Duan Qing （The Center of Eastern Literature Studies, Peking University）
Peng Jinzhang（Dunhuan Research Academy）

In the following we are going to show a fragment in Brqhm], which was unearthed from Dunhuang. Before we start with our main topic, we would like to say a few words on the archeological background of the new finds.
Dunhuang is a well-known name to everyone here. A hundred years ago, the discovery of a well-hidden library with thousands and thousands of manuscripts in very different languages and scripts made the small oasis city in Northwest China famous in the whole world. Today, due to numerous Mogau-grottoes at Misha mountain, which lies about 25 kilometers South-east of the city, Dunhuang still attracts people from different countries. The Mogau-grottoes of Dunhuang are divided into 2 sections: the southern part and the northern part. The southern part with 487 caves, stretching over 1000 meters, provides visitors and scholars alike with a spectacular view of brilliant wall-pictures and reliefs, both well preserved. In contrast, the northern part with 248 surviving caves has been neglected for a very long time. The situation began to change, when the Dunhuang archeologists started a project in 1988, which focused on a systematic archeological investigation of the northern grottoes. Under the leadership of Professor Peng Jinzhang, dean of the archeological institute of Dunhuang Academy, the archeological operation has been going on for 7 years. When the project came to the end in 1995, archeologists could be satisfied with the results, because the northern caves yielded plenty of manuscripts and relics. As one might expect, the unearthed manuscripts are written in different languages, in Tibetan, Mongolian, Uigur, Xixia (Tangut), and there is even one well-preserved manuscript in Syriac from the book Before and After.[2] Numerous manuscripts contain well-known Buddhist texts, such as Suvarzaprabhqsa-s[tra, Saddharmapuzfar]ka-s[tra, to name only a few examples. Dunhuang archeologists are planning to publish altogether 3 volumes reporting on their unearthed achievements. The 1st volume has appeared in print in 2000 under the title Northern Grottoes of Mogauku. Two other volumes are now in press.
The Sanskrit fragment we are presenting here belong to the recent finds from the Northern Grottoes of Dunhuang. Altogether two large and several small fragments in Brqhm] on paper came from two different caves. Fragments with the signature 464:18 and 464:75 came from the cave 464 which had also been visited by Pelliot in 1908[3]. The fragment with signature B 142:1 has been found in cave B 142, and B of the signature means a cave existing on northern side of the Mogaoku.
Generally speaking, those fragments from Dunhuang show a great similarity to the large portion of the well-known Turfan collection in Berlin. They are written in Brqhm] script, in the called "Nordturkistanischer Brqhm] Typ b". Judging from the contents, the fragments apparently belong to the Abhidharma class. It seems to us that the remaining fragments represent different chapters of a lost work of the Abhidharma class. We could not identify a parallel version so far, neither in Chinese nor Tibetan translation.
Of the few Brqhm] fragments, the one with signature 464:18 is the best preserved and of great interest. Three Buddhist calendars can be distinguished based on this small fragment. Our presentation will deal with this fragment alone.
About Fragment 464:18
The signature 464:18 means that the fragment was found in cave no. 464. The size measures 17 cm by 10.7 cm. The writing material is paper, which is broken on both sides, but the upper and bottom edges are well preserved, so we know that the original folio contained seven lines of writings. There is a hole in the middle. After a careful reading, we are convinced that the hole in the middle is not the string-hole as usually expected on such a folio, but due to damage. Through the content, recto and verso of this fragment can be determined.
By a lucky coincidence, an interesting part of the text survives. In the second line of the verso, the word po2atha appears twice. The spelling of the word betrays the school which the author of this manuscript might be affiliated to. As demonstrated by Haiyan Hu-von Hinüber, the form po2atha was preferred by Sarvqstivqdins while the M[lasarvqstivqdins used to write po2adha.[4] At one place of the preserved text, we can observe that the author had quoted from Guzaprabha's Vinaya-s[tra. Since we know that the Vinaya-s[tra is written by Guzaprabha, and that he was a famous Qcqrya of the M[lasarvqstivqda-school, it is really interesting to take note of a substitution of words[5]: While we read in the original Vinaya-s[tra of Guzaprabha the word form po2adha, we find in the quotation, as preserved in this fragment, the form po2atha. This is good evidence that different Buddhist schools were aware of distinctions of words, and they consciously selected their own forms of words.
