAa?i and ta-k'ik staods for dak.
The word Ni-sthdna is read by the Chinese as nis-thdna, and
consequently transcribed ^ -(0 $5 nt-*M (for wt) + t'an-na;
for a Chinese could never pronounce an initial st. The word is also
transcribed ^ ^ <'an, which proves that Julien's remark
that tan stands for thd before n is wrong.
In Akshayamali |$pj" ^ jfc /l-<s'a-ma, sAa is simply rendered
by ts'a (Fan-yih Ming-i, II, f. 10 recto). The syllable Aa is often
rendered by ^ ^ k'it-ts'a (Julien no. 500), which he have to read
k'(it)ts'a.
Kshamuda (Julien no. 2138) is transcribed -^r ^> ts'an-mu-to.
Julien says that ts'an stands for ksha before m. He is quite wrong,
for the old sound of ts'an was ts'am which stands for ksham; the
Chinese transcribed the word ksham-mu-ta repeating, as usual, the
final and initial m in ksham and mu.
Takshaka is not transcribed ta + kshaka, but tak -f- sAaa ^
^H. i Takshana is transcribed |lH jjjft tat-satna, the < of a<
becoming n by assimilation ; Takshagild is transcribed f* j|j
-f- saJ sild.
It was a terrible trial for a Chinaman to transcribe the word
Hrosminkan, and he saw no other way but to render it by fifc J|
^ git-lo-sit (for gir-ro-sit = Aro) wu'n, leaving even ci un-
transcribed. It was the same case with the transcription of the
word Hrish{ayamaua jjfc ^|J |J ^ 0t<-/t (for /riA) <a-ya (Fan-yih
Ming-i, XVI, fol. 4 verso); the syllable /jrwA must have sounded
anciently grish, as the A. S. gerst, gryri and the Greek xh*0*
prove ").
32) Benfey, op. cit., p. 1121 6.
32 D. SCHLEOEl..
Brahma pura is transcribed || j|| || f$ ff P'a-/a (for

Bra) 'i/> (for A or rather ch) ma pu-la; so we have to read Brachma
As it is impossible for a Chinaman to pronounce a syllable like

shpibh, he transcribed the word Anushtubh ^ i^t ^ .4-no
sut-tu-pa (Fan-yih Ming-i, XIII, fol. 19 verso); exactly as he trans
cribed the syllable stu, in Rohitavastu ^ jit IIlL ^ Jp: , by
The syllable faAin was simply rendered by IJH Win, as in the

transcription of Da-kshina (Fan-yih Ming-i, XI, fol. 32 verso; Ju-
lien, N. 2164) ^ (gg |R Ta Ts'in-na and $ % Tat Ts'in-na.
Or in pre-kshini, transcribed 7^)J Jj|^ ^ ptWi (for pir-ri, con
tracted to pri) ts'in-ni. Julien wrongly says: pour kchi
devant n". The Chinese transcribed ts'in-ni, repeating the initial n
of ni at the end of the preceding kshi(n) in order to give a precise
indication how the word was to be pronounced.
The application of the law was not necessai-y in the transcription
of Bhi-kshuni J% & pi-k'iu-ni or J pit-(s'u-7ii, because
the Chinese had already used the first two characters for Bhi-kshu,
a mendicant scholar. As well known, Bhi-kshuni is a female religious
mendicant.
Aspiration for the letter R.
17. The semi-vowel R is one of the hardest letters to pronounce,

even for most Europeans who, generally, replace it by a W. They
say bwight for bright, fawget for forget, pwick for prick. English
people generally avoid articulating distinctly the letter r in english
words, so that it is not heard in the pronunciation of words like
thorn, born, thorough, purslain, etc., which sound like thawn, baton,
thowo, pusslain, etc.
THH SECRET OP THE CHINESE METHOD ETC. 33
As well known, the Chinese can hardly pronounce this letter,

and if they do, they utter a guttural r as is now heard in Peking
as a consequence of mandchoo influence.
The Chinese of the central and southern provinces of China
replace the letter r by an A or au aspirated syllable, and by an /.
They pronounce e. g. the skt. word Prabhdratna as P'o-p'o-lat-na,
tfit lIsK jf?0 $5 i 111 which the first p'o represents pra, the second
bhd, and the character lat the syllable rat. The title Prabu, trans
cribed in full Ml P'-l (= p'a-la = p'a-ra for pra) -pu (for
6m), is rendered, in the History of Sumatra, simply by |^ p'o-bu,
where p'o stands as equivalent for pra.
Pradjdpdti is transcribed s|jj ^ ^ Jj^ pat-lo (= pra) dja-
pat-ti, but also ^ p'o-dja-p'o-ti, where the first p'o
represents pra. P'o stands for pra in ^ p'o-dja for pradjd ; the
same word is also transcribed $^ ^| ^ pat-lo-dja for pal-lo-dja =
par-ra-(pra)dja. The transcription giveu in the Fan-yih Ming-i,Y,
fol. 24 recto, jffi ^ pat-dja is incomplete; it should be jjj-^ ^ HU
pat-lo-dja.
p'o stands for pra in the transcription of Prasenadjit
iJJt Jj?f [^H P'0 (= Pra) -*s-?Jtifc (Fan-yih Ming-i, V, fol. 15 recto).
Bhadrika is commonly transcribed jj^ ^ ^ pat-t'i-li-ka,
where <'t-/t represents dri; but it is also transcribed jj^ ^ pat-t'i
and ^ ^ p'o-t'i, where <'i alone represents dri.
In siu-t'o (gudra) t o represents again dra. Dhrita is
commonly transcribed jjfc ^ ^ <i-/i (= dri) -to; but Dhrita rashfra
is transcribed ^> |tl Rt (= aVri) -to-ra-ts'a.
Dhri{aka is transcribed ^ ^ j^P <'i (= dhri) -to-ka. Maharashtra
is transcribed ffij ^}|J -(^ Ma-ha-lat-t'o where i'o stands for
(ra. In Surdshira ^ ^JlJ P^j su-lat-ts'a, fs'a (< a) stands for (ra.
Turfan is rendered in Chinese by pj^ ^ t'u-fan, where the
aspirated t'u represents tur; it is also transcribed more precisely by
3
31 u. schlegk'l.
i 'fl t'u-lu-fan, where <'u-/u is - tur-ru = <ur. The skt.

syllable aVirw is transcribed Jl ^ fu-/ (Julien, Me'thode, N. 2112).
Gridhraku{a is either transcribed in full by ffi jjfe |J jj^ Pi
fa'<-/t (= fo'r-n'< = ^n'd) -fo-/o (= dhra) -ku-fa, or simply by & ^
*'t-<u (= jn'dA) pjg or ijjjjjj for kufa, or by ^ |J f fl R
kit-lit (= kir-rit = gridh) t'o-lo (= dAra) -ufa. Fan-yih Ming-i, VII,
fol. 17 twso. Narindra (a king) is transcribed || |J Na-lan-
t'o or *{$| |^ Na-lin-t'o, where fo stands for dra. It is also
transcribed }t| |J j|f Na-lian-t'o-lo, where <"o-/o stands for dra.
The name of the Siamese usurper in A.D. 514, Rudra Varman,
is transcribed ^ ^fi Liu-t'o Pat-ma, where ^ t'o stands
for dra. The name of the B6dhisattva Bhadrapala is transcribed
>$t ^ HI Pat-t'o-po-lo, where again t'o stands for dra. In
Bhadrakalpa ^ ^ pat-t'o -kiap-po, t'o stands for dra
(Fan-yih Ming-i, V, f. 28 verso; Julien, N. 1488).
The syllable kro in ropa is rendered by Jj| k'u-lo, ^pj ^
k'u-lao, ^ffij 'u-/h or ^ J5[ k'u-lo, which last also serves for
transcribing gro in grosapam ^ for </ro, sat-pang for
sapam.
Other representatives of R.
18. I have already given, in 11 and 12, several instances

of the use of the characters J^j , ^ and |$pj" for the letter R.
But, besides these characters, the character jp] 6 (old sound at)
is also employed, as e. g. in the transcription of the name of the
Arguna-tree (Pentaptera arjuna) Ndgardjuna jpj jjifjj $P
Na-ka-at-lat-ch'iu13)-na. The Fan-yih Ming-i, Chap. II, fol. 21 ero,
33) Amoy Colloquial ch'iu, in Book-reading jft or j?.

remarks that has the same sound as 3i). The reduction gives
Na-ka (= Ndga) ar-rat (= rd) ch'iu-na (= djuna).

It is also transcribed in the Fan-yih 05 tyft J^j 05 > old
sounds Na-ka-at-ch'iu-na, wherein naka-at represents ndga-r 35).
St. Julien (Methode, nos. 1251 and 1252) wrongly says that
|^ , which he transcribes ngo, represents a, whereby the final a
of ndga would remain untranscribed, whilst he says in no. 1252
that j|! jijjlj ngo-la (at-lat) represents ar.
19. In modern transcription JqJ, or /HI are frequently used for

representing the letter R, as in the transcription of Bolor j|t ^
U or ^ jH^ W> Po-lu-r or Po-lu-la (ra = r).
By the Java-Chinese la is also used, as in the transcription
of the word orang tjh'dik (an intelligent man) Jfffo Jô ^ jiffi "j]
O-lang (orang) ts d-la (for tjer)-lik (for dik); the chinese transcriber
having expanded tjer, which he could not pronounce, to tjera.
The Java-Chinese use ^ kat for kar, as in tukar (to exchange)
1^ -Hr i-*ai: bdkar (to bake) >^ ^ ba-kat, wherefrom appears
that this character is also employed for kat = kal in kalinga ^
jUt kat-ling-ka for kal-ling-ga, with the usual reduplication of
finals and initials.
We have seen on page 16, that the character ^ is used by
the Amoy-Chinese for the transcription of ri in bediri Jlf .
In the same Manual the name of the town of Rangoon is
transcribed ^fcjj ^fe giang-kong 3B).
In my youth I have known a young lady, who could not

pronounce the letter r and substituted for it an aspirated g
3*> HI
85) Not given by Eitel.
36) It is also transcribed, more correctly, lang-gong, in the Manual; lang
being the colloquial word for jiu, indicated by the little circlet to the right of the
character.
86 G. SCHI.EGKL.
(greek %). She called ruiten boer (knave of diamonds) Ghuiten bugh
(pronounce %uiten boox)- This proves that r can be replaced by g
as in the above examples ").
Assimilation.
20. We have already had occasion to speak of this feature

in 8; and we shall give some more specimens of it for illustration.
The Skt. word skandha is transcribed Jj || |J , old sounds
sik-ken-t'a, wherein s(ik) -\- ken represent skan; it is the same case
in the transcription of the word ekandhila ^ || JijJi jjs^ sik-ken-
ti-la. Likewise Qloka was transcribed ijrjjjj j^f j^P su-lo (for glo)-ka,
and pnaganam si-ma (for gma)-sia-na, final m being
suppressed.
The word Sphitavdras was transcribed H ^ ^> ^ jjjlj jjj^J
sip-pi (for sphi)-ta-fat (for dr)-/a-w (for ra<-sw = ra); the word
sphatika, ^ jp[ Qfi &!l sik-p'o (for sphd)- ti-ka, and the word gramana,
^ HI l tjl o-/a (for pra)-wia-na.
The name Balkh is transcribed pan-li for Bal-li. It is
also transcribed JjjjE pan-lik-k'it for Bal-kh' or even Q JH
^ pa-li-hik, pali becoming, by assimilation, i?a/ 38). The name
Probolingo is transcribed by the Chinese in Java ^fj 2$ ^ |J\
put-lo (for pro)-bo-ling-go. The malay word perw< (belly) is trans
cribed ^ put-lut for pur-rut = perut.
Julien, N. 1379, says that J|| /?i (old sound pit) is used for
vi in Vimala, and refers as his authority to the Fan-yih Ming-i, II,
37) I have since discovered that the same character Jfl' gi, used in the transcription
of the malay word leiih |TfJ lap-gi, has the sonnd fli/i as a phonetic in the character
Sfi "curi's of milW". We then get regularly lap-Hi for lab-bih with the usual
repetition of final and initial b. j|| -^J ^ , K'ang-hi.
38) Bretschneider, Travellers to the West, p. 47.
fol. 6 verso. In my new edition (fol. 7 verso) this is not found;

it has J|i m for Vimala. Eitel has also ^ J|j| |j! pi-mala
for the same word (p. 167 a).
Sub no. 2270 Jnlien gives ;/oue as representing vi in Fima/a,
without giving his authority. Volpicelli adopts Julien's reading (op.
cit., p. 179), where he gives for the old sound of )^ nguet. I think
it is get (Amoy Reading goat, Colloquial godh) as is shown in the
transcription of the name of the Getae ^ jj get-ti.
The Chinese wrote thus Vimala as Vitmala, when t before m
became also m by assimilation.
This proves that the characters |p. , pit and get all had
anciently a final t.
It is the same case with the transcription of the word Upagtinya
M $t f$ get-p'o-sun-djiak which was read ut (in Canton ^
is pronounced yHt) -pa-sun-nya; by assimilation of t with p, utpa
became upa.
21. In the above mentioned Malay-Chinese Manual, the word

Semut (ant) is transcribed ^ tjjfy sun-but, which becomes, by
assimilation, svm-mut = sfanut. The persian loanword Kishmish
(currants) is transcribed ^jj |Sjjjjo Jj^ kit-sa (= Irish) mi"-si (= mish).
The malay word melili (to flow) is transcribed tyf\ ^ ^ but-
1d-1d, the final < of bit becoming l by assimilation with the initial
l of ld, in order to show that the vowel e is short.
The word bedidih (to boil) is transcribed ^ JI but-lx-U,
wherein final < of &< becomes / by assimilation with initial l of lt;
U-U standing for di-dih, as the Amoy Chinese has no d.
The malay word Bebangkas (to sneeze) is transcribed )Q
j^l^ ba-bang-kan (with the lower short tone = <f<) ')
A
89) We have here the same feature aa in gan; the syllable kan having
no representative hit in the lower short tone.
38 G. BGHLBOBL.
kdt-si becoming, by assimilation, kas. The name of Iskandar is

transcribed i^L ^ ^ Yik-sze-kan-tap-r, wherein yik-sze
becomes, by assimilation, is-su = is, and tap-r becomes tar = dar.
The malay word adas (anise) is transcribed lj{t frjjj ^ a-lat-si,
lat-si becoming las for da, as the Ainoy-chinese has no d.
Hindustan is transcribed j|fL ^ Hin-lu-sit-tan, wherein
lu-sit stands for dus. It is also transcribed Jrft ^ ^0 ZJtn-
<u-ss-<an or ^ ^ '(B Hin-tu-sz-tan, wherein lu-sz stands
for dus.
The malay loanword istal (from the Dutch stal, stable) is
transcribed 5|? ^ i-sit-tat, wherein i-sit stands for is and tat
for <a/.
In the modern transcription of Khorlos (a mongol tribe) by
R rjjf ^ jfjf koh-r-lo-sze, lo-sze stands for los. In that of the
sal?e Turushka, p|Jj P|| jj^ jg'|J , tut-lu-sit kiaiu (Fan-yih Ming-i,
VIII, fol. 9 recto), tut-lu becomes first tullu = turru - turu, whilst
sit simply renders the sibilant sh, tut-lu-sit standing for turush;
kiam stands for kam; the full form must have been turushkam.
In a Pormosan-Chinese vocabulary, the word idas (moon) is
transcribed P^f ^ i-tat-sik, wherein tat-sik becomes, by assi
milation, tass - das. Aisennas (little star) is transcribed
fâ ai-sat-na-sik, wherein sat-na becomes san-na and na-sik
becomes, by assimilation, nas.
The old turkish word bulmyc (bulmish) is transcribed ^
but (= but) mit-shi, where mit-shi = mish (Cf. T'oung-pao, Vol. VII,
p. 184).
Ispahan was transcribed ifo ,fg, jjQ ZfL t-u (for is) -pa-han;
Tashkend was transcribed ^ ^ -f1 tah-shih (for <asA) -an (= end).
Shiraz was transcribed shih-lat-sz, lat-sz becoming, by
assimilation, raz.
According to the same rule the Sanscrit word agmagarbha was
THE SECKET OF THE CHINESE METHOD ETC. 39
transcribed [Sflf jjtjjj) |p ^ a<-u (for ag) -ma k'at (= gar) -p'o,
or |$flf ^ |p ^ |gf a<-st (for ag) -ma k'at-p'o, or |jf $| Jg
^1 ^ ai-sip (for ap) -;na k'at-p'o for garbha.
22. St. Julien (Me'thode, p. 57 aud p. 113, nos. 435 and 439,
and p. 96, nos. 206211 6) has assumed that the chinese transcriber
read the foreign words etymologically, instead of phonetically 40),
and thus he fancied that the Chinese read the name of Cashmere
(Kdgmira, transcribed ^ ^ ka-siap-mi-la or j^B
ljjj| ka-iip-mi-la) as kd-girii-ra. Now the Chinese did not thus
pronounce this name, but they pronounced it kdg-mi-ra, exactly as
the English pronounce it Cash-mere, the French Caehe-myre, the
Dutch Kash-mir. In modern transcription we find ^ ^\ ^
Kih-shih (for i-esA) -mi-r. Consequently Julien is wrong in saying
(no. 207): shi-mi for pmt in Kdgmtra, joining together sAt
and mi, instead of kd and Ai.
We have only to look at modern transcriptions to see that
Julien is wrong.
Kashgar is transcribed P$ -f-f* ^ ^ Koh-shih (for 2T<iA) oA-r
(for ar); Kharashar is transcribed P|f pjjjij *J^ ^ Koh-lah-sha-r,
wherein lah-sha represents rasha.
We have thus to read ^ jjjjjjji j| ^ as Ma-Mg-var-ra and
not, as Julien does, Ma-he gva-ra, for the old sounds of || and
were jt> and fat, which could not become gvat (= far) by
assimilation.
40) Thi is a.o. proved by the chinese transcription of the malay word orang ptlajar-an
(a seafaring man). The syllable an is a malay sufflx for forming substantives ; ptlajar menus
to sail, ptlajaran tailing \ vrang pShjurun a sailing man. Now the Chinese did not care a bit for
the etymology of this word, but transcribed it phonetically with tf^j [2, ]^ B9
olang pa-la-jiat-lan for oron; palajiar-ran=palajar-rati, ilouuliup the r on account of the short'
preceding vowel: jiar-ran standing for (pr.yara). So they transcribed the word ingat-an
(meaning) ^J- ing-gat-lan, with the usual doubling of ^ and /.
40 O. SOHLEGEL.
We have to read |$flf ^ and |Sflf ^| as ag-va, and not

