Mayanagi Makoto (2005) ‘The three juan edition of Bencao jizhu and excavated sources’ in Vivienne Lo and Christpher Cullen,
Medieval Chinese Medicine - The Dunhuang medical manuscripts, RoutledgeCurzon, London and New York, 306-321.

Wellcome History Issue 33, p.22, Winter 2006
Book reviews, Medieval Chinese Medicine: The Dunhuang medical manuscripts
By CHEN MING
…By comparing and contrasting various ancient manuscript recensions of Tao Hongjing's Bencao Jizhufrom Dunhuang, Turfan and the ruins of Fujiwara Villa in Japan, Mayanagi Makoto reaches some interesting conclusions about the exchange of medical knowledge between China and Japan. Resources such as the Dunhuang manuscripts overcome a tendency simply to rely on a limited repertoire of historical material, and certainly contribute to restoring or reconstructing the medieval Chinese medicinal scene. Non-specialist readers can gain from the book a more rounded knowledge of healthcare in medieval China. …

The three juan edition of Bencao jizhu

and excavated sources¹

Mayanagi Makoto

Translated by Sumiyo Umekawa

Introduction

Bencao jizhu 本草集注 (An Annotated Collection of the Pharmaceutical Canon), the text that laid the foundations of mainstream Chinese materia medica was edited by Tao Hongjing 陶弘景 (456–536), sometime aroundCE 500.²On the basis of the Dunhuang 敦煌, Turfan 吐魯番 and Japanese editions of this book, discussed below, we know that this work was widely distributed. However, with the circulation of Xinxiu Bencao 新修本草 (Newly Revised Pharmacopoeia), a materia medicacompiled by Su Jing 蘇敬 at the imperial edict of the Tang government in 659, Bencao jizhu fell out of use and gradually disappeared.³ Despite serious scholarly debate about the number of scrolls and circumstances of the construction of Bencao jizhu, in the absence of original manuscripts it was not possible to draw definitive conclusions. By throwing new light on excavated materials stored in Japan and other locations, this chapter attempts to resolve the debate.

The Dunhuang scrolls

It was originally believed that the British archaeologist, Aurel Stein purchased the Dunhuang edition of Bencao jizhu and that it was held in the British Museum.⁴ It later emerged that the scrolls were held at the Omiya library of Ryukoku University in Kyoto.⁵It seems that this scroll was part of the collection found in the Mogao 莫高 cave in Dunhuang. Some of the collection were bought in 1908 and brought to Japan by Tachibana Zuicho 橘瑞超, who was exploring the Western Region and Central Asia at the command of Otani Kouzui 大谷光瑞, the head priest of Nishi Hongan Ji 西本願寺.⁶ It is also said that it was Yoshikawa Shoichiro 吉川小一郎, on his journey to collect Tachibana in 1912, who had bought the collections from Wang Yuanlu 王圓籙 the Daoist priest who discovered the cave containing the manuscripts.⁷ But returning to Japan earlier than Yoshikawa it was, in fact, Tachibana who brought them to Japan.
As we can see in Figure 13.1, at the end of the text there is a note stating:

開元六年（718）九月十一日尉遲盧麟／於都寫本草一卷　辰時寫了記
The 6th year of Kaiyuan (CE 718), on the 11^thday of the 9^th month, Yu Chi Lu Lin. Hand copied edition of one juan of Bencao recorded at the capital. Written during the hours of chen [7–9 a.m.].