At first glance, because po2atha and different timetables are discussed in the fragment, it suggest that the text might be the remains of a manuscript of the Vinaya class. But we read in the last two lines of the verso "yug]nq iti . kali[y] " whereby "yug]nq " may falsely stand for the right form of yugqni, for kaliyuga appears to be the topic next discussed. Though the text at this place is broken, it is not to oversee that the author had changed his topic. Taking the changing of topics as an indication, we understand that the manuscript is arguing with topics, which concern Buddhist conceptions of time. Since conceptions of time are topics of Abhidharma, the fragment has an Abhidharma colouring.
It is amazing that in the small fragment two persons' names are mentioned. One is the famous Vasubandhu and the other is, as mentioned above, Guzaprabha. Two quotations can be ascertained respectively, one is from the Abhidharmako1a, the other is from the Vinaya-s[tra. Through the mentioning of Guzaprabha, the question when the text was composed and written down can be solved in broad terms. Chinese pilgrim monks Xuan Zang and Yi Jing from the Tang dynasty, both wrote about him[6]. Thus we know with certainty that Guzaprabha must have lived in the 6th century. The fact, that no parallel version of the text as preserved in the fragment could be identified either in Chinese or in Tibetan translation, affords another proof that the manuscript must have materialized at a fairly late date, not earlier than the 8th century.
This fragment in Brqhm] belongs to the rare finds of old manuscripts for which the archeological background is clearly recorded. Cave no. 464 also yielded manuscripts in other languages, as for example Uigur, Xixia and Mongolian, and thus provides with evidence that the cave might be lively occupied-roughly estimated-until the fall of Mongolian the empire in the first half of the14th century. Of course, we cannot read from a fragment a whole monastery at Dunhuang at such a later epoch, but we may quote here an appropriate Chinese idiom: With one leaf the autumn is realized. Through the discovery of several fragments in Dunhuang, the fact cannot be neglected that the Sarvqstivqdins, going out of Turfan oasis, survived until a late date even in Dunhuang.
Transliteration
B464:18-recto
///.y. de1ai.. nyqyena sa/vatstsare[7] caitro mqsa4 prathamo bhavati ///
/// 9pha)lguno mqsa1-caitra1-cqya/ vasanto 3tu4 tatra phalgune mqse k32za(pak2e [narqtri0 ///
/// 91rq0vaza1-ca var2q 3tus-tatrq2qfhe k32zapak2e [narqtri1 1rqva9za0 ///
///9n0(q) ma[8] mahadutstsava4 .rq+ pak2advaye 'pi tasmqd-uktam-i ///
/// 9mqr0ga1iro navama. (y)[+ 9po2ya mq0(gh)au 1i1iro 3tus-tatra po2ye ///
/// 9sa/0(va)tstsara4 prqrabhyate ·ta/n9a0y(e)na caitra_vai1qkhau vasanta 3tu ///
/// trezaivam-ukta/ t3t]yasaptamayor-ny[nam-iti· catvqro (mq).. ///

B464:18-verso
/// 2wa/ pq`cada1ikam-iti· bhadantaguzaprabhena tu vina(y)9a0 ///
/// 9po0(2a)thq bhavanti· tatra traya 9p0o2athq4 pa`cada1ikq bhavanti.....
///.(c)qturda1ika4 + .i + + mqsqva1i2wo 3tur-bhavati ...
///9qcq0 r(y)a[9]vasubandhunq tu ko(1a)91q0stra ukta/·hemantagr]2mavar2q(zq)9m0 ///
/// (nipqt0yate [narqtr] ：adhyardhe mqse nirgate eka/ 1e2e 'rdhamqse dvit)......
/// (bh). saivqtra tu wir dra2wavya4 || yatheda/ yug]nq iti. kali(y) ///
/// ava1i2yante da1a sahqsrqz]ti anqdyqn]ti ：prathama/ ///
Translation
B464:18-recto
///... countries... accordingly, of a year, Caitra is the first month.///
/// Phalguna month and Caitra month are the spring season. There in the dark half of Phalguna month, (the minus-night ...)
///... and !rqvaza month are rains season. There is in the dark half of Q2qfha month the minus-night. !rqvaza (month) ...
/// A great festival named ... also in the two-half-month. Therefore it is said...
/// Mqrga1ira (month) is the ninth ... Po2ya and Mqgha months are the pre-spring season. There is in Po2ya month...
...the year begins. According to this, Caitra and Vai1qkha months are the spring season. ///
... so it is said: "the third and seventh have the minus." Four months ...
B464:18-verso
...fifteenth." As (related ) by Referend Guzaprabha in Vinaya 9s[tra0...