as a-pca; BJj^ ^ pi-sia-pu as vig-va, and not as vi-pva; and so
in all the cases given by Julien. According to the same law, we
have to read the chinese transcriptions of the word Igvara or egvara
PWt&%tk< W M # H (or ^ H) 41) as '>ara' and not
as i-gvara, which is corroborated by the transcription of the name of
the king of Malacca Qri Pdramegvara ^ ^c|J ^ ||| ^ ^ |jf| ")
of which the old sounds, during the Han-dynasty, will have been
sit-li (= sirri, contracted to seri, the malay pronunciation of gri)
po-lo (= para) mi-sit (= meg) pat-lo (= vara) ; for mi-sit can become
wlf or mig, but sit-pat could never become f as St. Julien would
perhaps also have advocated.
Etymologically the name is composed of para + ma (highest)
igvara (lord): Supreme Lord; but the Chinese wrote down the name
phonetically (as they heard it from the natives) as para-meg-vara,
and not as param-e-gva-ra.
Tg is also rendered by ^ i-sit (i-shih) in Qanaigchara
B>J\ ~J*J \(X ^ i^T |H sia-nai-i-sit-tsiat-lo (Fan-yih IV, f. 20 era>).
Julien, no. 1205, wrongly joins 7jF with Jj[ which he reads ndi,
without seeing that the chinese transcriber read the word Qa-ndi-ig
( r? ) -cha-ra. In Skt. the i of not and of ig have been contracted
to one single i. Benfey, p. 941, reads gdnaigchara, and says that
the word is composed of ganaigchara -f- a. The same contraction has
taken place in the skt. word StkdnSgvara, transcribed -j^jj, ^Jjl
/111 ' o^ sounds Sat-t'a (= SAa) -ni-sip (= nep) -fat-lo (= ara),
for 5</ana + fara (Eitel, p. 132 b). In the title of "The Sutra
of the golden, enlightened, supreme king", in Skt. Suvarnaprabhd-
sauttamarddjasutra, correctly transcribed in Chinese
41) Also written ^jfr fjjjjP ^jjf /-iiap (for iy) -p'o-fo. Fan-yih Mitig-i, Chsp. Ill,
fol. 10 eerio.
42) Cf. Toung-pao, Vol. X, p. 465.
THE SKCRBT OF THE CHINESE METHOD ETC. 41
lPt ^ ^ ' o^ sounds: Sau-pat-na, p'o-p'o-sa, ut-to-mo,

prabhdsa-uttama, has been contracted to prabhdsottama. (Fan-yih
Ming-i, IX, f. 20 recto; Julien, Me"thode, p. 82; Bunyiu Nanjio's
Tripitaka, N. 126).
Julien (no. 837) errs when he says that li-sha in irshyd (ffi
"J? i-li-sha) represents rshya; for by this junction the letter i
is not represented in the transcription. On the contrary, when we
join i and li (= ?r) we get ir -\- sha (for shya). Julien forgot that
he bad himself said (no. 436 a) that i-li stands for ?r in Irshyd.
The word agvakania is as well transcribed ^ 2^
at-sip (= aj) -/?a (= vak) -kat (= ar) -7/a, as |$flf. ijfjjij ^|J at-su
(= ap) -a<-na (= karna) ; the omission of the syllable va in this trans
cription proves undoubtedly that ^ and || and [$fJ and have
to be joined, and not ^ and jjp^ . The word agvaghosha is not read
by the Chinese as a-gva-gho-sha, but as ag-vag-gho-sha |$flif g
^ ^J?, old sounds: at-sip-pak-k'u*')-sa, atsip representing ag, pak
representing vag with final g, in order to show that the word gho
had an initial g, according to the law of repeating the initial
consonant of the next syllable. In the transcription of agvadjit
(master of the horse, Fan-yih Ming-i IT, f. 11 recto) |$flj" f ifi
old sounds at-soat si-to, we see that the chinese transcriber read
the word as as-soad (for agvad) si-to (for djit); at-soat becoming, by
assimilation, assoat = agvad, and the to after si simply representing
the final letter t of djit.
The word is also transcribed in full by |$flf ipjjg $J #]
^4<-i (for ap) p'o (for va) dji-ti {- djit), according to the Amoy-
dialect. In mandarin these characters are read O-shu-p'o-yu-chi, in
which it would be impossible to recognize the Sanskrit word.
The same is to be observed in the transcription of the word
43> S ^ 5 #J, * ti, "P"d K'<*"9-hi-

42 6. SCHI.EGEI..
Agvayudja 5^ | ^ ^ , in mandarin Oh-shih-foh-yu-shay, in

Amoy At-sip (for ap) pat (for va) dju-dja (for yudja).
23. In no. 23 Julien again errs in uniting 3K ^ saN

i'a-/o, of the word Rdsl^ra J^j ^ 7$ ^ , and saying that it
represented A<ra, a combination of letters which no Chinaman can
pronounce. J^j ^ 7^ at-lo-sat (= ar-ra-sat) represents Rrdsh, and
<'a-/o fra.
The character 7^ always represents a< or sad and, consequently,
Julien is wrong in saying that it represents sha in Sarpdushadhi
j^H 7^ , whose old sounds were Sat (= sar) -pao-sat-ti. The
< of sat and <t being repeated according to the rule of repeating
initial and final consonant, of which anon.
This is abundantly proved by the transcription of the word
Upanishad by ^ ^ 5^ iu-p'o-ni-sat, the name of a portion
of the religious writings of the Hindus, and which is composed of
upa-ni-sad (Benfey, Skt. Dict., p. 124 6).
Nonobstant, Julien says that 5$|J represents here ska u).
The word is also transcribed fj? |?fi Tu-p'o-ni-sha-t'o,

where in sha-t'o, contracted, represents shad, which is represent
more concisely in the other transcription by 7^ sat.
24. Sut-tu represents stu in stitpa, transcribed

sut-tu-p'o, but also ^ su-t'o-p'o or ^ o-<ao-jt>'o;
but this is no proof that ^ ^ k<-<u always represents for
the Chinese avoided to pronounce this syllable, hard for them. So
the word vastu in Kapilavastu aud Rohitavastu was read by them
as vas-tu and not as va-stu and transcribed 2ft fat-sut (for
va) -<, or ^ 2j ^ p'o-sut (for mm) -'m. The word Anushtubh
44") Both indications of Julien, sub nos. 23 and 1205 are wrong. The; are found in
the Fan-yih Hing-i, Chap. II, fol. 19 recto, and Chap. VIII, fol. 18 recto, Julien giving
Chap. II, fol. 16 and Chap. XVI, fol. 16.
THE SECKET OK THE CHINESE MKTHOD ETC. 43
(Fan-yih XIII, fol. 19 verso; Julien, Methode, no. 1624) was read
by the Chinese as Anush-tubh and not as anu-shtubh as Julien
assumed. It was therefore transcribed |Sjf -^s. |jfjl ^ at-no-sut
(= an-no-sut for anash) tu-p'o (for tubh).
The malay word miskin (poor) is transcribed in the Manual by
SS dH: 32t (for Wits) -*tni because the Malays pronounce
mis-kin and not mi-skin. The dutch name Meester Cornells (near
Batavia) is transcribed ^ [|[ ^ 7^ fp^ min-sit-tit ko-nai-lit,
wherein min-sit answers to Mees and tit (= <ir) to ter, the word
being pronounced mees-ter and not mee-ster.
The transcription of pve<<t pwra by | ^ ^rjj sip-pui-to
pu-lo, mentioned sub no. 208 by St. Julien, has nothing to do with
the other cases, for the old sound of fei was pui (as in Amoy)
and had an initial p as sip had a final p, so that both words are
to be joined: shih-fel first giving sfei and then goi.
Julien (no. 1548) wrongly says that f|J joined to ^ represents
the syllable ska in BMskara, ^ ^ ||| , old sounds p'o-sik-
k'at-lo or p'o-sai k'at-lo, for the character ^ has two sounds: i
and sat. The Chinese read bhds -j- ara, and not 6Ad -f- skara, and
thus p'o-sik or p'o-sai gave Mas and kat-lo = kal-lo = kar-ra gave
ara, with the usual repetition of the final and initial 2 or r.
Transmutation of Consonants.
25. -K" before becomes L, as in the trauscription of valgu,
of wich Julien (no. 1462) only gives the first character old
sound pok and pak, as in Amoy Colloquial. We suppose the two
others to be ^ , which would give pak-lik-k'u = vak-lik-k'u,
where final k of vdk becomes / by assimilation with initial / of lik,
giving val-lik-gu, contracted to valgu.
45) Compare the transcription of phdlguna by jjpf Iffy fpj p'o-lii-i'u-na in the
Fan-yih, V, f, 28 w/o.
44 0. SCHLEGEI,.
K before L becomes R, as in the transcription of Hebrew (Hebrafa)

jfrj U 2f Bi-pik-lai = Hi-bir-rai.
26. Final K before P becomes L, as in Bhadrakalpa |J

31$ p'at-t'o-kiap-p'o - bhad-dra-kal-pa. Kalpa is also transcribed
$3t kiap (old sound probably a/>) -p'o for kal-pa.
27. X- before jT becomes !T, as in the transcription of uttara

M z&W. yk-to-lo (Fan-yih, XVIII, f. 10 vero). Julien (no. 2261)
wrongly says that yuk 46), which he reads yeou, stands for u in
uttara; for this word is also transcribed ^ ^ ^ ut-to-lo and
PS. "M. $j| ut-tat-lo (for <ar-ra = <ara) (Fan-yih, II, f. 29 redo
and f. 18 vero).
Julien (no. 2268) again says " ^ pour ou dans outtara". Now
<o stands for <a and lo for ra and thus the final t of ou< would
have remained untranscribed. But the old sound of ||| was ut, as
to the present day in Amoy, and so ut-to-lo gave quite distinctly
back the Skt. ut-ta-ra. The pronunciation ul is confirmed by the
transcription of utpala ^ jjs{| ut-pat-lo (= ut-pal-la = pa/a)
and of utsamga ^ ffli ut-savg-ka.
Uttara is also transcribed ^ jp. ^ u<-<an-oa<, which proves
that tan had also the sound tat 4"). Ut-tan-oat thus stands for
u<-<ai-va< = nMar-va<.
28. before 5 becomes S, as in the transcription of the city

Bosna (modern Ghuznee) ^ ^ Bok-sit-na; hok-sit becoming
hossit = hos.
29. N before M, B or P becomes M, as in the transcription

of Rama by |j| ^ Ldn-md for Lam-ma - Ram-ma - Rama; in
46) is pronounced yu in Canton, /to* in Amoy, and yui in Mandarin. Julien's

reading yeou, without a final consonant, is quite wrong.
47) Vide supra, p. 20.
that of Avalambin ^ U-lan-p'un (to hang down, be in

suspense); in that of Bimbisara, transcribed ^ ||| pin-p'o
(for bimbi) -sa-lo. Fan-yih V, fol. 15 recto; iD that of Lumbini jjjjjj
^ Jjfa Lun-bin-ni; in that of the malay word tumbuk djirawat
(pimples in the face) ^ ^ ^ Jjj^ 0 tun-bu dzia-lah-oat; in the
malay words sumur (a well) -J*^ ij^J sun-but, and (an ant)
^ ^ ?<n-6u<, both for sum-mur and i-mu<; in that of Sum-
pah (to swear) ^ Q sun-pa. The Chinese having no syllables
ending in urn, are obliged to replace them by n, as in Malay 6e/um
(not yet) jfo ba-lun.
30. before L becomes L, as in Uruvilva, transcribed

life ^ iu-1d-pin-lo - u-lu-pil-lo for u-ro-in7-fa; in which
transcription the syllable ^ p'o after lo has been omitted, giving
u-ru-vil-la-va, which two last syllables are contracted to l-va.
The town of Balkh is transcribed by pan-li, which
gives Bal-li for 6a/ (see above 20); final kh being suppressed.
31. P after M becomes M, as in the transcription of the

town of Semarang (Java) ^ ^ Sam-po-lang for Sam-ma-rang,
the e of Semarang being short.
32. Afy before K or G becomes or 6r, as in the transcription

of the malay word Bakaldhi (to fight) ^ ffi ^ bang-ka-ld-i;
the province of Bagelen (in Java) is transcribed ^ ^ j|t Bang-
k'it-ling, wherein ba is written 6an^ on account of the initial g of
geien, and gre is rendered by gil on account of the initial / of len,
git-ling becoming, by assimilation, gil-ling for gelhi; the province
of Bekasi is transcribed ^ Jjfl ^ Bang-ka-si, the e being short.
Likewise the skt. word Nagarahdra was transcribed
nang-go-lo-ho-lo (Julien, no. 1222) wherein ng before g
46 G. SCHLKGET..
became g: in Qa^dnkha gj *^ j^P siat-siang-ka; in Cankhara

US siang-k'i-lo or ^ siang-k'ia-lo 48); in yla<7
ffs HfU o<-o.7-*o; in Angulimdlya jSfc j? Jf|[ j|t mngr 49)-'u-
li-mo-lo, or ^ ^ iang i9)-k'ut-mo; in Kadjiiigara |^ tfjjJ! $|
^ k'at-ing-k'at-lo; in Konkandpura ffi $H $5 ijjjjj kung-kian-
na-pu-lo.
33. iV> before L becomes L, as in $g (or g) ^ ^

kiang-liang-ya-sia (Fan-yih, III, f. 10 vero) = kal-la-ya-sia, for
Kdlaya$as (Julien, N. 592). J iangr before n becomes kan, as
in Kikana ^ fS k'e-kiavg-na (Julien, no. 593).
When we now bear in mind that these characters are pronounced
in Amoy Colloquial kin", in Chang-chow kion, we see that the final
ng is changed into a nasal tone, which even disappears in the
japanese pronunciation kyoy and that thus the use of kiang for kd
becomes quite natural.
For the second case we must remark that n is a nasal n, so
that the chinese transcriber read the word kikana as ki-kia"-na.
Both cases are only to be explained by the Amoy-dialect.
It is the same case as in the transcription of the word Druma
by Jjji tung-lung-mo, which first became tul-lum-mo = dul-
lum-mo or dur-rum-mo, dur-rum being contracted to drum. (Cf. Julien's
Me-thode, XIV, E.)
34. Ng before B or M becomes M, as in Kumbhdnda 5^

kung-poan-ta or ^ kutig-poan-ta; in Kumblnra ^rj* JHJfc
kung-pi-lo; in Kdugatnbi -ffi. ^ ^ ku-siang-bi 50).
48) <q , vulgo is pronoanced i'ta in Amor-colloquial.

49) Canton old sound a^.
50) 3j@ , in Mi*Qdarin mi, it pronounced 4( in Amoy.
Bula Comba, a fortress in Celebes, is transcribed by the Chinese

3^ ffilt zfc 'tf bu-ld-kong-ba, wherein ng becomes m, by assimilation
of initial b of ba. Cp. 33, where lung-mo becomes lum-mo.
Pemambuan (a bamboo shed) is transcribed Q ^ o ^ pa-bang-
bu-an, where ng becomes m. Bombay is transcribed ^ bong-be 51).
35. iV^r before N becomes Ar, as in Qanaka ffi JbD siang-

na-ka for San-na-ka (Julien, no. 53).
The word Kurana is transcribed jjj! j| ^ft kut-lang-na, wherein
final t of wi becomes / by assimilation with initial / of lang, and
ng of lang becomes n by assimilation with initial n of na. We
have thus to read the word Kul-lang-na, for Kur-rang-na = Kurana
(Cf. Julien, M&hode, p. 59).
36. Ng before P becomes M, as in the transcription of the

malay word menumpang (to go along with), curtailed in Low-Malay
to numpang, and transcribed ^ ^ long-pang or J^|$ ^ long-pang
by the Java-Chinese.
37. Afy before Tor D becomes N, as in the modern transcription

of Pondicherry ^ -tyfc 1|| pong-ti-tsia-li (Hai-kwoh-wen-kien-
luh, Chap. II, fol. 28 verso).
38. Final P before T becomes T, as well in Sanscrit as in

Chinese: anavatapta and anavadata |$J $R ^ ^ ai-na (for
anwa = ana) p'o-tap-to and (JflJ ^ )|| ^> at-na-p'o-tat-to.
39. P before becomes also J/, as in the transcription of
51) Ib the well-known modern chinese geography fgl| fqj Jj[ ^,Cbap:II,
f. 28 tM0. The name i> evidently transcribed by an Amoy-Chineee, for the Mandarin pro
nunciation Wang-mai would not fit at all.
48 6. BCHLEGEL.
dtmane (pada) |$fj ^ ^ dt-tap-mat-ni, in which tap-mat be

comes, by assimilation, tam-mat for <ma, and mat-ni becomes
man-ni for mane; and in which word, besides, the usual doubling
of final and initial consonants is observed.
It is the same case in the transcription of the word Gdutami
(Julien 1716) f{|}!|$fi which we have to read kiao-t'am-mi.
This appears more clearly by the transcription ^ ^ J^l kiao-
tap-ma for Gautama, because tap-ma becomes, by assimilation of p
and m, tam-ma. (Fan-yih Ming-i, I, fol. 18 recto).
40. Final T before L or R becomes L by assimilation, as e.g.

in the transcription of the kingdom of VallabM ffi jjj^ fl^ fat-lap
(for val-lap) -|- pi = val-lab-bhi. Likewise in J^t pat-lo-ma for
pal-lo-ma - par-ro-mo = parama (Eitel, p. 90 a); in the transcription
of the word mrigala ^ $/fl ^ , old sounds mit-li ka-lo, wherein
mit-li first becomes milli, next mir-ri, and, by contraction, mri, a
syllable which no Chinaman could pronounce.
The word Mlechchha is transcribed in Chinese l|L , old
sounds Miet-le-ts'ia, which the Chinese read Miel-le-ch'a and con
tracted to M-lech'a. Julien (no. 1145) is thus quite wrong to say
"Mie avec ^ li forme le"; for it forms Mle exactly as in the other
transcription of the same word fjj|,j ^ ||t Mi-le (= Mle) -ts'ia and
3jj};j 1$ ^-# (= -cA'o (Fan-yih Ming-i, V, fol. 23 rero).
The kingdom of Kotlan (Kotula) is transcribed ipj fl|jj o-
tut-lo, wherein tut-lo first becomes <uWo and then tula.
The name Ullangha is transcribed 4gf jfjtt ut-lang-ka for
ul-lang-ka. Bunyin, Tripitaka, Col. 377, no. 47.
The city of Talkan or Talekan is transcribed PB jjjjj tat-lat-
kian, wherein tat-lat becomes first tal-lat, and then tal (Eitel, p. 140 a).
Kurana is transcribed Jjj || kut-lang-na, which first becomes
kul-lang-na, then kur-rang-na and finally -ra^-iya for &uraija.
Talas or Taras is transcribed p0 ^ tat-lo-sze - tal-la-sz

or Tar-ra-sz, the consonant being doubled, on account of the short
preceding a.
The word Barukacheva is transcribed jj^ jjj^ ^ , old
sounds pat-luk-k'at-sip-p'o, wherein pat-luk becomes pal-luk - par-
ruk for baruk on account of initial k of k'at ; sip-p'o becomes siv-vo
for cAea 51).
The capital of the province of Kedu in Java Magelan is trans
cribed R,^ fr Ma-git-lang = Ma-gil-lang, wherein the t of git
becomes / by assimilation with the initial / of lang. Bili ("to buy"
in Malay) is transcribed j|J but-li for bul-li - bel-li, contracted
to bili.
For the same reason Blitar is transcribed ty] Jjl ^ But-li-tat
= Bul-li-tar, Bul-li representing Bli. Blora is transcribed yfc
But-lo-la = Bul-lo-la, Bulb representing Z?/o. Singapore is called
Selat by the natives, which name properly means a "Strait". It is
transcribed by the Chinese Jj> ^ or jf ^ 5i<-/< = ,St/-toi =
<S/a<, the vowel e being short.
Tttok betong is transcribed [J|[ j ^ Tit-lo but-tong, Tit-lo
representing TU-lq, the e in Te/oyfc being short.
The malay word sirei (Andropogon schoenauthus) is transcribed
i^. ^-14, which is contracted to sil-li = sir-r4 - serei.
In the name of mountain Meru Sudargana, transcribed in the
Fan-yih Ming-i (VII, f. 19 recto) $| )g $i M Su-tat-li-sia-na,
i. e. Sudricana M), final t of tat becomes / (r) by assimilation with
the initial / (r) of li, giving tar-ri = dri. 53)
52) P||| is generally pronounced flap, but also sip ( JF^ tyj .
Vide o/W K'ang-hi.
53) The chinese transcription is etymologically correct, for darcana is composed of
drip + ana. Bcnfey, Skt. Dict , p. 3S8 b and 389 a.
50 0. SCHLBGBL.
$ 41. T is often used for the letter K, as in ^ oat-ts'a

for yaksha (Fan-yih, IV, f. 22 recto; Julien, no. 2275). also trans
cribed ^| yak-ts'a andi erroneously, according to the Fan-yih,
$L yo-ts'a, for the vowel a in yaksha is short54); < is used
also for k in the transcription of Akshamati ffi ^ ^ at-ts'a-mat.
(Bunyiu, Tripitaka, no. 74).
Likewise in Akshobhya |Sfl[ f^j at-ts'iok-p'o, for the vowel
a is short; and the Chinese have therefore transcribed the word as
ak-sho -f- bhya and not as a + ksho-bhya, as Julien (no. 2186) fancied.
Serawak, in Borneo, is transcribed J^J? dj ^ Sa-/o-o<tf; #<fya (a crow)
is transcribed jt^ go-kont; Lavek or Lawek, the old capital of
Siam, is transcribed |sj| /o-oa< (T'oung-pao, Vol. X, p. 292).
42. T before S becomes <S, as in Asokadatta [$flf. ^ ^ a<-