Figure 13.1 The Dunhuang edition of Bencao jizhu

Due to the light colour of the initial two graphs 開元 Kaiyuan, and a number of graphs in the second line, there was some doubt as to the authenticity of the dating. Detailed analysis of the calligraphy demonstrates clearly that the script of the colophon is identical to that of the main passages and confirms the early dating of CE718.⁸ Ma Jixing 馬繼興 gives another reason for the early dating of the manuscript: the graph zhi治, the posthumous name for the Tang emperor Gaozong 高宗 is not avoided.⁹However, Liao Maoxin 梁茂新 has argued convincingly that official regulations concerning exceptions that could be made to the rule of avoiding posthumous names when copying original texts were established at the order of the emperor Taizong 太宗 in 626 and by the imperial decree of Gaozong 高宗 in 660 and　recorded in Tang Hui Yao 唐會要 (Collected Essentials Relating to the Tang).¹⁰ The relevant ruling was that it was not necessary to avoid terms such as shi 世 and min民 unless they were used together, as in the deceased emperor’s personal name (Li) Shimin (李世民). Similarly the posthumous name for the first emperor Gaozong 高宗, zhi 治, was not tabooed when making a new copy of an old text. Therefore, it was not necessary to avoid zhiin this manuscript at this time. We, therefore, cannot use Ma Jixing’s argument to support our thesis.
The Dunhuang Bencao jizhuis made up of 51 sheets of paper pasted together into a scroll, and is about 28 cm wide and 1,997 cm long. The sheets of paper (each 28 cm × 40 cm) are identical to the papers used for Tang period bureaucratic reports and are made from tapa 楮 (a paper from the mulberry tree) and indented with regular lines traversing the paper. It seems that 32 graphs from the first three lines are missing, leaving 720 lines plus the identification. But, as we can see from the right-hand side of Figure 13.2, there is a different quality of paper in the five papers of the 46th to 50th inclusive. This section, corresponding to the juanknown as Yao dui 藥對 ‘Drug pairing’, seems to have been a later amendment replacing most of the 45th sheet. Yao duirecords the effect on the potency of specific drugs when mixed in different combinations. The amended sheets are written in the same handwriting, but the distance between each graph is greater and the quality of paper is worse. According to Sakurai, this is evidence that the amendments are not merely commentaries overlooked in the first copying, but that the yao duisection represents the interjection of additional material from another edition of Bencao jizhu.¹¹

Figure 13.2 Yao dui 藥對 ‘Drug pairing’ is on a different quality of paper and seems to be a later addition

Figure 13.3 Biqiu hanzhu jieben比丘含注戒本 (Bhiksu’s book of Precepts with Commentary) on the back of Dunhuang Bencao jizhu

On the back of the Dunhuang Bencao jizhu, as seen in Figure 13.3, there is a copy of Biqiu hanzhu jieben 比丘含注戒本 (Bhiksu’s Book of Precepts with Commentary, hereafter Jieben) narrated by a monk Shi Daoxuan 釋道宣 of Mt Taiyi 太一. This is a Buddhist text currently held at Ryukoku University. It might be that when Bencao jizhu was made redundant by the newer and extended Xinxiu Bencaoin 659, the back of the scroll was used for copying. Jieben begins at the bottom of the back of Bencao jizhu and finishes on the back of juan 50 of Dazhidu lun 大智度論 (Treatise on Salvation through Great Wisdom) which is adjoined to Bencao jizhu. Presumably because the original first sheet of Bencao jizhu suffered some damage, it has been trimmed where it joins to Dazhidu lun. This explains why the first three lines are missing. Luo Zhenyu 羅振玉, borrowing photographs of this manuscript from the geographer Ogawa Takuji 小川琢治, printed the Dunhuang edition of Bencao jizhu in his Jishi’an congshu吉石庵叢書 (Collected Editions from the Hermitage of the Lucky Stone) in 1916. However, there are some problems with his transcription, in as much as it omits the 66th line, alters graphs and supplements whole sections. In spite of these problems, the Jishi’anedition was reprinted in 1955 by Fan Xingzhun 范行準 and used for academic research. In 1997, a transcription, commentary and translation was made from the original manuscript held by the University of Ryukoku.¹²

Bencao jizhu: determining the number of juan

Problems

Based on the content of the Dunhuang edition, Takahashi originally argued: ‘there were two versions of Tao Hongjing’s Jizhu; a three juan version and a seven juanversion and both of them had a commentary; the circulation of the former was small, while the latter enjoyed widespread distribution’.¹³Okanishi disagreed:

The three juan materia medica might be the manuscript cited in Suishu 隋書 (History of the Sui Dynasty) as Shen Nong Bencao jing (Shen Nong’s Pharmaceutical Canon) in three juan which is different from Jizhu. . . . It is more natural to think that Tao Hongjing first edited a three juan pharmacopoeia, Shen Nong Bencao jing: sanjuan, on the basis of a former edition of Shen Nong Bencao jing, including sentences from Mingyi bielu名醫別錄 (Additional Records of Famous Physicians, attributed to Tao Hongjing c.510) and later made a seven juan edition supplementing his own commentary (Bencao jizhu). The seven juan materia medica is more useful than the three juan materia medica which does not have a commentary, and it is therefore likely to have been distributed more widely.¹⁴

It appears that Takahashi conceded to Okanishi’s opinion. Later on, however, Watanabe concludes: ‘Tao Hongjing had firstly edited three juan of Bencao jizhu, but later re-edited it into a seven juan because each juan of the former was too long and thus not convenient’.¹⁵ Recently Liao Yuqun came to the same conclusion as Watanabe.¹⁶ Although there are still disagreements over the various recensions, Okanishi’s assumptions, discussed in his work, Honzou gaisetsu 本草概説 (Materia Medica– Abridged and Summarised) remain the most influential.¹⁷There is some new evidence that suggests that the three juan Bencao jizhu was in wide circulation, which may refute Okanishi’s position. I will set out the argument below.

Tao Hongjing’s own comments

In Tao Hongjing’s preface to Zhouhou baiyi fang 肘後百一方 (101 Remedies to Keep up your Sleeve, c.500), Tao states: 凡如上（藥物加工）諸法、皆已具載在余所撰『本草』上卷中, ‘All the drug processing methods given above, are already recorded in the juanA of the Materia Medica that I edited’.¹⁸ By mentioningshangjuan 上卷 ( juan A) of his work, we know that the book was originally either a two juan edition with shang A and xia 下 B or a three juan edition with shang A, zhong 中 B, and xia C. ‘Shang’ juan ( juan A) refers to the Dunhuang edition of Bencao jizhu. Here, the 221st to 222nd lines state: 今撰此三卷、幷『效驗方』五卷、又『補闕葛氏肘後（肘後百一方）』三卷 ‘This three juan text that I now edit aligns the content of Xiaoyan fang in five juan and Buque geshi zhouhou (Zhouhou baiyi fang) in three juan’.¹⁹ This text was probably written in the same year as the preface to Zhouhou baiyi fang (c.500) and consisted of three juan. Here Tao Hongjing refers to a three juan text that he is editing together with two other prescription books. About this three juan edition, the 24th to 29th lines state:

今輒苞綜諸經、研括煩省。以『神農本經』三品（合）三百六十五爲主、又進名醫副品亦三百六十五、合七百卅種。精麁（粗）皆取、無復遺落、分別科條、區畛物類、兼注詺世用土地、及仙經道術所須、幷此序錄、合爲三卷。
Now I gather up the various canons and include those thorough investigations and omit vexatious details. I have mainly used 365 from the ‘three grades’ sections (combined) of Shen Nong Bencao jing as a base and put in another 365 from Mingyi fupin(Adjuvant Drugs from Renowned Physicians), which comes altogether to 730 varieties. I have selected both fine and crude without leaving any out again, dividing and separating into an order and assigning boundaries according to the nosology of ‘things’. Additionally, there is a commentary on applications and the soil, including essentials from the canons of the immortals and from the arts of Daoism. Together with this preface, I combine them into three juan.²⁰

The 34th to 35th lines state:

右三卷、其中下二卷、藥合七百卅種、各別有目錄、並朱墨雜書、幷子注、大書分爲七卷。
The 730 varieties of drug combinations in the latter two juan from the three juantext on the right-hand side, both separately contain catalogues, and align miscellaneous writings in red and black inks with commentaries in small graphs. I have enlarged the writing and divided it into seven juan.²¹