... are Po2athas. There are three Po2athas on fifteenth. ///
...on fourteenth ... ... is the season having the rest of the month. ///
As Qcqrya Vasubandhu had said in the Ko1a_1qstra: "Winter, summer and rains..."
... the [narqtr] happens. After one and a half months have passed, one in the rest of the half month, second....
... Here[10] "wir" shall be seen. As this "yug]nq", kali-yuga...
... are left. "Ten thousand", "without beginning", the first...
Notes and Commentary
Generally speaking, we are confronted with a text, which deals with different systems of the lunar calendar in which the months are subdivided into dark and bright fortnights consisting of fifteen days or fourteen days. It is indeed not possible to reconstruct whole systems based on a single fragment, but through some key words preserved in the text and work done by scholars like Claus Vogel, problems arising within the text could be resolved to a great extent, and three calendrical systems can be distinguished in the preserved short text.
caitro mqsa4 prathamo bhavati
"Caitra is the first month". Apparently we have a calendar of the year, which starts with the month Caitra. Thus it differs from the system of M[lasarvqstivqda affiliation, according to which the year starts with Mqrga1]r2a[11].
9pha)lguno mqsa1-caitra1-cqya/ vasanto 3tu4
"The Phalguna month and the Caitra month are the spring season." With this statement, we have a year divided into six seasons consisting of two months each. However, we meet here an interesting slight deviation from the normal Indian calendar: The spring starts with the month Phalguna. It does not coincide with the well-known season subdivision, as preserved in Xuan Zang's description. According to him, the spring begins with Caitra[12]. Herewith a different system of season subdivision has emerged and will be clearly seen in the next context.
tatra phalgune mqse k32za(pak2e [narqtri0 ///
"There in the dark half of the Phalguna month, [narqtri..." The manuscript is obviously explaining a calendar according to which a year is divided into 12 lunar months, each month into two fortnights - the white fortnight and dark fortnight, and each fortnight is reckoned at fifteen days or, with a few exceptions of dark halves, at fourteen days. Claus Vogel designated such a system as a lunisolar calendar[13]. His research showed that the system had developed fully by A.D. 380/381 and was later adopted by Buddhists, as shown evidently from literature of the M[lasarvqstivqda and Theravqda schools[14]. Though differently arranged, the words which appear here and in the next 4 lines confirm that the author of this manuscript was familiar with the lunisolar calendar. The dark halves limited to fourteen days possibly to be identified in the fragment are: Phalguna, Q2qfha, Po2ya.
91rq0vaza1-ca var2q 3tus-tatrq2qfhe k32zapak2e [narqtri1 1rqva9za0
Even though the line is broken, because of the similar sentence structure in line 2, the missing text before var2q 3tus could be restored. We expect here two months, and they must be Q2qdha and !rqvaza. Unusual here is the month of Q2qdha in combination with !rqvaza reckoned as the rains season. According to the Indian calendar, we encounter Q2qdha normally in the summer season. The shift could happen only in the situation in which the month of Phalguna is reckoned as part of spring, and there are still three months countable between Phalguna and Q2qdha, namely Caitra, Vai1qkha and Jye2wha, as Xuan Zang listed. We see also in line 5 that Po2ya has been shifted into pre-spring, although it normally belongs to the winter months. It is amazing that we are encountering here a calendar only recorded in the Mahqvyutpatti, a lexicon compiled in Tibet in the first half of the 9th century. Claus Vogel was aware of this strange calendar, and he is of the opinion that this calendar is specially designed for Lamaist canons and "erweist sich mithin als eine ganz seltsame Kreuzung zwischen dem indischen und tibetischen Jahreszeitensystem".[15] Here the strange calendar otherwise only known in Tibetan sources has appeared in a Sanskrit edition.
mahadutstsava4
Line 4 is mysterious. The first word seems to be "nqma", and there is only a faint trace of long q to be read. What kind of a festival the mentioned great festival does mean is not clear so far.
9mqr0ga1iro navama. (y)[+
This is an unsolved problem. mqrga1ira is the nineth month, if a year opens with the month of Caitra. But the meaning of the ensuing broken word is not clear.