<u( (for s-< = aso) -tat (for do<); a and 'a being left untranscribed
(Bunyiu, Tripitaka, no. 42); likewise in Agoka |$flf |jjtj at-sut
(for as-sul = ago) -+- a (Julien, no. 248); in DUshana ^ jjjH
Tut-sia-sat-na = 7us-ta-san-na ; tus-sia becoming <uA = dwsA, and
the t of a< becoming w by assimilation with the n of na (Cf. 43).
It is the same case in the transcription of the word Dushkrita
^ j&Z ]w i?l ^ i old sounds tut-sit (for dus/) -kit-lit (for rt<) -<0
(for ta) and in that of Takshagild || tak-sat-si-lo or
^ ^ /If! tak-sat-si-lo, which became, by assimilation, tak-sas-
si-lo. Julien (no. 1905), unaware of the old souuds of and
took them to represent ta, and made of sat ksha; further in Bussorah
ty] tf but-sa = Bussa (Porter Smith, p. 64); in Qagdnkha || 'jp|
j^P- siat-siang-ka for sias-siang-ka = gag-gan-ka.
43. T before N becomes N. This is best proved by the different

The secret of the Chinese method etc. 51
transcriptions of the name of the park Lumbini, or Limbini, or

Lambirii, or Lavini, transcribed ^| ^ ^ , Lwig-bi-ni; ffi ^ ^ ,
Lim-bi-ni; j[ , Lam-pi-ni and ||| ^ ^ Lap-fat-ni (Eitel,
p. G2 b). Lap-fat-ni becomes first, by assimilation, lav-vat-ni, and
then t before n becoming n, we get Lav-van-tii for Lavini, the v
being doubled, because the vowel a is short.
Likewise in Buddhavana (giri), transcribed ^ |J ^ $5 |_L| ,
old sounds Put-Vo-fat-na, where fat becomes fan by assimilation
with the n of na; we read fan-na for mn-na = vana, because the
a of va is short; also in Ana (to inhale in breathing), transcribed
|$flf a<-na or ^ $5 ow-wa (Fan-yih, XVII, f. 18 ro; Benfey,
Skt. Dict., p. 18 b) 55). When ratna is transcribed lan-lo in Bahu-
ratna ^ pao-hiu-lan-lo (Fan-yih, I, f. 12 verso), we must
surmise that the chinese transcriber first changed ratna to ral-la,
and, on account of the initial n of na, he then changed ralla to
ra?ina for ra<wa; a very common assimilation of t with n, as in
kanaka, transcribed JE^ k'at-ni-ka for an-na-a, or in pan*
naga, a snake, contracted from pad-na-ga (Benfey, Skt. Dict., p. 514 a).
Ghanam is commonly transcribed jj| tan-nam; but also |=f|
'aNnam for Ghan-nam (Fan-yih, XVII, f. 3 verso). Vyakaranam
is transcribed |nj 1$ fH jfllj |$j pi-ya-k'at (= -kar) lat (= ran) -nam
(Julien, no. 1213).
Kurana is transcribed ^ J5 kut-lang-na, wherein final t of
m{ becomes /, by assimilation with initial l of lang giving kul-lang
kur-rang.
Karanam in Vyakaran-am Pfljj JJJJ jjjlj jjj^ is transcribed k'at
(= Aar) lat-nam for ran-nam = ranam.
In Latin penna stands for a more ancient pesna, which was
55) To breathe out ( [Jj ) is called in Skt. Vrtna and ia transcribed ^ $5

p'o-na. Ibid.; Benfey, p. 613 tt. vulgo /wan, stands here for pri/, as in pradjM
jgv Eitel, p. 94.
52 G. SCHLEGBL.
still earlier petna, from Skt. pat "to fly". (Beufey, i.v. pat; Pott,
Etyinologische Forschungen, Vol. TI, p. 58).
44. What is the reason that the character ^( pin was used
in transcription for the syllable vi in the examples given by Julien,
sub no. 1437, and which he does not explain?
In Uruvilvd ^ iu-lao-pin-lo, the character pin stauds
for pil, because lao has an initial /. In Vivara yfi pin-p'o-lo,
pin stands for pip = viv, whilst p'o-lo represents vara. In Vidruma
y& j|i ^| Jjj|j pin-tung-lung-mo, pin stands for pit on account of
initial t of tung, whilst tung-lung becomes lu-lu for du-lu - dru.
Julien, Methode, XIV E and XIX.
In Pippala ^ ^ ^ pin-pat-li, pin becomes pip by assimilation
with the initial p of pat (pal) pip-pal-la; the vowels i and a being
short, the consonants are doubled.
45. Final T is used for S by the Chinese in Java, as in the

transcription of the inalay word keras (strong) ^ Jjjjj kat-lat =
kal-lat = kar-rat, the r being doubled because the vowel e is short,
and final t of lat becoming s.
46. The modern chinese method of transcribing, not only syl

lables, but also single letters, seems to have been casually followed
by the ancient Chinese 50). This method is called ^ ^fj or spelling
by the collocation of three characters, from the frequent occurrence
of triliteral syllables. Thus the characters fjflf a. ^ ni, ffi t, ^ ya
and [fil" a stand for a-n-i-y-a the manchu word for year. The mongol
word k'uit'un (cold) is transcribed kit, ,fj u, "e\ /, |g|j tu, ^ u, in,
for k-u-i-t-u-n ").
. 5G) Of. the Van-yih iliiig-i, Chap. XI, f. 21 recto.

57) Edkioa, A Grammar of the chinete colloquial language, Shanghai 1857) p. 78.
THE SECRET OF THE CHINESK METHOD ETC. 53
To this order belongs the transcription of the Sanskrit word

svastika by ^ ^| ^ J( $JP , whose old sounds were sai-pak-sit
ti-ka 58), which represent s(ai)a(k)s(it) for svas; the two syllables
tika being transcribed by syllabic spelling.
Svdgata is transcribed 'fy sa for s, ^ p'o for vd, jfixj
k'at-ta for gata.
47. This law is to the present day observed by the Japanese

in their transcription of foreign words. Iu the japanese translation
of the dutch medical work Anatomia reformata (168 7) 50), the name
of S. Blancardi is transcribed ^fy ^ ^ =5 put-latig (for Blang)
kah-r-to (for cardi) because, in fact, the name Blancardi is pronounced
as Blangcardi.
The dutch name Karsten was transcribed by the Japanese
-Hi "^0 9)' which name, according to Mr. Shinkichi Hara of
the Musenm of arts and industry in Hamburg, is to be pronounced
Katsu-ni-ten, ni standing for r. Here the japanese transcriber followed
the same system as the Chinese, who would read this name kat-li-tian,
kat-li becoming, by assimilation kal-li = kar-ri = kars.
The author of the Chinese-Japanese vocabulary published by
Dr. Edkins in the Transactions of the Asiatic Society of Japan,
Vol. X [i], 1882, applies very frequently the law of assimilation 6 l).
58) FanyiA Ming-i, XV, fol. 1 verso. Julien, Methode, no. 1552.
5ti) This translation was published in 1822 under the title flu-lan Yaku-kei jJP
fill 0i ' "Mirror of Uutch TheraI'y"-
CO) Tijdschrift voor Indische Taal-, Land- en Volkenkunde van Ned. Indie, 4th Series,
Vol. VI, where, instead of -^jj ten, Mr. Serrurier baa put ^jj kiu, and reads katsten,
omitting the transcription of the character j^r| ni (= r). He seems not to have been
aware that the japanese Medembriki was the town Medemblik in Holland.
61) This author, called gjJ^ Sieh-itiun, was a native from yjj? Ting-choa,
in Pechely; but, although born in North China, he was, as Dr. Edkins remarks, in the
habit of using southern sounds to no small extent. For this reason I give the Amoy- and
Chang-chow pronunciation to the chinese characters he uses for transcription. The numbers
placed behind the japanese words refer to those given by Edkins and Satow in this vocabulary.
54 O. SCHLEGEl.
The phrase ikura ka (how many? no. 205) is transcribed by him

' Hk. ^ 'ft ,'i-*S-fot kai, the t of it becoming k, by assimilation
with ko. He pronounced the jap. word ikura as ik-ku-ra.
The word kusaru (347) is transcribed ^' fffi kut-sai-lo =
kus-sai-lo, the of u< becoming s, by assimilation with the s of nri.
The word kuraishi (dark, 17) is transcribed 'p|' ^ ^fc kut-lat-
sui, the < of /a< becomes , by assimilation with the of rat =
kur-ras-sui, and so in numerous other cases.
N before D becomes D, as in the transcription of the japanese
modote ko (to come slowly, 139) $j(f ban-t'o-tik ko for
mon-do-te (= modote) -ko.
In the transcription of the word yaseta (thin, lean, 344) ^
-j- ^ ya-sip-ta, final p of becomes t by assimilation with to,
and so we get i<-<a for seta.
In that of makura (a pillow, 254) ^ ma-kut-lai (= ra),
we have to substitute for final t an /, and first read ma-kul-lai
and then ma-kur-rai = ma&Kra.
48. A curious instance of assimilation is found in the trans

cription ^ J|l ^ tsing-li-gin for the Malay Tjiringin (Ficus
pyrifolia),. in which ^ <tw<7 stands erroneously for jip tstng,
pronounced tsin in Colloquial. Instead of Jl /i, we should bare
expected /tw^ ^ or tiangr.
Dingin (cold) is transcribed 5^1] ^ /i-în for di-gin, wherein
the n is suppressed.
Likewise the town Cheribon (jav. Tjiribon) in Java is transcribed
4fc Jjl '5C Tsing-li-bun, because the character <ingr is pronounced
tsi* in Colloquial.
Gemination of Consonants.
49. I have already often had occasion to call the attention
of philologues upon that interesting feature in chinese transcription
of foreign words 6J).
The law may be thus circumscribed:

As a general rule characters are chosen which end with the same
consonant with which the following syllable begins, when this consonant
is preceded by a short vowel.
Thus, e. g., the skt. word mati is not transcribed ||| Jg^ mati,
but ^ Jjg^ or j|& Jgj mat-ti, because the vowel a in mati is short.
When the preceding vowel is long, the final consonant is not
repeated, as in ^ | or /$fl ^ kd-t'o or kd-t'a, the transcription
of Gdthd.
This law is not restricted to the Chinese alone; it is also found
in many other languages.
Pott, Etytnologische Forschuugen, Vol. II, p. 257, says:
"Die Gemination ( Verdoppelung) eines Consonanten in der Schrift
ist oft nur Zeichen der Kiirze des vorhergehenden Vocales, weil,
einen langen Vocal vor einem geminirten Consonanten ohne Pause
zwischen der Doppelung auszusprechen, schwer fallt".
The french verb appeler (pr. appelay) has only one l, because
the preceding e is long.
But in the present, future and imperative tenses: j'appelle,
fappellerai and appelle, the l is repeated because the preceding e is
short 63).
62) Toung-pao, Vol. V, p. 171174, 1894; VII, p. 189, 1896. Cp. also J 1.
63) Une regie geWrale, et qui ne souffre que tris pea d'exceptions, c'est que qaand
les consonnes sont doublees, c'est presque tonjours parce que les syllabes qa'ellea forment
sont breves. Les consonnes qui se redoublont le plus ordinairement par cette raison sont
/, m, n, p et t; comme dans motile, pomme, courontu. Les mcmes coosonnes sont simples
dans les mots poite, dSme, trdne, tempite, parce que les syllables qui les precedent sont
longues. (Laveaux, Dictionnaire des difficulty de la langue francaise, p. 168; and p. 78,
i. Appeler).
56 0. SCHLEOEL.
The letter t is doubled after the letters a, e, o and u, as in

patte, battre.
We find further acclamation, affermir, apparaitre, etc.
In Euglish the infinitive of the verb travel has only one l, but
in travelling, traveller, travelled, the / is doubled, because the preceding
<f is short. The same feature is observed in revel and reveller, in
occur and occurrence, occurrent, etc.
50. We shall now give some further illustrations of this law

in chinese transcription.
Thus the skt. word kapotana is transcribed in Chinese ^ ^
$5 kap'po-tat-na (for tan-na). The first syllable has a final p,
because the following pd has an initial p; the third syllable has a
final n, because the 4th syllable has an initial n.
The second syllable is not rendered by a chinese character in a
short tone ( y\ ), because the d in kapotana is a long vowel.
The skt. name of the scorpion kapphina is transcribed ^|
kap-pin-na.
The name of the flower Champaka is transcribed ||| ^fj jjjfl
or fjl^l Jjjf) Tsiam-pok-ka, where pok has a final k, because ka
has an initial it.
The bisyllabic word Padma is rendered in Chinese by three
syllables ^jp fyfc pat-tik-mang ; the Chinese having introduced
the character ^ beginning with a t (d), because the preceding
syllable pat has a final t (d). The Chinese read and understood the
word as sounding pad-d-ma.
The trisyllabic skt. word pudgala is rendered in transcription by
îfi ^Nf "ftf iH pu-dik-ga-la, because the syllable pud has a final </.
The name of the mountain Kukkuta is transcribed as Kut-kut-ta
^5 H; tQe ûd syllable beginning with a k, kut-kut became,
by assimilation, kuk-kut; and as the 3d syllable had an initial t, a
THB SECRET OP THE CHINESE METHOD ETC. 57
character was chosen for the 2d syllable having a final t. Jnlien (no.
687) says quite wrongly " ^ kiu pour kou dans koukkou{a", whilst
he ought to have said "kiu, ancienne prononciation koul, pour kouk
dans koukkouta."
The skt. word dhvadja was not read by the Chinese as dhva-dja,
but as dhuad-dja and, consequently, transcribed ^ , ancient Bouuds
t'oat-sia (Julien, no. 2056; Fan-yih Ming-i, XX, fol. 7 recto).
51. Julien (no. 1706) says that jp. tan stands for ta in pu-
tana j^L^ft; this is quite wrong; the Chinese read pu-tan-na,
tan having a final n because na has an initial n, and to indicate
that the vowel a is short.
Sub no. 1715 he says that ^ fan has lost its filial n and
stands for da in Damya gij ^jjjj t'an-miao. But the old sound of
'M' was t'am (as in Amoy), aud so we get t'am-miao, t'am having
a final m because mya had an initial m.
Iu the transcription ^ ^ ^ |sf| u-t'am-pat-la (udumbara),
t'am stands for dum, aud pat-ra for bar-ra, because the vowel a
in bara is short.
Julien (no. 1718) is quite wrong to say "t'an par exception
pour doun (doum devant une labiale)". There is a great phouetic
difference between t'an and dum, but no such between dam and dum.
Madhydntika is transcribed ^ |JJ jJjD mat-tian-ti-ka or ^
BJ mat-tian-ti *).
The name of the kingdom of Magadha is transcribed
mo-k'at-t'o or |p i4 ^ mo-k'at-t'i for ma-gad-dha, final and
initial < (</) being repeated.
The word puttra is transcribed |]fJ ^ put-t'at-lo, wherein
initial < of 'a< is repeated in final t of put; fat-lo, contracted,
represents tra.
64) Eitcl, Skt.-Chin. Diet., p. 63 a, writes in both cases t^J yew instead of |JJ îan.
Cf. Julien's Methode, no. 2019.
58 G. SCHLEGEL.
The word Ardjnna is transcribed $ff J||| $5, which St. Julien,
no. 265, wrongly reads an-chun-na, instead of Oh-chun-na, old
sounds At-sun-na, in which final t of at represents r, and final and
initial n of sun and na are repeated.
In the transcription of atharvana jj^" ^ at-t'at-p'o-na,
initial t of Cat and final t of a< are repeated, and final t of |H
Vat represents an r (thar).
52. 2T is repeated in the transcription of the word Bagheldn