Tao Hongjing’s comments tell us that this Bencao jizhu was originally a threejuan text, of which we know that the first juan consisted of a preface since it contained the passages translated above. The second and third juanconsisted of 365 varieties of ‘drug processing’ selected from Shen Nong benjing written in the red ink. Another 365 varieties of ‘medical processing’ were taken from Mingyi fupinand written in black ink. Tao Hongjing’s own commentaries, ‘Additionally, there is a commentary on applications and the soil, including essentials from the canons of the immortals and from the arts of Daoism’, are given in smaller script.²²
It is also clear from Tao Hongjing’s supplementary notes that his commentaries to the second and third juan were later enlarged and re-edited into the seven volume edition: ‘(I) have enlarged the writing and divided it into seven juan’, with another commentary written in small script. From a note at the 720th line which states not shang juan (juan A), but Bencao jizhu diyi xulu 本草集注第一序錄 ‘Bencao jizhu: no.1 Preface’, we can assume that the Dunhuang manuscript just contains the preface and that this is the first juan of a seven juan version of the Bencao jizhu.²³ From the references to the three juanversion quoted above, we may also assume that the two editions had the same preface. Thus, the evidence of commentaries made by Tao Hongjing shows that he himself edited and commented upon both a three and a seven juan edition of Bencao jizhu. We can therefore refute Okanishi’s interpretation that the three juan version might have been the book seen in Suishu as Shen Nong Bencao jing sanjuan and not Jizhu. I will argue that it was, on the basis of bibliographies contained in the dynastic histories.

Bibliographies in the dynastic histories

Let us consider the Bencao materia medica (pharmacopoeia) texts attributed to Tao Hongjing, or Tao Yinjju 陶隱居 as he is sometimes referred to in Chinese and Japanese bibliographies in the official dynastic histories. The Jingjizhi經籍志 ‘Record of Canonical Books’ inSuishu (656) states, ‘In the Liang 梁 period (502–555) . . . there was Tao Yinjju’s Bencao: ten juan. . . Tao Hongjing’s Bencao jing jizhu: seven juan’.²⁴ The entry for the fifteenth day of the fifthth month of the Emperor Kanmu’s reign period (787), in Shoku Nihongi 續日本紀 (A Later History of Japan) (797) records: ‘The Ministry of Medicine states:Xinxiu Bencao with commentary by Su Jing 蘇敬 and Tao Yinjju’s Jizhu Bencao (Pharmacopoeia, Annotated and Collected), select together and increase by 100 strips . . . please go forth and use it, (here is) permission’.²⁵ Nihonkoku Genzaisho Mokuroku 日本國見在書目錄 Bibliography of Texts Seen in Japan (875–891) states:

神農本草七（卷）陶隱居撰、…本草夾注音一（卷）陶隱居撰、…注本草表序一（卷）陶隱居撰
Shen Nong Bencao: seven (juan) edited by Tao Yinju . . . Bencao jiazhuyin 本草夾注音 (Pharmacopoeia with Interspersed Commentary and Phonetic Entries): one (juan) edited by Tao Yinju. . . . Zhu Bencao biaoxu 注本草表序 (Annotated Pharmacopoeia with Table and Preface): one (juan) edited by Tao Yinju.²⁶

‘Record of Canonical Books’ in Jiutangshu舊唐書 (Old History of the Tang (940–945) ) states: ‘Bencao jijing: seven juan edited by Tao Hongjing’.²⁷ ‘Yiwenzhi’ 藝文志, the bibliographical treatise of Xintangshu新唐書 (New History of the Tang (1060) ) records: ‘Tao Hongjing’s Jizhu Shen Nong Bencao 集注神農本草 (Shen Nong’s Pharmacopoeia, Collected and Annotated) in seven juan’.²⁸
References to Bencaotexts attributed to Tao Hongjing in later dynastic histories such as the ‘Yiwen zhi’ in Songshi 宋史 (History of the Song) are simply copied from earlier dynastic histories. The entries that seem to refer to Bencao jizhuinclude:

a seven juan Bencao jing jizhu recorded in Suishu

Jizhu Bencao 集注本草 (Pharmacopoeia, Annotated and Collected) recorded in Shoku Nihongi 續日本紀

a seven juan Shen Nong Bencao神農本草 (Shen Nong’s Pharmacopoeia) in Nihonkoku Genzaisho Mokuroku 日本國見在書目錄

a seven juan Bencao jizhu本草集注 (Collected Pharmaceutical Canon) in Jiutangshu 舊唐書 (Old History of the Tang)

a seven juan Jizhu Shen Nong Bencao集注神農本草 (Shen Nong’s Pharmacopoeia, Collected and Annotated) in Xintangshu新唐書 (New History of the Tang).