With line 6 we are introduced to a new calendar, according to which a year apparently opens with the month of Caitra, and Caitra andVai1qkha are reckoned as two months of spring again. This is the familiar Indian calendar with 6 seasons to a year.
t3t]yasaptamayor-ny[nam
"The third and seventh have the minus." With this statement we are introduced to another system of dividing the seasons, whereby a year is divided into three seasons -winter, summer and rains, and each season has four months or eight fortnights. This is a specific calendar of Buddhist tradition as stated by Abhidharmako1a[16] and by Xuan Zang[17]. In the whole year there are altogether six fortnights counting fourteen days. Reckoned at eight fortnights of a season, the third and seventh fortnights must be reduced, because two lunar days end in one solar day"[18]. Buddhists had to be familiar with the calendar, because the important ceremony of Po2atha happens fortnightly. Each school might have a different subdivision of the seasons, but the dark halves ending on the fourteenth remained the same. Such a calendar, according to which a year is divided into three seasons and so on, is in existence in the Theravqda school, as revealed by Claus Vogel. In each season, there were altogether eight uposatha days, and two of them- the third and seventh - fell always on the fourteenth. Herewith we may quote Buddhaghosa's Kaxkhqvitaraz] and the translation from Claus Vogel:
tattha hemantagimhavassqnqna/ tizzam ut[nam tatiyasattamapakkhesu dve dve katvq cha cqtuddasikq, avasesq awwhqrasa pazzarasikq ti, eva/ ekasa/vacchare catuv]sati uposathq.
"Of these (Uposatha days) six fall on the fourteenth, two each - in the third and seventh fortnights - having been laid down for the three seasons winter, summer, and rains the other eighteen fall on the fifteenth. Thus (there are) twenty-four Uposatha days in one year."[19]
What the sentence t3t]yasaptamayor-ny[nam means, has been well explained by the quoted text. But we have in "ny[nam" a neuter form. The synonym for the omitted day must be ahan instead of tithi.
vina(y)9a0
The work mentioned here may be vinaya_s[tra or vinaya_s[tra_v3tty_abhidhqna_ svavyqkhyqna, because both are works of Guzaprabha. It is apparent that the author was quoting Guzaprabha from either of his above works, and the passage concerned may supposedly be the following from the Tibetan version according to the Peking edition.
gso sbyox bya'o || zla ba phyed phyed kyi tshes bco lxa la'o || …… gax ?ig tshe bcu b?i dax bco lxa pa `id la 'di mtshan mo gcig gis ma chog pas der de rix `id du tshes bco lxa ?es bya ba bya ba ste| gso sbyox byed de bas na bcu b?i pa 'dis bsdus pa yin no || gax ?ig mdo las gso sbyox ni g`is te | bcu b?i pa dax | bco lxa pa'o ?es gsuxs pa yin no || [20]
[qcq]rya(?)-vasubandhunq tu ko1a_(1q)stra ukta/. hemanta_gr]2ma_var2q......
......[nipqt]yate [narqtr]
Here is the quotation of Vasubandhu from the Abhidharmako1a concerning a subdivision of a year into three seasons. The sentence can be filled out with help of the extant Sanskrit edition of the same work, as follows:
qcqrya-vasubandhunq tu ko1a_(1q)stra ukta/. hemanta_gr]2ma_var2qzqm adhyardhe mqsi nirgate 1e2e 'rdhamqse vidvadbhir nipqtyate [narqtr]
The Chinese translation of the quoted text:
寒熱雨際中，一月半已度。於所餘半月，智者知夜減。[21]
However, there is a difference notable. It is in the second pqda:
The restored text： 1e2e 'rdhamqse vidvadbhir nipqt]yate [narqtr]
The existent edition: 1e2e 'rdhamqse vidvadbhir [narqtro nipqtyate[22]
於所餘半月，智者知夜減。
Why has the difference come into being, we would let the question open.
wir
The text is too broken in order to judge if it were a mistake of the writer. We would prefer not to read a syllable as a mistake if it can make a meaning. wi is a grammatical sa/j`q and means the last syllable of a word.
yug]nq iti. kali(y)
"yug]nq"must be a mistake. The right form shall be yugqni, it is the plural form of yuga, as we can read in the next word kaliyuga. It seems that the author is changing the topic here, and that he starts to explain what is yuga and kaliyuga , -these are conceptions of time according to Buddhists. As stated above, the changing of topic allow us to suggest that this fragment belongs to a text of Abhidharma class.
To sum up, fragments in Sanskrit found in northern caves of Dunhuang are not numerous. However the few finds can speak for some facts: There were influences of Sarvqstivqda in Dunhuang which came from Turfan oasis. As proved by other finds from the northern caves, the fragments must belong to a work of later time. As demonstrated by the fragment with signature 464:18, the tradition of Sarvqstivqda school writing on Abhidharma seems to be well maintained even in a later time, and the fragment itself belongs to a lost manuscript. The question is: Who might write down a work which demands such rich knowledge. Perhaps the question will be open for a very long time.