M pak-ka-long.
The Chinese did not transcribe the name Bhadrika, as Bha-drika,
but as bhad-diri-ka jj^ ^ ^JA , old sounds pat-t'i-li-ka, the final
(d) of pa< indicating that the initial consonant of <'t was also a
t (d); t'i-li, becoming, by contraction, dri; the name is also trans
cribed ^ ^ ^ = bad-di-ri, or simply by ^ ^ pat-t'i, repres
enting Bhadri, abbreviation of Bhadrika.
The word Vikramdditya is transcribed fji& ^pj ||| || |$flf ^
(for J^c) ^ pi-k'o-lo-mo-at-tit-to 6S), wherein 'o-/o represents kra.
53. Julien (Methode, p. 82) says that in this word J|| |$flj"
(wa-a) represents mo; but this is thoroughly un-chinese.
The chinese transcriber read vikrama-dditya, and thus |$J repre
sents at, with a final t (d) in order to point out that the syllable
dit had an initial d, dt-tit representing dd-dit = ddit; the character
tit, with a final t, was chosen in order to indicate that tyd had an
initial t. It is the same case as in the transcription of the word
Bdldditya ^ j|| |$pjf J^c ^ po-/o dt-tit-to i. e. Bdla-dditya, and of
Ciladitya f* || |$pjf ^ ^ i. e. Qila-dditya (Fan-yih, V, f. 16 recto),
65) <", *s a phonetic, was pronounced ria, as in the character tia in

a-<ia or tia-tia, "father" (in Magyar- atya, English daddy). Tia is
largely nBed in Amoy (Douglas, Diet- of the Amoy-dialect, p. 493) Pontiancti, in Borneo,
is transcribed Pun-tia-na.
and of which Julien (no. 2017) himself says that the second part
A
represents Aditya, this being one of the names of the sun 86). In
admitting the first reading of Julien, we would get ditya, which
does not mean anything in Sanscrit. It is the same case in the
transcription of Qakrdditya ^ j^D ^ |$flf ^ ^ siak-ka-lo (for
krd) dt-tit-to for aditya.
In the transcription of Ratnagarbha Jf$j ^|J ^fj ^ft $g ^
At-lat-tan-na-k'at-p'o, at-lat becomes arrat = Rrat and tan,
with final n, was chosen to indicate that the syllable na had an
initial n.
54. The word Kanakamuni is transcribed ^pj ^

k'u-nak (Canton no) -ka-mo-ni, in which wa has a final k because
ka has an initial k, or ^pj ffi ^ ^ , old sounds k'u-im-kam-
mo-ni, wherein the m is doubled in kam and mo. In the trans
cription of kokila ^fpj ^ k'n-ki-lo, the is not repeated because
the d of kS is long.
In Amala ^* J^l jj^ am-mo-lo (Fan-yih, XVI, fol. 28 recto;
Julien, no. 15) m is doubled, because the vowel a in amala is short.
55. In the afore mentioned Chinese and Japanese vocabulary

the word Imdto (younger sister, no. 63) is transcribed ^ it-
but-to, in which word the final t of it becomes b (= m) by assi
milation with initial b (to) of &ul, whilst the final and initial t of
tu< and to are repeated. In the transcription of the word kutu
(modern kuchi and kutsu, a mouth, no. 209) 'f!|' the same
rule is observed. Likewise in the transcription of kotoba 2p ^
u<-o-6o; of fa<a (pret. of ki, to come, no. 138) ^ ^ i<-a; of
gatari (to talk about, no. 151) |^ ^J" ^ kiat-ta-li; of <a<at (to beat,
66) Benfey, Skt. Dict., p. 76 b, 2. This ia confirmed by tbe chinese commentator that
vikramdditya means jfcj{ fj "le soleil transcendant", as Julien translates it.
60 G. SCHLKGEL.
no. 163) j^l ^1 lat-tat-ke; the chinese translation and transcription

gives ^ tjjt ^ Sf+ f which Satow (no. 163) reads shia tataka,
for hito talaku. But ^ ^sin^-rt represents slid, the jap. pronunciation
of the chinese ijffi siang, pronounced in Amoy Colloquial san; sing-a
giving nasal sa", and meaniug "reciprocally, each other". Tatald is
"to beat"; sho tataki is to strike one another; of hataru (to kill, no.
180) 3^ P0 J|5Afl<-<a<-/on 67); of iu<u (six, no. 323)^ ^ but-tsu;
of Tatsu (to stand, no. 118) ^ ^ ta<-<s!t 68); of riko (clever, no. 104)
")\ lik-ko, wherein the k is repeated 69). Final t is repeated in
the transcription of the phrase fu shitta (not understand, no. 183)
^ 3t Iff Put 70); in the word ikita (alive, no. 200) ^ 1fl
ihV-<rf and of kayeri kita (to come, no. 138), transcribed 'fof JfP ftg
^ Ao (= ka) -ye-li kit-ta.
56. In the above-mentioned Malay-Chinese Manual the name

of the place Magetan is transcribed RB} Ma-git-tan or jj|
j ^ Ma-git-tan, where the initial consonant of tan is found as
final in git, and in the transcription of Kedah or Quedah, on the Malay
Peninsula * |J tfiM'o, ^ $| AV<-fa and $| K'at-la.
The malay word Tek&kan kaki (hollow or bend of the knee) is
transcribed 3^ ^ ^f"- Jljfl 3j!t tik-kok-kan ka-k'i, wherein the k of <iit,
and kan is repeated because the e of ?e is short and because
kiik has a final k and a short vowel; the suffix an is a common
malay one for forming substantives.
The malay word Tamaha (greedy) is transcribed |||
67) Satow snggests hatasu (to finish); but long is pronounced ra by the Japanese and
stands for ru.
68) Comp Note 12
69) Edkins' remark that final k of lik is dropped is wrong. It is repeated because
ko begins with a *. Satow's proposed reading ^jj|J pj (sub no. 123) is unnecessary.
70) Both Kdkins and Satow have left the word unexplained ; fu is chinese loanword
Jfs put, "not"; and shitta comes from shiri, "to understand".
THE SECRET OF THB CHINESE METHOD ETC. 61
Tam-ma-ha, with fiual and initial m because the a in ta is short.

The malay word Pe-sakit-an (a criminal) is transcribed Q
j[j jfy pa-sa-kit-tan.
Mendjdngan (deer) is transcribed ^ P|| ban-dziang-gan, the
fiual and initial g of the two last syllables being repeated.
Rambutan (Rambut -f- an, Nephelinm lappuceum) is read as
Rambu 4- tan and transcribed ^ -ft ldm-mo*-tan.
Magelan (Java) is transcribed RS} Ma-git (for git) -lang.
Tangeran is transcribed ^ ||j tang-git-lan, for tang-gir-ran,
in order to show that the e is short.
K&jil, small, is transcribed ^ kit-tsit for the same reason;
foi< stands for tjil.
Sum$nap is transcribed yj| ^ ^ su-bun-lap, wherein the / of
lap becomes n by assimilation with the n of bun, the two h's being
repeated in order to indicate that the 8 is short. The Chinese read
su-men-nap.
The t is repeated in the transcription of Sukatan |^ jjy ffi
so-kit-tan for the same reason.
The k is repeated in that of Panarukan J, ~p Pa-na-
lok-kan.
The word Kuning (yellow) is transcribed ^j* kun-ling, for
kun-ning; the word Tgrang (clear) is transcribed i|| tit-lang =
til-lang - fir-rang, contracted to tgrang.
The place called Binamu is transcribed ^ ^ bin-na-mon
and not Bi-na-mqn, because the transcriber wanted to point out
that the i in Binamu was short.
In the well-known chinese modern geography yjij p|| Jjj^ ^
(Wylie, p. 48), Chap. I, f. 29 verso, Venise (Venetia) is rendered
by ,0S* Pip Bin-nin-sin, the ?i being doubled on account of the
short e in Venise.
62 G. SCHI.EGEL.
57. If the vowel of the preceding syllable is long, a character

is chosen haviDg no final consonant, as e. g. in the transcription
of the skt. word Kapotika (dove) j&P j&Jl ka-p6-tik-ka; but for
the third syllable a character having a final k (tik) was chosen,
because the fourth syllable ka began with a k, and the vowel i of
ti being short.
The skt. word Uddna is transcribed ^ ut-t'o-na, wherein
t'o stands for da; but when the vowel is short, the final consonant
is repeated, as in the transcription of the skt. word ud&nam Jfafa f^/[
pj| u-tan-nam, wherein n is repeated to show that the vowel a
is short.
The name Suri is transcribed 2p s<-/t = wr-rj, because
the vowel u is short; but the word Surya is transcribed |^
su-li-ya, because here the vowel d is long. The word Rdksha is
transcribed 1|| 3t Lo-ts'a, ^ ^|J Lo-sa< or Pj|$ [ffc -M> (for
-ts'a, because the vowel of Rd is long. But the word Laksha is
again correctly transcribed by the Chinese ^ lak-ts'a, because
here the vowel a is short.
Mdlava is transcribed Jlp ^ mo-lap (for fan) -p'o (for va)
because the d of md is long; but Malasa is transcribed ^ ^
mat (for ma) -lo-sa, because here the a of ma is short.
As an illustration of this rule, I quote the following examples:
Agama |$flf ^fjflj ^ A-ka-mo;

Amalaka |$flf. ^ ^ j|| A-mo-lak-ko ;
Jnanda |$flf ^ J-na?i-i'o;
-4rya |^ Ifft A-li-ya;
BMma flit ^ pt-/wo;
Bhutas $Tj| ^ ^-<o;
5<5a7ii p-<'t;
X>lra ^ ^ fi-p'o;
THE SECRET OF THE CHINESE METHOD ETC.
Pdrddjika ^ ^ g, p'd-/d a-i (= -*o;

Pdnjdlra Jj! ^ j|l p'd-li-i-to-lo (lo-lo = tra);
Pdfali ^ Hi H p'd-ts'a-li;
Purana pt<-/aw-9ia;
Sdgara J jjj^ sd-k'at (for ôr) -/o (for ra);
<td-/t :
Vdirochana Wftfc ff^ ^ Jj$ px-lo-tsia-ua;
Vdi{aka ^ />i-so-a;
rittMa tyl R $1 p?-Wo;

Ffnuana Bflfc $| Jfjj ^ P!-/o-sn-wn;
VUya Wlk $J pWt-ya;
Yoga Jg|f -f^fll yu-ka, and so in many other cases.
58. The rule is, however, not constantly observed. The name
Djela, in wich the vowel c is long, is as well transcribed jgjj; |J
CM-t'o, as ^ , old sound tsiat-to, for djet-ta. Mdldkufa, in
which the first & is long, is transcribed j& |[f| ^ fl mat-lo (=
mal-la for md/d) -hu-t'a. Eitel, p. 70 6.
7?w/Ai is transcribed ^ ^ pd-rt, but Bddhila ffi $|
put-ti-lo.
64 O. SCHLEGEL.
Brdhmana is transcribed jj^' pat-lam-mo for Pal-lam =

Par-ram - Brdhm, but also, more correctly: ^ p'6-lo
(for bra) -men.
The word ^m?'a is transcribed, either simply by ^ Sj|| fim-/o
= dm-la = dm-m, or by 3 syllables ^ j|f| dm-mo-lo = dm-ma-
la = dm-ma-ra, or [$flf ^ at-rnat-lo, at-mat becoming first, by
assimilation, dm-mat, wherein mat represents the letter m, and the
final t the letter r, the Chinese reading dm-mar-ra for dmra.
Shanmuka, in which the u is long, is transcribed jj ^f. j^D
siang-mok-ka - ghang-muk-ka, instead of being transcribed nhang-mv-kn
( j^J -f-J: j$fl ). Likewise dnwa (cliarity) is transcribed t'dn-na,
instead of td-va.
Dushasana, in which the vowel u is long, is transcribed ^
tnt-sia-i>at-7ia; but here tut-sia has become, by assimil
ation, tussia for dusha; sat-na is = an-na for sana.
Hiranya, in which the vowel i is short, is transcribed ffi ^
i-lan-na, where we should have expected it-lan-na for ir-ran-nya.
Kashdya is transcribed igj ^ ka-sha, where we should have
expected ?gj kat-sha, as in the transcription of Kaganna ffiy Jjjj
J|5 kat-siang-na.
Likewise we find #|5 ^ }Hf *n-na-t'i-p'o instead of ^|{| pjfl
kan-na, the vowel of a in kanadeva being short.
Kapila is transcribed j^P J^J Jj5| ka-pi-lo; but in kapilavastu it
is correctly transcribed J;(j ijsjl hip-pi-lo. Fan-yih, VII. f. 9 vro.
Kapitha is correctly transcribed J;[j -flfj, kap-pi-t'a; but kapira
JfrP ig. ^ ka-pit-si.
Kapdtatia is correctly transcribed mH kap-pd-t'at-na
for tan-na; but kapdtika, incorrectly, by j^P ^ ^5 j^P ka-pd-tik-ka,
where we should also have expected jjjjj ^ kap-pd.
When the Chinese transcribed the word mdkiha by yfc
THE SECROT OF THE CHINESE METHOD ETC. 65
mok-ts'a7>), they were quite right; for the root-word is moksh.

Of. Mokshacila (Benfey, Skt. Dict., p. 727). In the transcription of
Afdhhala by ^ mo-ts'a-lo, the long d is correctly rendered.
59. We thus see that Julien's verdict (Me'tbode, p. 50, XVI)

that the Chinese do not like to figure twice the same Sanskrit
letter, was based upon the modern pronunciation of the chinese
characters.
So he says that the Chinese wrote Valabhi for VallabM. Now
this name was transcribed 4fc Bjjjj of which the old sounds were
Fat-lap-pi. Fat-lap becomes, by assimilation, Fal-lap, and thus both
Vb are represented in the transcription; the character lap (= labh)
was chosen in order to indicate that the next syllable began with
bh; so here also the consonant is doubled.
He says that the Chinese wrote kukufa instead of kukkufa. The
name is transcribed ^jjj Jjj rt kut-kut-t'a in which final t of the
first kut becomes, by assimilation with the initial k of the second
kut, also k; thus we have kut (for kuk) -kut-t'a, where again the t
of kut and t'a is repeated; it is the same case as with kukkura,
which, according to Julien, the Chinese transcribed kukura.
Now the Chinese did not: they transcribed ffl (for^) _J^. ^
(Fan-yih, VIII, f. 10 verso) k'ut-ku-lo, by assimilation, kuk-ku-lo for
kuk-ku-ra 7i).
Julien quotes further pipala for pippala. The word is transcribed
71) Canton muk, Amoy bok, ancient sound muk, according to Volpicelli, op. cit., p.
163, no. 1156; but who, notwithstanding, thinks this muk to represent md in mSktha,
instead of m6k, and in no. 1737, where, instead of giving the old sound of U'a, he
says that it represents kiha. The Chinese did not pronounce m6-ktha but mdk-t/ia.
Likewise they transcribed the name Takthafila by tJi or (lB
i& jSI ' ^ *00D^' 'a*-*,'"""' And t'at-tt'a-si-lo, because the radical is takth.
Both Julien and Volpicelli are thus wrong in saying that Afe stands for la in
takthafita. It stands for tak (cf. ,16).
72) Cp. ( 41.
5
66 G. SCHLBQBI,.
Jji of which the old sounds were pit-pat-lo, pit-pat becomiug,

by assimilation, pip-pat, and pat-lo became pal-lo for pala.
According to Julien (no. 2268) the Chinese should have trans
cribed the word uttara as utara; but this comes that he was un
acquainted with the old sound of the characters ^ ^ ^ : ut-to-lo
(as to the present day in Amoy); thus each skt. syllable was
distinctly transcribed.
Page 82 of his Me*thode he says that in the title of the
Suvaniaprabhdsottamaradjasutra, transcribed ^ ^ f^l ^
M\ ^t' ^11 ^ ^ i wîch ê reads, according to the modern sounds,
Sieou-fa-lo-na-p'o-lo-p'o-so-yo-to-mo, so-yo is contracted to so. Now
the old sounds of these characters were sau-fat-lo-na (for Sauvarna)
p'o-lo-p'o-sa (for prabhdsa) ut-to-mo (for uttama) in which each skt.
syllable is correctly transcribed. It is in Skt. that prabhdsa-ultama
has been contracted to prabhdsottoma. Cf. 22. The Fau-yih (IX,
f. 20 recto) has, instead of fat-lo (var), ^ pat (= var), and for
p'o-lo-p'o, simply p'o-p'o, in which the first p'o represents pra and
the second bhd. Cf. 17.
The same contraction has taken place in the word Buddhoshmsha
for Buddha-usfmisha, correctly transcribed jfy fijfc |^ ^
Put-t'o ut-sit (= ush) ni-sa.
Deva-indrd, contracted in Skt. to devendra, is again correctly trans
cribed Hf gjj 0 |H T't-p'o /n-ro-/o (Fan-yih, IV, f. 15 eo).
Indra, correctly transcribed ^ gij Q | ^ mo-ho in-
t'o-lo, is contracted to Mahendra, which is literally transcribed
M i @ 1$ SI ,no AWw (for Ww) <,o-/o (= dra)-
The speech made by Buddha in Ceylon called Lankdvatdrdsutra
is composed of Lanka (= Ceylon) + Avatdra (descent) and correctly
transcribed by the Chinese 1p /fJ/U |$flf Jj^ ^ ling-ka (= lanka)
o-pat (= aa<) -io-/o (= <tfra). Eitel, p. 61 6.
The word srotdpanna, composed of srota (stream) and apatti
THE SECRET OF THE CHINESE MKTHOD ETC. 67
(oue who has entered into), is therefore correctly transcribed by the

Chinese ^ |J |$flf ]j| sut-lo (= sro) o-pal (= a~pan) -nang
(= na). Eitel, p. 132 a.
The name Djayindra, composed of Djaya + indra, is correctly
transcribed by the Chinese ^ IJJJ fgj |J ^ st'a-ya (= djaya) -in-
t'o-lo (= indra). Eitel, p. 37 a. Julien, Methode, no. 2237, is thus
quite wrong in saying that ya-in in the chinese transcription stands
for yen.
The Skt. name of Corea Kukkutegvara, which is a contraction of
kukkufa (cock) + igvara (lord), is correctly transcribed by the Chinese
H g H H M H Ke-k'u-t'a I-soat-lo. If St. Julien had paid
due regard to this transcription, he would not have made the mistake
in no. 1747 of his "Me"thoile" to contract fl t'a-i and said
that these two characters represented fe; for the Chinese did not
read kukku-{egvara, but kukkufa-igvara.
60. Such contractions are very common in Skt. Craddha-utpdda

(producing faith 'fpj ) is contracted to Qraddholpdda; Qraddhd-
indriya (root of faith -f^f ^) is contracted to Qraddhendriya;
Mahdydtia uttara is contracted to Mahdydnottara; Nara-indra, a
king of men, is contracted to Narendra; bhutamma (gold) is a
contraction of bhu-uttama; Kapi-indra (lord of monkeys, one of the
names of Vishnu) is contracted to Kapmdra. Benfey, p. 157; etc.
It is the same case with the transcription of the words mentioued
by Julien in XV, where he says that the Chinese wrote Avidha
instead of Aviddha, Budha for Buddha, viguda for viguddha, anirudha
for aniruddha, gudhddana for guddhddana, sidha for siddha, etc.
Now this remark is totally wrong, and if Julien had paid due
regard to the old sounds of the chinese characters used in the
transcription of these words, he would not have unjustly accused
the chinese transcribers. These characters and sounds were:
O. 8CHI.BGKt.
Aviddha ffi ^73)|J6 o-pit-t'o = a-vid-dha.