With the exception of Shoku Nihongi, each entry records a seven juanwork, and includes the term ‘Bencao’ in the title. In contrast Shen Nong神農 ‘Shen Nong’, ‘Jizhu’ 集注 ‘Annotated Collection’ and ‘Jing’ 經 ‘Canon’ are not common to these titles. It is therefore likely that Okanishi did not realize that there was a three juanversion of Bencao jizhu.

Wooden strips excavated from the ruins of Fujiwara Villa

As we can see in Shoku Nihongi續日本紀 (A History of Japan, continued), the Ministry of Medicine in Japan reported to the throne on the fifteenth day of the fifth month, CE 787 that ‘they would like to adoptXinxiu Bencao commented by Su Jing which has hundreds more sections than Jizhu Bencao of Tao Yinju’. This is evidence that Jizhuhad been in use in Japan at the Ministry of Medicine sometime around CE787. Therefore no one has questioned the assumption that the Bencao studied by students of medicine at the Ministry of Medicine in Taihou 大寶 Command (established in CE 701) was Bencao jizhu.²⁹
Moreover, wooden strips excavated from the ruins of Fujiwara Villa ( fl. 694–710) contain phrases such as ‘taihou sannen’ 大寶三年 which indicate the year 703, ‘tenyaku’ 典藥 ‘the drugs’ and ‘(Ben)caoji Bencao jizhu . . . Ben Bencao（本）草集　本草集注…本本草’ (no.72). Another contains the phrase ‘Bencao jizhu shangjuan’ 本草集注上卷 (Annotated and Collected Pharmacopoeia: juan A).³⁰These are materials related to the Ministry of Medicine, found all together in a drainage ditch (SD105) which fell out of use around CE 703, since one wooden strip bears that date. Additionally, according to the database for wooden strips set up by the Nara National Research Institute of Cultural Properties, Japan, the term Bencao is only written on four strips (including no.72 and no.74), of a total 25,755 excavated from the SD ditch at the ruins of Fujiwara Villa.³¹

Figure 13.4 Wooden strips excavated from the ruins of Fujiwara Villa ( fl. 694–710)

A facsimile of no.74 is on the right-hand side of Figure 13.4 with my transcription to the left-hand side.³² The upper part of the wooden strip (164 mm × 21 mm × 2 mm) remains intact. Unfortunately the lower part is damaged and, according to my estimation, does not contain any words below ‘Bencao jizhu shangjuan’ (Annotated and Collected Pharmacopoeia: juan A). No record suggests that there was any other text with the same title as Tao Hongjing’sBencao jizhu at that time or in earlier periods. In addition, according to Tao Hongjing himself and on the basis of the total number of graphs in one edition (a matter which I will give detailed attention later), we will rule out the possibility that Bencao jizhu could have been a two juan work. It is also unlikely, given the value that the Japanese placed on importing authentic Chinese cultural artefacts, that a sevenjuan edition was intentionally edited down into a three juanedition. Thus, the wooden strip edition of Bencao jizhu shangjuan must be the proof of the existence of the three juanversion of Bencao jizhu and we must bury Okanishi’s assumption once and for all.