[1] This paper has been firstly presented on the XIIIth Conference of the International Association of Buddhist Studies. Herewith we would like to express our sincere thanks to Paul Harrison and Oska von Hinüber who kindly spared time checking and polishing our writings, even though they were very busy during the conference.
[2] Duan Qing: "Bericht über ein neuentdecktes syrisches Dokument aus Dunhuang/China", in Oriens Christianus, Band 85, 2001, pp. 84-93.
[3] Cave no. 464 had also been visited and examined by Paul Pelliot in 1908. In his record, the cave is listed as no. 181. At that time, he could return home with above 900 woodblocks for printing in Uigur script and hundreds of manuscripts in Uigur, Chinesese, Tibetan and Mongolian alone found in this cave. The finds are preserved in the Bibliothèque Nationale in Paris.
[4] Haiyan Hu-von Hinüber, Das Po2adhavastu, Vorschriften für die buddhistische Beichtfeier im Vinaya der M[lasarvqstivqdins，Studien zur Indologie und Iranistik, Monographie 13, Reinbek 1994,p.1.
[5] Guzaprabha, Vinaya-S[tra, edited by Prof. P. V. Bapat & V. V. Gpkhale, Kashi Prasad Jayaswal Research Institute, Patna.
[6] 季羨林等校注《大唐西域記校注》，中華書局1995年版，第398頁。王邦維校注《南海寄歸內法傳校注》，中華書局1995年版，第204頁。
[7] "sa/vatsara"is the correct form. "ts" is doubled due to an influence of the Tocharian language, as orally pointed out by Lore Sander.
[8] Left above "ma" there is a small point. After comparing with the anusvqra in the above two lines as for exemple in the word "sa/vatsara", we are quite sure that the point is not an anusvqra, but a spot. In line 2 under "mqse", there is also such a spot.
[9] Only "ra" is readable.
[10] "(bh). saiva" belongs to the sentence before and has no clear meaning for me. My translation starts with "atra".
[11] As shown by Claus Vogel: "On the Date of the Po2adha Ceremoney", in: Baudha- vidyqsudhqkara4, Studies in Honour of Heinz Bechert on the Occasion of His 65th Birthday，edited by Petra Kieffer-Pülz and Jens-Uwe Hartmann, Swisttal-Odendorf, 1997, pp. 673-687.
[12] 季羨林等校注《大唐西域記校注》，中華書局1995年版，第168頁。
[13] Cf. Vogel, op. cit., p.680.
[14] Cf. Vogel, op. cit., p.680.
[15] Claus Vogel, "Die Jahreszeiten im Spiegel der Altindischen Literatur", in: Indica Et Tibetica, Band 37, Vividharatnakarazfaka, Festgabe für Adelheid Mette, herausgegeben von Christine Chojnacki, Jens-Uwe Hartmann und Volker M. Tschannerl, Swisttal-Odendorf, 2000, p. 303.
[16] "於一年中分為三際。謂寒熱雨各有四月。十二月中六月減夜。以一年內夜總減六。"引自玄奘譯《阿毘達磨俱 舍論》，《大正新修大藏經》第29卷，Nr. 1558, 第62頁中欄。
[17]. 季羨林等校注《大唐西域記校注》，中華書局1995年版，第169頁。
[18] As explained by Sir. M. Monier-Williams under the word "[na ". Sanskrit English Dictionary, p.221.
[19] Cf. Vogel, "On the Date of the Po2adha Ceremony," op. cit., p.677.
[20] Quoted from $dul_ba4i mdo4i rgya_cher 4grel_pa，Vinayas[trapiw]kq, 《西藏大藏經》（The Tibetan Tripitaka Peking Edition）, edited by Daisetz T. Suzuki, Tokyo-Kyoto Tibetan Tripitaka research Institute 1957，Volume 124，p. 175b.
[21] 《阿毘達磨俱舍論》，《大正新修大藏經》第29卷，Nr. 1558, 第62頁中欄。
[22] Abhidharnako1a & Bhq2ya of Qcarya Vasubandhu with Sphutqrthq commentary of Qcqrya Ya1omittra,critically edited by Swqm] dwqrikqdqs !astri, Bauddha Bharati Series-5,6,7,9. Varanasi, India 1987, p.537.