Buddha ffi |J p<-ro = bud-dha.
Viguddha Wfo $ |J pi-sut-t'o = vi-gud-dha.
Aniruddha ffl & $ |J atari (for an-nt = am) -/u<-<'o = a-m-
rud-dAa.
Quddhodana ^ 7A) fp| soat-t'o-t'an = gud-dho-dan.
Siddha j( ^ sit-Vam - sid-dham, etc.
Gemination of vowels.
61. Gemination or doubling of vowels, which is a very

common feature in german languages in order to indicate that the
vowel is long, as in broom, loom, boon, between, seen, tree, three, etc.
in English; in boom, loopen, koopen, room, rooven, etc. in Dutch;
in Aas (carrion), Moos (moss), Loos (lot, share), etc. in German,
is of extremely rare occurence in chinese transcription.
I have only met with a few instances:
One in the transcription of the skt. word Pradipa (a lamp),
itfe tL $C Pat (= Pra) it-i (= ^0 V i I-tsing; oue in that
of Pdrudjika ^ B p'o-lo sia-i (= dji) ka; one in that
of Tathdgata -ftjj, |jT $H |J to t'a-o (= <Aa) -ka-t'o, also trans
cribed H |$pf #g <a< fa<-o (= thd) -k'at (= #a<a). Fan-yih, I,
fol. 3 rec<o; one in that of Musalagarbha ^ Jff- i ^ ^
mu-hu (for mu) -a /o (for lag) -kat-p'o (for garbhci). Fan-yih, VIII,
fol. 13 rec<o75); one in that of Pratidegydmi jg ^ ^ J|fl
pat-lo (= pra) -<t ft (= de) -sia-ya (for pya) -wit. Fan-yih Ming-i,
78) Here only the transcriber thought of the phonetic sound which Jjj^: has in
the character ||^| , pronounced to the present day pit in Amoy.
74) The phonetic "JJ is pronounced tout in j in Canton ihiit.
75) The Tan-yih writes erroneously 1^.(^4^ ^ mn-hu-p'o-kat-lol-p'o,
with />'<> for ja and transposition of kat and fci.
THE SKCRET OF TUB CHINESE METHOD ETC. 69
Chap. XI, fol. 13 recto; Julien, no. 1106; and one in Upddhyd(ya)
HI $t |J W$ !x-p'o (= upd) -fo-ho (= dhyd).
Suppression, contraction, omission or

metathesis of syllables.
62. This occurs very frequently, and St. Julien has already
pointed out some in his M4thode:
The word i|| r% i|? (p. 4) is to be read ^ ^
Pfil > W ^5 l-U Mu-chin-Un-Co; Hoa-yen Shih-shan, which
represents the name of the mountain Muchilinda, called ( ) in
Chinese "Stony Mountain". Here the words yen and |JL|
shan are omitted, whilst and ips have been transposed. Besides
the rule of repeating the initial consonant of the following syllable
is applied in Chin-lin, wherein the n of chin becomes / by assimilation
with initial l of lin, giving chil-lin for chi-lin.
In the transcription of the skt. word cvbha stri, a virtuous
women, by ^ 5fHJ Su-po<7,)-7,)-/t (old sound), Su represents
(7u, pat represents bhas, and eit-li str\. The < of pat becoming by
assimilation with the of <. We thus get su-pae-sit-li for cu-bhas-stri.
The transcription is incomplete, for we should have expected sh-
pat-sit-ti-li 7*) for u (fu) -pas (6Aa) -ait-ti-li (stri).
The syllable mtn is suppressed in the transcription of andgdmin

|f ai-na (for anna = and) -gam.
76) Canton
77) The character ^> , of which the ancient sound was tiit, is often used for the
character whose ancient sound was til, as to the present day in Amoy {Fide
K'ang-hi, i. v.).
78) Cf. Julien p. 74, where he gives flj 5^|J nt-ttit-U for /r<, without quoting
his authority.
70 Q. 8CHLEGEL.
Da is suppressed in Ananda |SjJ" ^ d-ndn, transcribed in full

|$flf HI |Jfj d-ndn- t'o; da is also suppressed in Pindada 3j
pin-da; ddha is suppressed in Aniruddha |$flj" ^ IS ai-nt (for
an-ni = ant) -hi (for ru), transcribed in full JB, ^ at-ni
lut-t'o; ra is suppressed in Antara 7^ an-i'o; ka and medial
are suppressed in Apasmdraka jj^ Jj|| Jj^ o-pai (for pas)
md-lo (for ra).
Wheu Bhadra is transcribed jj^ pat-t'o, the aspirated <'o

represents ora. In full it is transcribed ^ j|| ^ pa< (bhad) -tat-lo
(for tar-ra = dar-ra, contracted to dra).
Bhadrika, transcribed in full jj^ ^ ^ j^P pat-t'i-li (= art) -a,
is curtailed to jj^ ^ ^ pat-t'i-li, without final a, or even to
l!pC pat-t'i, wherein the aspirated t'i represents dri.
The skt. name of the jackal Qrigala is transcribed in the

Fan-yih (Chap. VI, f. 14 recto) by ^ $H ^ sit-ka-lo = sir-ga-la.
Here sir has been transposed to sri. Cf. 15. But it may be,
perhaps, the transcription of the Pali-form Sigdlo.
63. When the epitheton of Buddha Akshobhya is transcribed

|W S3 tfit 01 simply, abbreviated, |$^" 5^ , we have to read the
two first characters at-ts'iok 79), for the vowel a in ak is short.
The Chinese have read the name at-ts'iok-p'o for ak-ts'iok-p'o 80).
St. Julien, no. 2186, wrongly says that ^ represents ksho, a
combination of letters no Chinaman could ever have pronounced.
W ^ ^> "f^J . in Mandarin ti'uk, in Amoy tt'iok. Cp. Fan-yih Ming-i,

Vol. I, (tel. 13 ero. In modern transcription the word Chocolade is transcribed ^CJJ -jjj'
Chirku-Ut or fgj ^ ^ Shuk-ko-Ut. Vide tupra, 41.
80) Cf. the transcription of Dakshina by ^ ^ tattt'in-na, Fan-fih, XI, f.
31 verso; Kitel, p. 32 4; that of VirUpdksha i.e. vi-rdpa-aksha (having deformed eyes) by
$L W /"'-'*(= wr-r)-p'o( pa)-at-ti'a(= aksha) and that of Haritaki
by ^ ^ M*-t't-K = ha-ri-tak-ki. Fan-yih, VIII, f. 2 Krn>.
The name of king Agdka (Pali Asoka) is generally transcribed

|$flf f|j j$JJ At-su (= As-su) -ka or |$flj" f $J ^4<-sw (for ^4-u) -Wo
(Fan-yih Ming-i, V, fol. 15 vero and VII, fol. 24 rec<p); but it is
also abbreviated to |$flJ ^ at-yok or at-yuk, which is unexplainable.
The character ^=f is the primitive form of the character $f , a
cicada which has not yet cast off its skin 81); the cast off skin of
such a cicada is called sot (modern shut) M). By a first clerical
confusion between j$fc ana" yuî tne chinese transcriber has
put ^ yuk instead of which is pronounced oe (sAw?) in the
characters ^jfj, and ijrj. We should then yet |Sflf at-soe",
which became, by assimilation, as-soe = ^4o, in which transcription
only final a would have been suppressed.
It is only by admitting this hypothesis, that the use of ^ yuk
for so in the Pali word kappdso (cotton) ^ j=|^ kap-p'o-iok (Fan-yih,
XVIII, fol. 14 recto, Julien, Methode, no. 2264) is to be explained. In
the other transcription of karpdsa $M ^ ka-p'o-lo, there is a trans
position of p'o and lo; it should be read ka-lo (= kar) -p'o (= pd)
with suppression of ^ sa; for the word ka-p'o-lo would answer to
Skt. kapdla or karpara, "au earthen waterpot". Cp. JJ|P |j|f for
kapdli(kas), in which transcription the syllable s is suppressed.
That this is a clerical error, is proved by the fact that the
character yuk is used for yuk in the transcription of the word
Yuktavddi (one who speaks, discusses, reasonably) by ^ ^
yuk-to-p'o-t'i (Fan-yih, X, f. 4 rc<o) and in that of the name of the
flower Yuthx (Jasminum auriculatum) by j=| jffc yuk-ti (Fan-yih, VII,
fol. 6 verso), where, according to the law in 27, final k before t
becomes t. Yuk-ti thus stands for yut-ti, just as ^ ^ yuk-to-lo
becomes ut-to-lo for uMnra (Fan-yih, XVIII, fol. 10 verso).
80 41 *| 5fc <!& ^ # f W v. if -
> 0 m F , # ft'. *fe *f J & **,

i.v. ijjft
72 G. SCHLEGEL.
Prasenadjit (Pali Pasenadi) is transcribed J Iffi [ p'o (for

pa) -ê (for se) -m' (for wad). (Fan-yih Ming-i, Chap. V, fol. 15
recto, Julien, no. 1262); the syllable di not being transcribed.
It is transcribed in full by ^ ^ pat-lo (= parra
= />ra) -se-na (for djit), or by $fc $S #f frf ^ pa<-k (=
pra) -sze-na si-to (for djit) 83).
64. The word crdmanira is generally transcribed in full by

||| 3^ ^ si7-/o (for cra) mat-ni-lo, wherein mat-ni becomes
first, by assimilation, man-ni for marê; lo, of course, is ra.
The buddhistic Dict. Fan-yih Ming-i (Chap. III, fol. 22) says
that the word is transcribed in the Tripi,aka of the T'ang-dynasty
M Jfl $Msit-li-mo-lo-ka = crimdldkaM). The Ki-hcei-chuan**)
says that he who has already been instructed in the ten precepts
(cikshdpada) is called crdmana 8e); that the (novices) from 7 to 13
years are called Khu-u Sha-mi (scare-crow (?) cramanfiras). From 14

to 19 they are called Ying-fah Sha-mi (cramaneras, who respond to
the dharma), and those of 20 years and more are called Ming-tsz
Sha-mi (titled cramaneras) 87).
We thus see that Shami is an abbreviation of Qrdma(n6ra), or
83) Eitel, op. cit. p. 95, has erroneously di for 'j1^: ri or shi.
84) This would confirm Benfey's supposition that (ri comes from (ri (to ripen), which
is found in the latin Cere).
85) This is the ^ ^[ , the record which the chinese pilgrim I-lring
sent home.
Cf. Takakuaa's I-tsing, p. LXI and the Fan-gih Ming-i, Book III, fol. 17 veno
86) Takahua, op. cit , chap. XIX, p. 96, line 22, only translates the first phrase :
"After the priest has been instructed in these (ten) precepts, he is called Qramaiyira" ; bat
be has skipped the following words and titles given by I-tsing.
rather of the Pali-form Sdma(nera); for the old sound of ^J? sha
was sd, as in the Amoy-dialect, which also appears from the
Siamese Samanen and the Singh. Sdmanero.
The feminine Qrdmanerika is transcribed in full ^ ^HJ J |j|
3-S Qrimdtialika, and curtailed to JjJ? jjf|| ^, Samini, for Samane
(rika).
Likewise the transcription ^ (ll^Jj fo'-sia-u is not that of
the skt. Gridhrakufa (which is transcribed #p Jj| |S ^ B
H<-/i/ (= gridh) -t'o-lo (= dAra) -u-{a), but that of the Pfdi-form
Ghidjakabo, or rather Ghidjaku(o 88).
65. The syllable drish in Sakdya dfishti, is transcribed j|| ?(5jJ

H> ^ tat-li-sit-ti, wherein tat-li-sit first becomes tal-li-sit, next
tar-ri-sit; tarri becoming contracted to dri whilst lit represents final
sh (Fan-yih, V, f. 10 verso).
It is represented also by jfa J] ^ ti-lik-sa in dfish^dnta (Julien,
no. 1980, 1601); ti-lik-sa = diriksa = dirish = drish.
Final ma is suppressed in the transcription of Atma |SJ ( ||t )
a-i'aw(ffio), contracted to tma, meaning "self, own" (Fan-yih, XI,
f. 28 verso).
A curious instance of mutilated transcription, which has even

baffled the sagacity of St. Julien, is that of Ldkegvara by ^ I^J
M. $tfe Lo-i-soan-lo '*).
Julien (nos. 451 and 1592) says that t stands for % in icvara
and siuen J3. for gva in igvara. Now the word is composed of
Loka + tfvara, Lord of the world, in Chinese jj ^ , contracted to
LSkigvara. The chinese transcriber ought to ha*e written ^ fyft
(/o-o) p| ^ i-soan (ig) and the omitted (war); /o standing for
88) Cf. 17.

89) Fan-t/ih Ming-i, Chap. I, fol. 12 atrio.
74 O. SCHLEGEL.
ra ; for igvara is usually transcribed f| ^ ^ i-p (for (ff)

/a<-/o (for ara) 90). It is the same case as in the transcription of
Sthdneivara $k M M M & M saUl'a (for stM) (for nif)
fal-lo (for vara). Eitel, p. 132 6.
In the trauscription of Uruvilvd by ^ |^ ^ ^ in-lo-pin-lo,
the last syllable va ( ^ p'o) is omitted ; the n of pin becomes /,
by assimilation with the initial l of /o, giving pil-lo = vil-lo or billo,
the Pali-form of Skt. vilvd.
The name of Mdiireya, transcribed in full by
moe i'a<-/e (for tre) -ya, is incorrectly rendered by ^| R?|l j| 1$
moe-li-le-ya 9I) and mostly curtailed to ^ jj^f mi-lik for ^
mi-tik-lik = mat te-re for <re.
66. Two of the most striking instances of abbreviation (if, at

least, St. Julien, Methode, p. 19 no. 14 and p. 21 no. 38, is right
in his identifications) are those of fj ^ t'a-p'o for vishtambha
and of fj^ ^ ko-ta for skandha. The first word ought to have
been transcribed tyfc }5 fl^ ^ pi-j'< 91)-'a-/A>, and the second
should be an abbreviation of ^ $JP aa (or |^ 'ok-ka, Julien,
p. 21) ko-ta.
In the first place Julien erroneously joins ^ and R and says
that these two characters represent the syllable shta, a combination
of letters which no Chinaman would be able to pronounce 9J).
The chinese transcriber joined pi and sit in order to represent
90) Cp. above $ 22.

91) Fan-yih, I, fol 12 recto; in this transcription li stands for di = ti, di-U intended
to represent trl.
92) Old sound, according to Volpicelli, fat. Canton shat, in Amoy tik with final k
instead of Snal t.
93) As less as a Frenchman or an Englishman are able to pronounce Sehl in my name
Schlegel, which I have seen written in a french and english scientific journal Schelegtl,
with introduction of a vowel e between h and l. The german province of Schletien becomes,
in french transcription, Sileiie, in english Silesia.
THE SECRET OP THE CHINESE METHOD ETC. 75
the syllable vish M) and took t'a-p'o to represent the syllable {ambha.
He must have heard tabha, m before b often disappearing in Skt.,
as e. g. dabh and dambh 5). We have thus to read the characters
flt ^ as tabha.
If ^ ^ ka-ta represents kanda in skandha, the chinese trans
criber must have heard this word pronounced skadha, without
medial n, exactly as this word, in its meaning of "shoulder", has
become in Latin scapula, without n. In A. S. sculdor the n has
changed to an /. The word is generally transcribed ^ $|| |J sak-
kan (for skan) -t'o (for dha). Fan-yih, XVII, f. 1 recto.
The above-mentioned contractions are not a bit worse than the
english alms and french aumone from the greek Ixsyniovuvvt, or the
english proxy for procuracy, or english provost, german Probst and
Provost, french prdvot from the latin praepositus, or the french
dveque (for Ivesque) from episcopus 9B).
Suppression of finals and letters.
67. This is a very common feature, probably because, in

speaking, the finals were not pronounced.
Thus the skt. word Chhdyd (shadow) is simply transcribed in
the Fan-yih Ming-i (Chap. IX, fol. 1 verso) by ^ ts'ai. Julien
(no. 2137) is thus quite wrong to say ts'ai pour tchhd dans
tchhdyd". Volpicelli, op. cit., p. 178 (64), follows in the wake of
Julien, without having confronted his statement with the Fan-yih.
Adbhuta Dharma is transcribed simply [$flj" j?j ^ Ad-bhu
9i) Cp. the word vifvakarman, transcribed fflf^ ^jjj rj-jg JjA pi-rip (vif) pak
(pak) kat (kar) mo (ma), final a being suppressed.
95) Benfey, Skt. Diet., pp. 386 a and 387 a.
96) Pott, Etymologische Forschungen, Vol. II, p. 282 et
76 B. SCHLEGEL.
Tat (= Dhar) -mo (Fan-yih, LX, fol. 17 recto); the final ta in adbhuta
being suppressed in the transcription.
Final a is farther suppressed in Ndr-iyana Jft $i Na-lo-yen
(Jalien, no. 2245).
Final in in Andgdmin is suppressed in the Chinese transcription
r$J .A<-nd-^dm, for An-nd-gdm.
Final m is always suppressed in the transcription of the word
varman |p fat (for car) -mo.
Final na is suppressed in Brdhmana ||i pat-ldm (for
Brdhm) -mo, and final a in the transcription ^ jj^ p'o-/o
(= Brdh) -mun (for mana).
Final a is suppressed in Ferghana ffij ijljl po-han or ^ ^
hok-han; in Mahdkatyaydna |p gfij j^D ^lE Mo-ho-ka-tsien-yen;
in Nibbdna yH ni-p'an or Jfjj? ^ nip-p'an; in Qrdmana JjJ?
sa-mun; in Sanghdrdma ^ -fy} ^ Sang-kd-ldm; in Chandatta jjj
^[ tsien-tan; in Chhiasthdna ^ ^ Tsin-tan or ^ <#t>n-ian;
in Buddhochinga |g yg Pnt-to-t'ing 97); in Bokhara Jff g
Po-k'at (= Bokhar), transcribed in full ^| pok-k'e-lo in the
Fan-yih, VII, f. 15 redo; in FVfroiarman rfl |$ f ]jf J'-P
(= we) -pa/: (= va) -a< (= ar) -7o.
Final an is suppressed in Mahdndman jfpj J|J mo-ho-ii<<m\
final t or m in Qakyamuni ^ j^B ^ a-a-muri.
68. Final n is suppressed in Kdloddyin ^|jf |J |f ia-tiu-

<'")! in ></a IpJ g tfo-iui-/o or |J{1} jg| Kut-tut-lo;
97) In Mandarin c/S>n^, bnt in Amny /'iV/, in Canton colloquial <aj (Wells Williams),
according to the old aonnd of ita phonelical part c/kv or tang, which gives the same
sound to the characters y^ , , and j^gf . The characters
, , Jj^gf - |>$' and are used in the transcription of tan (tang) in the
name Mdtahja (Jalien, nos. 1925, 1927, 1931, 1933 and 1934).
98) Iu Canton Ka-lau-t''o-yi.
THE SECRET OF THE CHINESE HETHOD ETC. 77
in (Jikhin f* || fikhi; in varman jj^ || pat-mo or ^ |p /a<-

mo; in Shadumdn jQ, ^ i7u (c/) -lo-mo (Eitel, o. c, p. 29);
in aamangat (soul) ^J? j]| na-man-gat (Malay), in which the
nasal man stands for man.
Final kan is suppressed in Hrosminkan Ijfc ^ ^ ^ git-lo-sit
(= #ros) -nun (Eitel, o. C, p. 45).
Final kas is suppressed in kapdlikas j^P ^ ^ ka-p'o-li.
Final in kumbhdndas ^ ^jfc kiu-poan-ta.
Final in Mallika ^ ?fH) mat-li and Jj|| ^|) mo-li.
Final a is suppressed in Aranyakah ^ ^ (for ran)
-ôA (for nya).
In Arada kdldma, transcribed |$flf. |^ JjjjJ ^ at-lo-lak-ka-lam
for ar-ra-dak kdldm (Fan-yih, V, f. 2 two), final a is suppressed.
Final ta is suppressed in Brahmadatta ^ ^ ^ Bam-mo-lat.
69. Final r is suppressed in the transcription of the roalay

word sdyur (vegetables) J|J? Jjjlô sa-io, in menir beras (broken rice)
Jp| gjf| ^ mi-Nt but-lat-si.
Final in that of manis (sweet) ^ P4|s ma-ni; in that of
Banjumas ^ Boan-giu-ma or -giu-ma.
Final / in that of Kendal ^ kan-la (for da); medial / is
suppressed in that of palm (false) fljvft pa-su, and of selderi
(celery) jJ?o [J J|I sd-tit-li; medial r in that of bertahu (to acquaint
with) ^ ^ ba-tao-n.
Medial y in that of t/awn bayam (spinat) ^ ^ flg /a-un
(for <faun) ba-am.
Final is suppressed in the transcription of Pontianale ^ ^
pun-tia-na; in analt (a child) $5 in baih (good) bai;
in orang kdteh (a small man) Jjjk ^jjf\ ^ o-lang ka-td.
Final 22 is suppressed in Sudagar darat (a freighter) y
5f Su-la-ga la-lat; in persegi (square) G, jj? jit pa-sd-gi;
78 G. 8C HI. KG EI..
in tSrlalu (excessive, too) ^ ta-la-lu; in peng-ar-ti-an

(knowing, understanding) Q JJfl ^ pa-nga-ti-an, wherein pan-
gati represents peng-arti, and an, an.
Final Sibilants.
70. The Chinese having no syllables ending in , they are

obliged, in order to express final s in foreign words, to add after
the last vowel a chinese character beginning with an s.
In this way nearly all such characters could be used, but the
most common ones are fgj, ^g, -fc, ff, 3||, jfg,
, ^, JS, 4$, , fcj?, ft, ffi,

^, ^, w. h#. j^. je&. . ^. *. p^. *a and
We shall give a few examples of each of these characters:
is employed in Skt. manas |p $5 J^Jf mo-na-2e; in Mongol

Khorlos ^[J ^ || ^ koh-r-lo-sze; in the transcription of 7aras
or Talas llB $f <a*-/o (for taWo or tar-ra) -sze.
J@L in SAt'ra2 j{|J ^ Shilahsze; in Iskander rffi

Yih-sze-kan-tah-r; in Kirghiz ^HJ ^ ^ Kit-li-kit-sze; in
Genghis Khan ffi ^ ^ Ching-kih-sze Kan.
in Pushpa ^ po-i<-p'o; in ushnisha ^ ^J?

2-'<-m-a or pjg[ ^ ^ ^ u<-<-m'-a; in Fas/{i E jj^!
yat-sit-ti (Fan-yih, VII, fol. 30 i-mo); in 7uruA H|| ^
tut-lu-sit; in Hfhh{aya(mana) g ;js|J Hp |Jj ^JJ git-li-sit t'o-ya
(Julien, no. 547); in Hrish-ma (modern Zshkeshm, Eitel, p. 47)
j |p J|t git-lit-sit (- hrish) -mo (= ma); in the name of
the monastery Hushkara ||| ^ $N ^ Hu-sit-ka-lo; in Namas-
kara ^ ^ |j| a mo-' (= wicw) -kat-lo (= ar-ra).
$g in VigvabhH jjjgj ^ pjfij- pi-sa (= vig) p'o-fu. (Fan-yih, I, f.

14 recto).
-jfc in Tashkend ~"J~" tahshihkan; in Pdramegvara ||| ^
HI po-lo-mi-sit-pat-lo; in Bishbalik ^|J /V JH pt<-
si< (= 6is/) -pat-li (= bal-li).
in Kdcmira j^D ^ ka-siap-mi-lo', in Igvara ffi ^ ^
^| t-'a/> (= ?f) -/>'o-/o (= vara).
<ff* in Cashmere ^ -ff* ^ kih-shih (= eA) -mi-r.
^ in Mahegvara || j ^ jUl (= -/a<-/o; in
Igvara PMfaM T-siP (= *) in 4fa |&J $|
(or ^ ) ash-vak or asA-a ; in Vigvakarma P9jfe ^ ^ Jjj!|
pi-sip (- vic) -pak (vak) -kat (kar) -mo; in Vigvabhu jff?
^ "nP pi-sip p'o-pu.
^| in Igvara ffi (or ? ) yj|| ^ (or ^ I-sip-fat (or a) -/o;
in pdrgva jjjj JjjSj y|l p'o-lit-sip (= paVf) -pa (= a'). Julien,
no. 829, wrongly contracts lit-sip-pak to rpua; in vigva W$fc ^
pi-sip (= w'f) -pa (= va).
in Meester Jp ^ mi"-sit (for mm) -<t<; in miskin (poor)
$B ^ iJC mi"-sit (for mt) -in (Malay).
g| in Bhdskara ^ |^ jjj^ p'o-sai (= bhds) -kat-lo (for kar-ra =
kara). Julien, no. 1428, p. 173. Julien is not consequent: in
no. 1428 he says correctly that p'o-sai stands for bhds, whilst
he says in no. 1548 that |f sai-kat stands for ska.
^| in DushkfUa ^ ^ "jjf tut-sit (for tussit - dush) -kit-
lit (for krit) -to; in Jasmin (jessamine) ^ ^ ya-sit-ming
or ?St ya-sit-mit; in which ya-w'J is = yas 98); in Hosna
(modern Ghaznee) ^ 51(5 Hok-sit-na, wherein hok-sit becomes
hossit = hos.
Encyclopedia /ji^- Jjj* , Chap. 73, f. 14 verto, and my Dutch-Chinese Diet., i.v.
Jaimijn. The name Jasmin is arahic.
80 O. SCHLEGKL.
y is much used for final by the Chinese in the transcription

of formosan words, as in idas (moon) jJ^r j^l i-tat-sik; tat-
sik becoming, by assimilation, tas-sit = das; in aisennas (little
star) J||| ^| ^ di-sat-na-sik, wherein na-sik becomes, by
assimilation, nas, and sat-na becomes san-na.
|^ occurs for final sh in the transcription of the old northern turkish
bulmyg (bulmish) ^ but (for bul) mit-si (for mish).
jjjjjy occurs for final s in Assakanna (Skt. agvakarna ffi jjijjij |J ^fl )
a<-su (for aa) -kat (= ion) -na.
2t occurs for final s in vastu z^L jtjf fat-sut-tu,fat-sut becoming,
by assimilation, fassut = vass = a.
is exclusively used for final s by the author of the Malay-Chinese
Manual in the transcription of malay words, as in mas (gold)
J|| f% ma-i; rampas (to rob) /V f% lam-pat-si; betas
(paddy) frjjj but-lat-si for ber-ra-s; tikus (mouse) ^fl ^pj"
^ ti-kut-si; bdlas (reciprocate) fjj[j ^ ba-lat-si; in fea
(quick) ^ ^ ^ la-kat-si; kitrus (lean) ^ ||| -/u<-si;
in all which cases final t becomes , by assimilation with initial
of st.
^j? occurs for sh in Kharashar P|f pjjjij fjj? ]^ koh-lah-sha-r, lah-sha
standing for rasha = raA; in kash of Kashgar k'e-sha;
in <2A g . (Cf. T'oung-pao, Vol. IX, p. 198).
occurs for final sh in pushkalavati "jS f Jgj po-sik
(=push) -kat-lo-fat-ti (= kal-la-vat-ti); in Kanishka jÔ JjjJ ^ $D
a nt-iX; (= m'sA) -*a; in Djyotishka ^ Jjgj "jS jJjP s?< (= d;y<5)
-ti-sik (= <tA) -a, etc.
jjjj represents final A in pushpa ^ pu-sip (= puA) -p'o
(=
jfj stands for in msthdna Jjfo 'jg 05 m'-sa (= m) -<'an-na
(= thdna) or ^ ^jp ni-sze (= wi) -J'an. Fan-yih, XVIII,
fol. 11 recto.
THE SKCRET OF tHE CHINESE METHOD ETC. 8l
^ occurs for g in m'fva-bhdWjfc ^ f^-pi-sia (for vic) -pu (for bhti) ");
iu Avalokifegvara ffi ^ )M. ^ % & o-p'o (= ava)
-lo-kit ti-sia (for teg) -p'o-lo (= vara), Fan-yih, II, f. 7 rec<o,
Col. 1 ; in Agva (horse) |{flf ^ a<-'a (= as-a = ap) -p'o
(= va).
stands for in Rdkshas jjj^ ^ | lo-ts'a-sa = rd-ksha + ; in
Vrihaspati ty] Rjg g^T | g Jfg (W) -/w.,a (Aa) -p'0-<t
(jmJi). Fan-yih, IV, f. 20 i*mo.
and 'fQ are used in the Malay-Chinese Manual for final , as
in the transcription of Celebes [J .1 'ffi Se-li-pdh, or in that
of fy> (fins of fish) ^ kd-pdh, wherein final is represented
by final h.
stands for sh in the transcription of the town Chdsh in Turkestan
^ B^F cAawAi (T'oung-pao, Vol. IX, p. 198). Eitel, p. 141 a.
flfjj) occurs for in sphitavdras 1f|> fj$fc ^ jjjlj flfji| sip-pi (= pAt)
-to-fat-lat-sze. Cp. 26.
|$ occurs for final in Qriyas (lucky) f* ^ jjjty st-/t (= pre)
ya-sin (= yas). Fan-yih Ming-i, VI, f. 4 mo.
pi^ stands for final c in Fmce ja* p$ bin-ni-sin ^
Jj^ H^, Vol. I, f. 29 ero): a modem geographical work. The
final e in Venice is not heard in the pronunciation; the name is
pronounced as Venneei.
J[fif represents s in Abhdivaras |Sflf ^ P|| ljj| JFflF a-p'o-tsoe (for soa)
-ra-si (for ras).
j?Ef represents * in Kirghiz ^ j. ^ |S} K'it-li(=kir)-k'it-si(=ghiz).
r? represents p in igchara ffi IU1"s1i (= 'ff) -ia~'0 (= âra).
Fan-yih, IV, f. 20 wro.
yJS, represents final s in Parasmdi(pada) ^ ^ yj| ^ poan-lo (by
assimilation paWo = par-ra = para) -sap (= ) -moi. Eitel, p. 90 ft.
' .. . ..... l.i.:J

09) The syllable va ia luppreued in the transcription. Cf. Eitel, p. 170 4. "'
6
82 G. SCni.EOEl.
R|J occurs for final ia Kas-tadguna j^D R >Jt $R ka-ts'ia (= faw)

'tq-kiu-na. Fan-yih, XIII, f. 12 iwio; Julien, p. 20, no. 26.
71. A general law in chinese transcription of foreign sounds

is to place a mouth P or a man -f before the character intended
to represent the sound 10), or to represent it by a character which
is only chosen on account of its phonetical part.
Thus they write ^ ch'i in order to express the sound dya
(Julien, no. 1823); ^ to for tyd (Julien, no. 2030) because the
phonetic ^> to has the sound tia in J^".
They wrote pg. $ij for Talkan where P0 jfcjj |H would have
sufficed. They used ^ yik (it) for <i, tit (Julien, nos. 436 a and
455). By preference the character jjfH kia is used to render skt. ka
or ga. The Java Chinese use jjjJJ instead, without the prefix of ^.
They write ^ where _J^. would have done the same service;
likewise ^ instead of ^ , ^| instead of ^ , ptf; instead of ^ i
0^ instead of , ^ instead of 1g , P|| ^jp instead of J^j Jji ,
etc. jfuji is used for ^ because the phonetical part was pronounced
tak and was chosen to represent the skt. syllable dak (Julien, no.
224, 225). ^ shui is used for |^ shuo (aoat) (Julien, no. 259);
\% eh'a is used for or fS i'o (Julien, no. 1025); $| pt is
used for || pit ( 61).
The thibetan formula Om mani padme hun is transcribed o^f
5^ PjS 9V PS! P^*i witn li^fcle mouths to the left of each character
in order to show that they are to be read phonetically (T'oung-pao,
Vol. VIII, p. 114).
Instead of the character jjjljL, whose ancient sound was ts'iak
or ck'xak ( i jjft fj o ^ #J #J) during the T'ang-dynasty, they
used the character jjL, whose ancient sound was tok (jfy "fjTj,
100) Wells Williams, A tonic dictionary of the chinese language in the Canton Dialect,
p. vu.
THE 8ECBET OF THE CHINESE MKTHOD ETC. 83
t. e. <(iok -f" Vide ^ ^ ), as to the present day in Amoy.

This only can explain, why this last character is employed for dja
and chd in the transcription of Vidjagham ^ J^L |if|J and VicMram
jji. ^ , which we have to read Pi-ts'iak-kang and pi-ts'iak-lam
= Vi-djag-gham and vi-chak-ram, etc. "")
Conclusion.
72. The present paper is intended for an introduction to a

comprehensive dictionary of all the characters used by the Chinese
in their transcriptions, not only of Sanskrit words, but also of those
of the other foreign people with which they had intercourse. In
order to avoid all difficulties of pronunciation, this dictionary shall
be arranged according to the radicals, giving to each character the
phonetical value in the transcriptions for which it is used.
For the older transcriptions, especially those of Sanskrit, and
the newer ones made by Amoy Chinese, we have to give to tho
Chinese characters the old, and not the modern sounds. For it is
impossible, except by guesswork, or by having positive authorities,
to identify the foreign words in their modern chinese garb.
1 shall only quote a few instances of the queer identifications
Julien has arrived at in basing them upon the modern pronunciatiou
of the chinese characters.
101) la Julien's Me'thode not less than 34 characters with a prefixed mouth are quoted,
for which the principal characters could as well have been chosen:
r^j no. 3, no- 3*, no- 344, Pfl D0- 316' BSf no. 371, U> no. 3S3,
H^D no. 385, no. 380, |1^ no. 389, P no. 390, pjg no. 438, pjg no. 564,
pj^lj no. 746, P|| no. 785, jffl no. 809, |tg| no. 813, no. 815, p|| no. 1066,
$3|no. 1022, Pjlgno. 1064, P| no. 1065, no. 1214, p|| no. 1216, p^no. 1216,
Rfno. 1235, Pj^ no. 1308, fl$f no. 1644, B^no. 1746, 2024, P^j-no. 1753, Pj| no.
1897, P$ no. 1972, [j^no. 2164, no. 2221 and Pjjjjno. 2280.
84 0. SCHLEGKL.
73. In no. 2272 of his "MeUode" he says that j|g, which

he reads youe, has been employed by mistake for -fjjfc fa (va) in
Reoata, Iffej&ft? (Fan-yih, II, f. 17 recto); in Vadjra, j$ ||
(Ibid., VIII, f. 11 ero) and in Aoivartti, ffi f j$ |fc (/<?.,
XII, f. 14 vero). Now the old sound of the character was oat
(= tca<), 118 to the present day in Amoy 101), and represented vat,
vad and var in all these transcriptions, which sounded originally
Li-wat-to, Wat-dja and O-wei-wat-ti, initial and final t being repeated
because the preceding vowels were short.
In the transcription of Chakravartti (Pali chakkavatti) |fiF JjflJ
, old sounds Tsia-ka-lo-wat, wat stands for var or vat lM).
In no. 2276 he says: " ^ youe pour gou (sic!) dans Qouddhodana"
^ ip| ^ youe-t eoti-tan (Fan-yih, V, f. 14 ero), adding: "Le
signe correct e"tait "||* cheou". There is an error of transcription,
for the transcriber had only in view the phonetical value of ,
which is pronouuced sut in |j 104). He should have transcribed
Wi 3l 1fi sut-t'ao-t'an, and then sut stands for sud, t'ao for dh6
and Van for dhan, the aspirated syllables dho and dhan being equally
rendered by aspirated chinese characters.
The character ^ in the transcription of dja in Rddjagriha is
simply a clerical error for SB sia (dja). The author of the Fan-yih
Ming-i, VII, f. 11 recto, himself observes that it is to be transcribed
-iff ^ HH R~ra" <lja- There are more such clerical errors iu the
Fan-yih.
The character 0 oat, wat, which is unisonous with oa<, is
used by the Java Chinese to transcribe the syllable wat in the
102) In Annainene it ii pronounced Viet as in Viit-nam jj|Jj jfi , the native name
of Annum.
103) The tkt. letter 3T, which the English transcribe ca, Bounds, according to professor
Kern, more like the dutch mi.
104) Amoy toa/, Canton Mandarin ihtooh, old sound- probably /t\ Properly it
is the- transcription of the Singh. Suith6dana. Kilcl, p. 135a (Pall Suddhodani).-'
THE SECBKT OF THE CHINESE METHOD ETC. 85
malsy words Keliwal (excessive) jjflj ^)| Q ka-li-oat and bebuwat (to
make) )f 0 ba-bu-oat, etc.; and for war in iawar (insipid)
il* 0 <ao-oa<; in tndwar (Arab "Rose") Jjfo Q ma-oat; in keluwar
(Malay "to break out") }jfl J 0 ka-lu-oat.
74. In no. 780 782 Julien says tbat 1$J le stands for la in
Amlaka, for ri in Kharika (^ |J jgffl ) andki/W(|| ^).
Now the old sound of |J was lt (Amoy) or lak (Canton),
and it stood for lak in amlaka, for lik (rik) in kharika and for rw
in rishabha (ris-sha-bha), final of /tA becoming , by assimilation
with initial s of sAa; the consonants are doubled on account of the
preceding short vowels.
Julien is not at all consequent in the application of his principles
of transcription.
In no. 1489 he says correctly that Hj[ lik stands for final l in
phalguna (^ (or jjj|() |Sjf ||? ^ pa< (or p'o) -lik-k'u-na); but in
no. 7834 he says that ^ ^ lik-k'u stands for lgu in phalguna.
/a/t stands for (ZAa in virudhaka \f$fc ^ pi-Hu-lak-
ts'a, also transcribed, more correctly: Wife 1$J j&B pi-lo-lak-ka.
In Pratyeka Jjl stands for pra in Jj| ^ Jjgj -fj{||| pit-lak-
tsi-ti-ka (Fan-yih, II, f. 2 rero), pit-lak - pil-lak = pir-rak = pra,
and <i = tye; the character eAi (old sound H) is redundant in
this transcription. Julien, no. 1842, says it represents tye in this word ;
but what does he do then with Jfjj <t?
jj^f lak stands for rag in Mah&raga ^ ^ mo-hiu-lak
(Julien, no. 362). It stands for lak in lakeha 1$J ^ or ^
lak-ts'a or lo-ts'o (purple). Fan-yih, IX, f. 9 recto.
$ 75. In no. 1918 he says that , which he reads te, stands

for da in Udaka ( ^ j&P ), for (a in Kanaka ( $|| ^= JfjP ) and
for di in Indica. This is totally wrong, for the sound of at the
8G O. 8CHI.EGEL.
time of transcription was not te, but tik (as in Amoy) and tak (aa
in Canton); and so Julien ought to have said ij^f for dak in Udaka,
for tak in Kantaka ,05) and for cftc (di) in Indica. The character
which Julien reads te (no. 1913), does not stands for ta in takshafild
( ^ 5$lJ jjll)' Dut for ia^; does not stand for (a in Pitaba
( ^ ^7D ), but for tak, pi-tak-ka, with the usual doubling of the
k as the vowel in taka is short. The word is also transcribed ^
(Fan-yih, IX, f. 13 rec<o), which now sounds pi-t'eih-kia,
but whose ancient sounds were pi-tak-ka, the character ^ being
pronounced chak in Canton, in which the vowel a reappears.
In no. 1916 Julien says that ^j, which he reads te, stands for
ta in takshaka (|jj[ J&E ) and for <t in kapotikd ( j|P j$0 ) ;
it stands for <a in the first and for tik in the second case. The
character ^[ being pronounced tak in Canton and tik in Amoy.
76. In no. 1627 Julien says that 3|a jjj^ stands for sru in
srughna, and in no. 650 that k'in stands for gh before n in the
same word.
The old sounds of 2p jj^ were Sut-luk-k'in-na, in which
sut-luk stands for surruk = srugh, k'in for ghin and na for na; final
k (gh) of ru<jr/ and initial k (gh) of k'in being repeated, as also
final n of grAin and initial n of no.
77. Let us quote a few more examples already referred to in

1. Under the heading tchi in Julien's Me"thode, p. 201 204, we
find a great many chinese characters now pronounced chi, as re
presentatives of skt. ti, ft, di, tya, dhi, and even ki, as e. g. ^ i
and for di in kundika IpT . These three characters are
all pronounced ti in Amoy and may thus stand for di (Fan-yih,
105) Or rather for tak in Kantaka, the Singh- pronunciation of Kanfaka.

XIX, f. 6 recto. Julien's indication, sub nos. 1850 and 1876: Vol.
XVIII, f. 18, refers to the old edition of 1603. It is also transcribed
tfn ft ) rea^ iQ Amoy-dialect kun-ti-ka.
In no. 1836 Julien gives ^ chi as representing ki in kirdti.
In Araoy this character is pronounced tsi in Bookreading, but ki
in Colloquial, which sound it must have had also at the time of
transcription. He gives ^tg chi as representing ki in kokila ^f|5j ^
(no. 1862). Now the character ^ , pronounced in Araoy-reading
ts't, has as a phonetic , which is pronounced in Amoy-colloquial
ki', and thus the characters Pj , ^ , ^ and ^ are all pronounced
ki in this dialect. The latter character ^ (a harlot) is still read
ki in Mandarin, and thus the old sound ki for ^ is guaranteed by
the Amoy Chinese and the Sanskrit.
Sub no. 1849 Julien quotes iffy chi as representing ti in
Bhartrihari ( tffc B^F TftJ )- '^ne character jffi is pronounced ft
in Amoy, and such was also its ancient sound loe). At the time of
transcription these characters were pronounced fat- (= bhar-) ft

(for tfi) m) -ha-li and represented as well as could be the name
Bhartfihari, which is not recognizable in Julien's reading Fa-tchi-ho-li.
In the next no. 1850 Julien gives JJjj^ chi, as representing dhi
in Samddhi ( * Hi yjjj, ). fjfo is pronounced <t in Amoy-dialect,
and the whole word is read sam-mo-ti, sam ^ being chosen in
order to indicate that mo (md) began with an m.
In no. 1852 Julien says that ffl chi stands for (i in Yashti
( P| ^ ). In Amoy ^SB is pronounced ii. The character
being pronounced iai (ya<) ,0*), ^ ^ iat-sik stands for yash and
it for (i.
106) Psjj? 7ft) $J <' = 'flk + l* Vide ^ apud K'ang-hi.

107) Cp. <i 17, on aspiration as representing r.
108) 7^ j^jj . ^ is to be pronounced / + (*i)a<. faa-yM, VII, f. 30
verto.
88 0. >CH LEG EI..
78. Iu no. 1861 Julien gives 0, chi for (i in Iffim ffl

jffi,). This character is pronounced ti in Amoj.
It is the same case with the characters Jj, and ^ cA't, used
for (At in /fofAina jfe (or ) and kha((ika (nos. 1860, 1864
and 1865). The characters and ^ are pronounced to the present
day t'i (thi) in Amoy.
Sub no. 1872 he gives erroneously 31 instead of ^ as representing
(i in Sakayadrislni (1^ M Wfr & M M 3%. > ân-yih, XV, f. 10
verso). The old sounds were sat-kd-ya-tat-li (= tar-ri) -sit (for <frA)
-ti, the actual pronunciation of the character ^ in Amoy. The
same character is used for ti in -4m-arMi ( |$p| jj^ ^ ), whose
ancient sounds were o-pi pat (for var) -ti, also transcribed ffi
i09) ^ o-teei oa< (= wa< for var) -ti.
In no. 1878 Julien gives cAt as representing fa in Kanaka,

Singh. Kantaka (^ The character |Jf is pronounced i't
in Amoy, and could thus stand as representative of tak; for the
chinese transcriber read the word as Kan-tak-ka (in Amoy Kian-
t'ik-ka) und thus |5/ does not represent (a but tak.
79. Sub no. 13 Julien gives ^r, which he reads 'an, for an
in angdraka ( ^ Rf ^ $JP ). The character ^ , pronounced yan#
in Mandarin, is pronounced iang in Amoy and ong in Canton;
is pronounced go in Mandarin, ngo in Canton and ngo in Amoy.
The Chinese have read ang-ngd-la-ka, which were the old sounds
of these four characters.
80. jjUj) (no. 42 and 43) is used for saii in sandjayin, and
for san in Sannaddha ( ^ |J ). It is pronounced shan in
Mandarin aud Cantonese, but san in Amoy, which has preserved
109) This character, pronounced yueh in Mandarin, is still pronounced oat in Amoy.
THE SECRET OV THB CHINESE METHOD ETC. 89
the old sound; ^ is pronounced nah in Mandarin, but nat in

Canton, lat in Amoy. Julien (no. 1206) erroneously reads the
character Nai, and says it represents na. He seems to have con
founded ^ nah with nai. The character ^ nat represents nad
in San-nad and |J{J t'o represents dha; it represents na< for nar in
^fs ^ i8!E nat-lak-ka for nar-ra-a = naraa (hell). Fan-yih, VII,
f. 1 verso; Eitel, op. cit., p. 81 b; Julien, no. 1207, who, again,
erroneously reads the character as nai.
^ represents nat in [J^ ^ ^ at-ndt-md (for an-ndt-ma =
Skt. and<md (Fan-yih, XI, f. 28 dro; Eitel, op. cit., p.,9 b). It
represents nd in ^ J^Jj ^ j^j? naMt (for nad-di - nadi, the
vowel a being short) -ka-siap-p'o (Nadikdcyapa).
It is the same case as with 3fj|J (no. 46, 47) representing a

and an. The character is pronounced ghan in Mandarin, but san
in Amoy.
81. When Julien (no. 267) gives to jrjj, the mandarin pro
nunciation eul and says that it represents nt, it is because he was
not aware that this character is pronounced dzi in Amoy; the old
sound was m".
^Likewise when he reads (no. 268) ^ as eul, though it stands
for ni in viffini. In Amoy it is pronounced nin (Colloquial It).
The character ft is used for pun in pundarika (no. 290). In
Amoy Colloquial it is pronounced pan, which was the old sound.
It has become hun in Amoy Reading and fen in Mandarin.
82. The character foh is pronounced pok in Amoy Reading,

but pak in Colloquial. It thus stands for Bag in Baglan (Julien,
no. 315 a), for vak in Mdnavaka (^ ^| Ê)> old sounds mo-
nap-pak-ka for md-nap-va-a, with the usual doubling of the
consonants (Julien, no. 313).
90 0. SOHLKGRL.
The character ^| , now pronounced fu, is pronounced in Araoy

Colloquial pu and thus stands correctly for pn in purusha, pushya,
purohita, etc. (Julien, no. 817).
83. When ^ joh is used for na in Tallakshana (Julien, p.

50 and no. 475) we may remark that this character is pronounced
jiok in Amoy-Reading, but that, in Colloquial, the word na" is
used. (Douglas, Diet, of the Amoy-dialect, p. 334 b).
This appears more clearly by the collocation of Jj[ and
J|| ^*.in Sandjna j Jj( ^ an<* Manddjiia J|| ^ ^ and
which value is only to be explained by reading these characters
i-na and si-na; for i-jo and shi-jo would never respond to djiia.
It is the same case in the transcription of djna in Sarvadjna j|
|Sj| s<*t-p'o-sin-jiak, where we must read sin-jiak as sin-na
Bunyiu, Col. 395, no. 34, mentions a foreign Craniana, who

translated one sutra, called ^ |^ j|g[ , which he transcribes Zo-lo-
yen; we have to read the name Na-ra-yen, i.e. Narayana.
The best proof of the identity in sound of ^jj* jiak and va
is found in the two transcriptions of the word Aranya, j|fj ^
at-lan (= ar-ron) -jia (Fan-yih, IV, f. 5 ero and XX, f. 4 rec<o)
and j|U UP at-lan-na (Fan-yih, XI, f. 30 wmo), and a still
stronger proof in the transcription of the name of the river
Hiranyavati j5 5^ ^ jj^ ^ hi-lien-jiak-pat-t'i in the 7n-i (
^TCii Julien, Methode, p. 13), quoted by
Julien, sub no. 886, and by that in the Fan-yih, VII, f. 21 recto
jfjlj Ija 'fjfc J55 hi-lat-na (by assimilation for /an-na = ran-na)
-fat-ti, which more than proves the identity in sound of ^* joh
and Jp wa at the time of the transcription.
Kanyd (a girl) is transcribed in the Fan-yih (Book VII, f. 13
verso) which we must read aJ-na, and not kat-joh, for
THB SECRET OF THE CHINESE METHOD ETC. 91
final t of kat can only change into n, when the next syllable begins
with an n, and so we get regularly kan-na, the a in kan being short,
for kanyd (Pali kannd).
Sometimes ^ is, for convenience sake, used instead of ^ ,
old sound djia (jia), as in Djndna ^ for ^ $5 djia-na
(Fan-yih, IV, f. 9 verso; XIII, f. 12 verso); in Matiodjnaghosha
^ 3^ ^ -HI 0? OTai-?!2 (= wjan-no) -djia-k'u-sa (Fan-yih, II, f.
29 rc<o); in Yadjnadatta ^ ^ ^ ^ yan-djia-tat-to (Fan-yih,
V, f. 23 vero; Julien, no. 2247); in the name of the 6th of the
six brahmanical opponents of Cakyamuni ( ) Nirgranthadjiiati
JL tit |5 3? ni-kian-t'o-djia-fi (Fan-yih, V, f. 26 recto). The
skt. word Pradjna is transcribed ^ ^ poan-joh, but, in this case,
K'ang-hi tells us that poan is to be read pat and that joA has the
sound jia (^Îfc^W^ô^lrl;)- Final 1 otanding
for r we get par, and, by metathesis, pra; pra-djia thus stands
correctly for pradjna.
84. For the same reason the character p ju is used to render the
syllable nya (Julien, nos. 479 480), and $p j=| ju-yen for rendering
the syllable nydn (no. 481). The character $P is pronounced ju (dzu)
in Amoy-Reading; but in Colloquial it is not used, but replaced by
na* (Douglas, op. et loc. cit.), as in the phrase ts'oang nan Hd-ldn po,
buxom as a dutch woman. $P would sound n&n-ian in Amoy,
and only in admitting this, the transcription of skt. nydn (= na-ydn)
is to be explained n0).
The name Punya, generally transcribed ^ ^ p'un-jang (Julien,

nos. 1530 and 470), is transcribed in the Fan-yih (III, f. 16 recto)
110) It is carious that the characters Tfi* and ^JJ are not used in their pronun
ciation joh and j% in transcription of skt. sounds. Cf. Julien, Methode, { XVIII a and
nos. 8S4886, 1530. Julien remarks quite properly that the characters ^Jfc ( anj
represent in Skt. sounds impossible to divine (Methode, p. 52).
92 G. SCHLEGEL.
-^fj $H po-j% and in the Nei-tien-luh ^ ^ put-joh (Julien, nos.

480, 481, 1491, 1530, 2248 and 304). In Japanese these characters
are read Hon-jo, Fu-nio and Suts-jo, but in Amoy for both the
characters ^ joh and $P ju the word ni" is used, and thus
^jftj $H and iffi ^ represent faithfully pu-n&n or put-n&* for
pun-na" = punya.
Tbe character ?ia is used in transcription of skt. sounds

for na and for nya or nya (Julien, nos. 1199 1203). The same
character, now only pronounced na in Mandarin, has in Amoy both
sounds nd" and nidn (Douglas, op. cit., pp. 334 a, 336 a).
85. (âkyamuni's park Lumbini is transcribed jfjj^ ^ ^ 'Mn-

min-ni; it is also called Limbini and transcribed ^ ^ Lin-wi-ni.
Both transcriptions are unexplainable in Mandarin; but Jj^ min
is pronounced 6in and w;", (t, whilst is pronounced Km in
Amoy. This proves that this was the old sound of these characters
at the time of transcription which was Lun-bin-ni and Lim-bi-ni.
In the first transcription the n of lun became m by assimilation
with the initial b of bin (See 43).
Julien's explication in no. 1157: min, pour bi, devant un
n dans Loumbini" is radically wrong, because he has not paid due
regard to the chinese method of doubling the consonant n, the
first t in Lumbini being short.
86. It is only in Amoy and Swatow that the character ^ loh

also has the sound lak, which Wells Williams rightly says was the
old sound. Therefore it is used for lak in Skt. dmalaka ^ ^
tfffl , potalaka |Jfj ^ ^TO , in Amoy-dialect po-t'o-lak-ka (Fan-yih,
VII, f. 18 verso); for ra in Ndraka ^ j&R , Amoy : nan-lak-ka
(Fan-yih, VII, f. 1 verso). It is used for lak in the uigur word for
a horse ulak ^ and for lag in the tibetau ulag.
THE SECHET OK THE CHINESE METHOD ETC. 93
If ^ (no. 2155) is used for dji in ddjivaka, it is because it

was anciently pronounced dji, as in Amoy. The sound tseu or tsz
is of later date.
The characters , yjpj, ^ aud IflJ , all pronounced ye in

Mandarin, are prououuced ya in Amoy, which has preserved the
old sounds. Therefore these characters are used to render the skt.
syllable ya (Julien, nos. 2227, 2229, 2230, 2231, 2234, 2235, 2242).
87. The skt. word svdha (well-said) is transcribed

so-ho in modern pronunciation; but in Amoy-colloquial, the char.
4g} is read soa, as in }Jj ^ hoan-soa, "to roll about the ground
as a horse"; so that Julien's remark in no. 1603 that ^ so
stands for svd is unintelligible. The character is prouounced so
in Araoy-reading; but its older sounds soa and sa have been
preserved in the Amoy-colloquial.
It is the same case as in the transcription of Svara by |gj
soa-lo and of Sodgata by soa-ka-t'o (Julien, no. 1603,
Fan-yih Ming-i, II, f. 19 recto).
When manushya is transcribed ^ T^JJ mat-no (for man-nu)
-sa, or ^ ^ ^J? mo-nq-sa, it is because it does not represent the
Sanskrit word, but the Pali equivalent manussa, the character ^J?
being pronounced to the present day sa in Amoy. The genitive of
the pluralform ma-no-sia-nam (Fan-yih, V, f. 4 recto)
represents the Skt. manushydndm.
The old sound of or |g as a is confirmed by the trans
cription W '(Jlf ^ sa-ka-lo of sdgara (Fan-yih, IV, f. 21 ereo;
Julien, no. 28); by that of vydsa by WJ IfJJ 'J? pi-ya-sa; by that
of i'aJea i$ jj? pu<-a (Eitel, p. 1 64 b) ; by that of the Pali word
Isadhara jj? ^ i-sa-Vo-lo (Eitel, p. 46 a); by that of
Musala Q fj? muk-sa-lo for mus-sa-la (Eitel, p. 78a; cf. 28
94 G. SCHLEQET..
and T'oang-pao, Vol. VIII, pp. 492 -497); by that of ddsa J|fc |
t'o-sa; by that of prabhdsa ^ 0S (Fan-yih, IX, f. 20); by that
of Sari by J|J , the capital of Tabristan in Persia; by that of
the river Selenga by |g '/X Sa-ling-kiang, which latter word
means "river"; by that of hama (a goose) ^ ^ kang-sa (Fan-yih,
VI, f. 18 ow), which transcription proves that, at that time, the
word was still pronounced gaiisa in Skt., which is corroborated by
the german and dutch gans; by that of Fanasa (Jack fruit) ^j? 0ft
p'o-na-sa (Eitel, p. 88 b); by that of Qanakavdsa ||Jj jl^ ^
siang-nak-pak (- vds) -sa (Eitel, p. 121 a); by that of Mahigasaka
M Si ^ l& $R mo-At-s!a-sa-*a (Eitel, p. 69 6); by that of JtsJW
dra gjjT ^ jjj^ mo-ho-sa-lo, the present Masar, about 30 miles W.
of Patna. Eitel, p. 67 a; etc.
88. The word Sainika (a soldier, guard or picket, Julien, no.

1577) is transcribed in the Fan-yih (V, f. 24 verso) g f*J% jj/J),
in Amoy Colloquial sai-ni-ka, rendering exactly the skt. sounds.
In Mandarin the character |J is pronounced si, in Canton,
Amoy, Foochow and Swatow sai, which proves that this was the
old sound of the character.
The word is also transcribed -^q siart-ni and jjj^ san-ni
= Skt. send111), the n being doubled in order to indicate that the
e of send is short; ni stands for na (Julien, no. 1267). De Harlez
has translated in his "Vocabulaire Bouddhique" (T'oung-pao, Vol.
VIII, p. 173) the characters j| ^ by "un here'tiquo, un sectaire".
He must have been misled by the chinese translation 7p| |p[
"he who has the instruction (or possesses the secret precepts) of an
army '- As means also tirthakas in Skt., he has translated
ft\ ^ y "heretics", not aware that tirtha means also "instruction",
111) Sainika is composed of Send -\- iia. Benfey.

THE SECRET OV THE CHINESE METHOD ETC. 95
"a secret precept" (Benfey, Skt. Dict., p. 365 b, nos. 7 and 8) and
was also the designation of certain persons of a king's court (Ibid.,
1. c, no. 12). Sendni means a general in Skt., and it is very probable
that the Chinese san-ni is a contraction of san-na-ni.
The pali word Namo (Skt. namali) is transcribed ^ , which
would be in modern Chinese nan-wu, which does not at all resemble
the indiau word. But if we read it in Canton-dialect, we shall get
Tiam-mo. The so often met with formula ^ |J
Namo Amitdbha (Hail to Amita Buddha) is often explained by
the Chinese in Canton and Auioy that there is no (f^ mo) Buddha
in the South ( |J| nam).
89. I have called the attention of linguists upon the fact

that it is an error to believe that the buddhist monks who taught
the Chinese to transcribe Indian words, spoke classical Sanskrit
(See 1). Probably these monks spoke Sanskrit as badly as a
roman catholic monk or a protestant parson speak Latin. Sanskrit
was at that time a dead lauguage for them, exactly as Latin has
become one for us, and they were too much accustomed to speak
Pali or any other local dialect, in which the Sanskrit words had
undergone very serious changes. Most, or nearly all the medial r's,
the palatal sibilants and lingual f's had disappeared in the 5th
century B.C. "*)
Instead of Dharma they said Dhamma; for Gupta, they said
Ghitta; for Tamralipti, they said Tdmalitti; for OridhrakCiâ, they
said Ghidjaktifo; for Karpdm, they said Kappdso; for Karpura,
they said Kapp&ro; for Agoka, they said Asoka, etc.
This is easily proved by the chinese transcription of these words.
The Sanscrit Dharma is transcribed tat-mo for Dhar-ma;
but when we find ^ J|S t'am-mo-kuk-to, we must read these
112) See Toung-pao. Vol. till,' p. 497408.

96 0. SCHLSflBL.
characters as Dhammagutta, the Pali-form of Dharma gupta in).

Likewise J]) ^ J^J Jfjg tam-mo-lit-ti or J|t ^ ^ to-mo-li-ii
represent Tdmalitti and not Tamralipti; ^ ^ ki-sia-k'ut repre
sents the Pali Ghidjakvfo and not the Skt. Gridhrakvfa, which is
transcribed #m Jj| |S ^ E *1<-'< (= S""1'^) -i'o-'0 (= rfAra)
-&w-<'a.
The characters ^ ftf k'ap-p'o-yuk, 3$ J| k'ap-poe, "j^f J|
ku-poe or J| kat-poe (for kap-poe, by assimilation of /> and )
represent the Pali word kappdso, and not the Skt. karpdsa ; ^|
does not represent -4po/ra, but the Pali pronunciation Asoka "*).
All this is easily enough to understand.
The Hindus came much earlier to Java, than the Chinese to
India, and they introduced there the Pali words kappdso and kappvro,
as the Malays call the cottonplant Kapas and the camphor of Su
matra Kapur-Barus from the locality where it is found 11s).
When we find in the Fan-yih (II, f. 3 recto) the transcription
^ |S3fj VM> old sounds Su-t'o-pan, it does not render the Skt. word
Srdtdpanna, but its Pali equivalent Sotdpanna, in which the medial r
had since long disappeared. The scientific and post-factum made
transcription |S [5rf J| sut-lo (= sro) -t'o (= ta) -6~pat-
nang (= apanna), given by Eitel (o. c. p. 132), is not even mentioned
in the buddhistic dictionary Fan-yih. .
Bhagavan is transcribed ^ 4jf& ^ pok (= bhag) -ka . (= ga)
bam (= van), but its Pali-form Bhagavd is rendered by ^ /fJ/U ^
p'o-ka-p'o = bha-ga-va.
113) For the transmutation of before / in /, see { 27.

114) See above ( 63.
115) The. malay word Kapur is now also used to designate lime (mortar, plaster); bat
mostly in combination with other words, as Kapur tiroh for coarse mortar, Kapur tirth
or Kapur makan for the fine lime mixed in the sirih for chewing, etc. Von de Wall,
Malay Diet, p, 480 a. As the old Javanese did not cement their buildings, which were ad
built of bamboo, wood and palmleaves, they, probably, learnt the mortar from the Hindus,
who colonized Java, and confounded it with -the camphor called Kappuro by them.
THE SECRET OK THK CHINESE METHOD ETC. 97
The transcription j pi-kiu is that of the Pali-form Bhikkhu,

rather than that of the Skt. Bhikshu.
Likewise in ^ ^ ^ j:i{E p'o-t'i-sat-to we have the transcription
of the Pali-form Bodhisatto, and not that of the Skt. Bodhisattva,
so that Julien's statement in no. 2025 that to represents tva in
Bodhisattva is incorrect. nip-poan and ^jjy|| ni-p'an are
not transcriptions of the Skt. nirvana, but those of the Pali Nibbdna :
^ uk-k'i-p'o is not the transcription of the skt. utkshepa,
as Julien (nos. 1318 and 515) fancied; but that of its Pali equivalent
ukkhepo, and so in hundreds of other cases.
Due regard has to be paid to this fact, in order to enable the
student to trace back to their real form the chinese transcriptions
of Indian words.
In publishing this paper, I have only had the intention to
fulfill Julien's wish, expressed in page 35 of his Mdthode, "that
"other philologues, availing themselves of the entirely new materials
"for study which I offer to them, may continue the laborious
"enterprize which has occupied me for more than fifteen years, and
"may succeed in still more extending and enlarging the road which
"I have been the first to open".
Errata.
P. 38, line 8, for lu-sz read tu-se.

42, 19, where in read wherein.
42, 19, represent read represented.
52, 7, /ao read /o.
CONTENTS.
PSf
Introduction m
1. Insufficiency of the Methode of St. Julien 1
Examples for demonstration 2
2. Confusion occasioned by Julien's ignorance that chinese words now
ending in n had formerly a final m 4
Repetition of consonants 5
On the rendering of skt. [a and tha in chinese transcription ... 5
3. On the value of final t as r in chinese transcription 6
4. On pretended final n standing for r 7
5. Analysis of a Malay-Chinese Manual 7
Diacritical signs given by the author to indicate differences in sound 1 1
6. The seven tones in the Amoy-dialect 12
Method of indicating short tones of characters which have none . .13
7. Reason why final n appears to stand for final (, I or r 14
Exception to the rule 15
Confusion of n and r in the pronunciation by the Malays and Siamese 15
Final ri and bill rendered by ;/i 16
Kappal and Kappan 10
8. Examples from which appeal's that final n does not stand for final r 17
The double sounds tan and tat of the same chinese characters. . .18
Assimilation of consonants 20
9. On the value of sound of fffc and ^ 20
10. Hirth's Parthum explained 21
11. Old sound of J^j proved by examples ... 22
fj| for har 23
12. Phonetical value of |$flj" 23
Phonetical value of the characters PJ , jffe , j| and ^ 24
|$J and 41f 1'^present a double initial H 25
Con-non and Condore 25
Value of the characters ^ R$<J 20
13. On pretended m for r in 26
Suppression of r in sanskrit transcriptions, proved by modern trans
criptions of malay words . 27
100 CONTENTS.
Page
14. Transcription of TamnlipH and Gupta 28
15. Metathesis of kir for kri, in Crigata 29
Change of p before m to m 29
Transcription of Kshetra and Manlri 30
10. On the rendering of ksha, kshe, kshau and kshun 30
Phonetical value of |jfc 30
How the Chinese divided the sanskrit words in their transcriptions . 31
Impossibility for a Chinaman to pronounce and transcribe some Sanskrit
syllables 32
17. Aspiration for the letter H.
Examples for elucidation: Prabhdratna, Prabu, Pradjdpdti, Prasenadjil,
Bhadrika, (judra, Dhrita, Dhritaka, Turfttii 33
Gridhrakuta, Narcndra, Rudra, Bhadrapdla, Bhadrakalpa, Krofa
and Grdsa 34
18. Other representatives of B.
Phonetica| value of ^ in Ndgardjuna 34
19. Phonetical value of |j , jg , , || , g and flj 35
Substitution of r by g or gh 30
20. Assimilation.
Phonetica| value of |j| , and ^ 36
21. Assimilation in transcription of Malay words 37
In Formosan and Turkish words 38
22. Error of .lulien in assuming that the Chinese transcribed sanskrit words
etymologically instead of phonetically, proved by modern transcriptions 39
Transcriptions of Igvara, Paramefvara, Candifchara, Sthdnifvara and
Suvarnaprabhasauttama 40
Transcriptions of Irshya, Afvakarna, Afvayusha, Acvadjit and Afva-
yudja 41
23. Transcription of ISdslttra, Sarjmushadhi and Upanishad 42
24. Transcription of Sldpa, Kapildwtslu, llohitavasiu and Anushtubh. . 42
Transcription of Miskin, Meestcr Cornelia, (Jvclapura and lihdskara . 43
25. Transmutation of Consonants.
Transcription of Vah/u 43
Change of K before L : Hebraia 44
26. Change of A' before P in Bhadrakalpa 44
27. Change of K before T in Uttara 44
28. Change of A' before S: Hosna 44
2Ji. Change of N before M, B or P 44
30. Change of N before L 45
31. Change of P after M 45
CONTENTS. 101
Page
32. Change of Ng before K or G, proved by the transcription of the words
Bakaldhi, Bagelen, Bckasi and Nagarahdra 45
33. Change of Ng before L, proved by the transcription of Kikana and
Dmma 46
34. Change of Ng before B or M, proved by the transcription of Kumbhdmla,
Kumbhiru and Kducdmbi 46
By that of Bula Comba, Pemambuan and Bombay 47
35. Change of Ng before N: Canaka and Kuraha 47
36. Change of Ng before P: Menumpang 47
37. Change of Ng before T or D: Pondichcrry 47
38. Change of P before T: Anavatapta, Anavadatta 47
39. Change of P before Af 47
Atmanc and Gdulami 48
40. Change of T before L or ii : Vallabhi, Parama, Mrigala, Mlcchchha,
Kotlan, Ullangha, Talkan, Kurana 48
Talus, Barukacheva, Kcdu, Magclan, Belt, Blitar, Blora, Saint, Telok
belong, Serel, Sudar?ana 49
41. jf'used for K: Yaksha, Akshamati, Aksliubhya, Serawak, Gdgak, Lawek 50
42. Change of T before S: Asokadutla, Afdka, Dushana, Dushkritu, Tak-
shapild, Bussorah, Cacdhkha 50
43. Change of T before N 50
Lumbiiii, Limbini, Larnbini, Lavini, Buddhavana, Ana Ratna, Kanaka,
Pannaga, Ghanam, Vydkaranam, Kurana, Penna 51
44. Pin ffi for vil 52
45. T used for S: Keras 52
4(i. Modern and ancient alphabetical spelling 52
Svastika and Svdguta 53
47. Japanese transcriptions of Blancardi, Karslen 53
Japanese transcription of Ikura ka, Kusaru, Kuraishi, Module ho,
Yaxc ta, Makura 54
48. Curious instance of assimilation in the transcription of Tjiringin, Dingin
and Chtribon 54
49. Gemination of Consonants 55
50. Illustrations of this feature: Kapolana, Kapphina, Champaka, Padma,
Pudgala, Kukkuta 56
51. Mistakes made by Julien by not applying the law of gemination of
consonants 57
52. Gemination in the transcription of Bagheldn, Bhadrika, Vikramdditya,
Ddldditya, Ciladitya 58
In Batnagarbha 59
54. In Kanakamuni and Amala 59
55. Gemination of consonants in the transcription of japanese words . . 59
1 02 CONTENTS.
56. Transcription of iiagetan, Kcilah, Tekiikan and Tamaha .... 60

Transcription of other raalay words and place-names. Transcription of
Venice 61
57. Gemination of consonants not iiFed when the preceding vowel is long 62
58. Exceptions to this rule: Djeta and Djetta, Bddhi and Budhila . . . 63
Brdhmana, Antra, Shanmuka, Dushasana, Hiranya, Kashdya, Ka-
canna, Kanadeva, Kapita, Kapilavaslu, Kapitha, Kapica .... 04
Kaputana, Mdksha, Mokshacila, Mokshala 65
59. Julien's verdict in Methode, p. 50, XVI, unwarranted
Vallabhi, Kukkuta, Kukkura, Pippala 05
Uttara, Sumrnaprabhdsottama, Buddhdshnisha, Devendra, Muhendra,
Lankdvatdra, Srdtapanna 66
Djayendra, Kukkuti'roara 07
00. Contractions in Sanskrit 67
Error of Julien in his accusation that the Chinese dit not write twice
the same skt. letter in their transcriptions, demonstrated by examples 07
01 . Gemination of Vowels 08
tj 02. Suppression, contraction, omission or metathesis of syllables
Muehilinda, C.ubha stri, Andgdttiin 09
Ananda, Aniruddha, Antara, Apaamum ka, Bhadra, Bhadrika, {Irigata 70
03. Akshdbhya 70
Acoka, Kappaso and Karpasa 71
Yuktavddi, Yuthi, Uttara 71
Prasenadjit 72
S 04. (Jrdmanrra, Shami, (^ramanerika, Gedjakah) 72
05. Sakaya drishti, Drishtanta, Alma, Ldkccvara 73
ft _
Icvara, Sthane'Qvara, Uruvilva, Mditrcya 74
00. The curtailed transcription of Vishtambha and Skandha 74
Impossibility for english and french people to pronounce the syllable Sclile 74
07. Suppression of finals and letters.
Chhdyd, Adbhuta Dharma 75
Ndrayana, Anagdmin, Varman, Bruhmana, Ferghana, Mahakalya-
yana, Nirvana, Sdngharama, Chandana, C.hinasthdna, Buddho-
chinga. Bokhara, Vicvakarman, Mahandman, Cdkyamuni . . . 70
08. Suppression of final n in Kdlktdyin, Kollan 70
In (îkhin, Varman, Shadumdn, txnnangat 77
Suppression of final syllables in Hrosminkan, Kapdlikas, Kvmbhdndas,
Mallika, Aranyakah, Arada kdldma, Brahmadatta 77
09. Suppression of final r, l, k and medial y in Malay words ... .77
70. Final Sibilants.
Characters used to express them 78
CONTENTS. 103
71. On the characters p| and ^ prefixed before Chinese characters . . 82

72. Conclusion 83
73. On the va|ue of 6ound of , ||9 and Q 84
74. On that of 85
75. On that of , ^ , ^ and $g 85
70. Transcription of Srughna 80
77. Transcription of <t, ti, di, hja, dhi and hi; Kundika 80
of Kirati, Bhartrihari, Samadhi, Yashti 87
78. of luini, Kathina, Khattika, Snkdyadrishli, Avivartti, Kantaka . . 88
79. of Angdraka 88
80. Transcription of Sandjayin, Sannaddha 88
of Naraka, Analmd. Value Jjjjlj *n, ^jj nat and J||}J san . . 89
81. Value of , jjjf, # 89
82. Value of jfsjj|: Mdnavnkn 80
Value of 1j 90
83. Value of Jfc and J$ 90
84. Value of #p , ^ and ffi 91
85. Transcription of Lumbini 92
80. Value of ^ : P&tataka, Amalaka, Ndraka, Ulak, Ula;/ .... 92
Value of ^f. $ and ^ 93
87. Value of and J^J? in Svdha, Svara, Manushya, SSgara, Vydsa,
Vatsa, hadhara, Mnsala 93
in Prabhdsa, Sari, Setenga, Haiisa, Panam, (ânakavdsa, Maltifasakd,
Mahdsdra 94
88. Transcription of Sninika and Send. Error of De Harlez 94
Transcription of Namo 95
89. Proofs that the buddhist monks, wo taught the Chinese to transcribe
Indian words, did not speak classical Sanskrit but some local dialect'. 95
Transcription of the pali words Kappdso, Kappuro; the malay words
Kapas and Kapur explained 96
Transcription of Sotdpanna and Srdtnpnnna: of Bhagavan and Bhagavd 96
Transcription of Bhikkhtt, Bddhisatto, Nibbdna, Ukl;he}io .... 97
Importance of the Pali and other dialects for the identification of indian
words in chinese transcription 97
Julien's wish fulfilled 97

The Secret of The Chinese Method of Tran

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Secret of The Chinese Method of Tran

Uploaded by

Copyright:

Available Formats

XGxd+S^a tic'** r / ^/'^

THE SECRET OF THE CHINESE METHOD

OF TRANSCRIBING FOREIGN SOUNDS

Reprinted from the T'oung-Pao, Scries 11, Vol. I.

As we have said in our Conclusion 72, the present paper is

He refers Gavampati (uo. 277) to Chap. XI, fol. 15, whilst it

Iu no. 2208 he gives jjja wei for vi in Vichtioii and refers to

In no. 2240 Julien says that the character ye ^ has been

In no. 2274 he says that [H| is erroneously employed for dja

Besides Julien often only gives the value of one syllable of a

Only 011 p. 52, XVIII a, he gives as transcription Ta-lo-la-

old edition of 1603, with a slight difference in the number of the

Leyden, June, 1900. G. SCHLKGFX.

OF TRANSCRIBING FOREIGN SOUNDS

1. I have already, on different occasious, called the attention

1) Toung-pao V, p. 173; VII, p. 184 icq, 568 scq ; VIII, p. 490.

arise from his grave could converse, without much difficulty,

3) = = JJjjig "an owl", pronounced tsi in Amoy, would seem to make an

pou dans Pouroueha; <|j Fou-leou pour pour dans pourqa".

pronunciation of ^ (to float) is p'u and was probably also

5) A Chinese Malay Manual, Chapt. ^ >|j (Numerals). I shall preaently return to

In XIII A, Julien says that final n is very often left out

transcription, and were thus very appropriate to transcribe the

have given abundant examples of this feature in the transliteration

2|5 , old sound sut, for sur in Suri ^ %\ .

k'at kar in Kama |

k'at kar in Kavpura | ^ .

These examples could be multiplied by thousands. The isolated

6) Journal China Branch Royal Asiatic Society, Vol. XXI, 1886.

4. But there always remained to solve the riddle why, in

As is seen, the chinese word ^ , meaning I,

The second partition has to the right ^ you,

The third partition has to the right ffi he.

The two volumes contain XVIII Chapters treating respectively of:

Chapt. I. P Jp| i jf^f Parts of speech.

This preface is dated from the year Kia-shin of Kwang-su

This seems to be the same man, mentioned as Tan T'ien-pit

He then explains the method of his Manual as we have indicated

adding to some of them conventional marks as J^, O Q-

Experience has taught me that this O *s placed next to a

attention of the scholar is drawn. He indicates these tones in the

But if we look up for this syllable in the -j- 3l Dictionary,

and f% and for the ^ A w*. fe> UK Mld $' So

smaller type: ]L ^ 50? *s to 06 rea^ A(an + p)& = hit.

7. We shall now give some further illustrations of the reason

Final n before r represents also an r; e. g. in 4|

has no y\. characters, the author used ban, which is to be

8. If we now turn to other chinese words wherein final ?i

10) In the name of the celebrated concubine Tat-ki T"** {TarkiV).

wherein t'at-lo again represents tra. Djina puttra is transcribed

In Buddha ksh&tra the last word is transcribed ^ ^fi j|fl

11) Eitel's //ty in both transcriptions is of course a misprint.

as in ^ ||i ttU (ft Sam-mo-t'at-t'a for Samotata, name of an

fo-yu-shi (ancient sound Atsippakjuja). For the character ^ is not

When (as in N. 1707), j|[ j|0 represents tram in SUtrarn,

9. The same feature is observed with regard to the syllable

as in the Sanskrit word pradjfia, which means in Chinese "wisdom";

10. Dr. Hirth's identification of ^ with Parthuva ) is

Po-wu ^f) was a place in Chao (j|j|), Vide 1 ^

This Po-wu was the name of the present town of P'ing-shan,

la-hou-la and ^ tifa Jjp la-hou-la represent the second form of

12. Many quibbles could have been avoided respecting the

80) Jalien, p. 134, N. 744 ^jj ^ for rtha.