Broken strips excavated at Turfan

General condition³³

About a hundred years ago, Central Asia and the Western Regions of China were a mecca for archaeological explorations: expeditions from Britain, France, Germany, the United States, Japan and China all unearthed huge finds. Those made by the German explorers A. Grunwedel and von Le Coq on their fourth expedition to Turfan between 1902 and 1912, were stored at Preußischer Akademie der Wissenschaften in Berlin. Kuroda Genji photographed four Chinese texts at the Akademie in 1933. He reported that one of the fragments (T II T) is a book on miscellaneous subjects written both in red and black ink that ‘shows the original state of Tao Yinju’s Jizhu Shen Nong Bencao jing written in the Six Dynasties period’.³⁴ We cannot draw definitive conclusions on the basis of Kuroda’s report since there seem to be a number of errors and the poor quality of the black and white photographs makes verification impossible.
Since the Berlin Akademie was divided following the partition of Germany, inquiries at the Staatliche Museen zu Berlin, Preußischer Kulturbesitz in West Berlin alone were fruitless. After the reunification of East and West Germany in 1990, I discovered that the medical texts were stored at the Akademie der Wissenschaften in East Berlin until 1992, at which time they were moved to Staatsbibliothek zu Berlin Preußischer Kulturbesitz Orientabteilung. The director of the Staatsbibliothek, Dr Hartmut-Ortwin Feistel wrote to me that the reference T II T of Kuroda’s reports had changed to reference number Ch1036r. In June 1993 I received a full-size colour copy (Figure 13.5) and permission to use it for academic purposes. I was lucky enough to see it in Berlin with my own eyes in March 1998. We can see in Figure 13.5 that this text does indeed differentiate the sentences from Shen Nong Bencao jing andMingyi bielu with red and black ink, and sentences from ‘yao dui’ and commentaries are written in thinner ink and arranged in double lines.
However, the other extant Dunhuang edition of Bencao jizhu, held at Ryukoku University, is not differentiated with red and black ink. Some later editions of Xinxiu Bencaokept the tradition of using red and black ink: the Dunhuang edition, Pelliot 3714, is written in this style, while Stein 4534 and the Ninnaji 仁和寺 edition held in Kyoto have both dropped the style. The question remains: to which period can we attribute the Turfan edition which keeps the original style of a book on miscellaneous subjects written in red and black ink? Words likezhuzhi 主治, shihu 世呼, shibufushi 世不復識 and shizhong 世中 appear in this edition. Zhi is the posthumous name of the Tang emperor Gaozong, Li Zhi 李治 while shi is that of the Tang emperor Dazong, Li Shiming 李世民. Therefore, Xinxiu Bencao, compiled on the edict of the Tang government, carefully avoids these graphs and uses zhu 主, suhu 俗呼, subushi 俗不識 andsuzhong 俗中 instead. Six Dynasties influence can also be seen in the calligraphic style. On the basis of these two matters, the failure to avoid Tang taboos and the calligraphic style, Kuroda believes that this is a Six Dynasties copy.³⁵ Ma Jixing agrees that this is a pre-Tang copy.³⁶

Figure 13.5 The Turfan edition of Bencao jizhu

On the other hand, Watanabe maintains that Kuroda is too hasty. In his opinion, ‘it is a careful copy of a Six Dynasties edition which preserves the Tang or pre-Tang style of Tao Hongjing’.³⁷ It is also dangerous to assume the date of the Dunhuang version of Bencao jizhu on the basis of posthumous name avoidance, since we are not considering an officially sponsored compilation like Xinxiu Bencao and therefore need not necessarily adhere to the imperial edict. I agree with Watanabe’s assumption, and consider that it is an early Tang edition, c. seventh century. This view is supported by the opinion of palaeographic specialists who identify early Tang script with remnants of Six Dynasty calligraphic styles.³⁸ However, it cannot be disputed that the Turfan edition is the oldest extant Chinese materia medica.

In the lineage of a three juan or seven juan edition?

I would now like to consider whether the Turfan edition is in the lineage of a three juan edition or not. As seen in Figure 13.5, the Turfan edition cites the commentaries of Tao Hongjing in double lines in a thinner script. Since the Dunhuang edition of Bencao jizhu refers to enlarging the small graphs of the commentaries to the three juanedition to make the seven juanedition, it would seem to follow that the Turfan edition is in the three juanlineage. However, calculating on the basis of the whole length of the edition estimated from the number of graphs and the size of the Turfan text, Watanabe suggests that the Turfan edition may be pieces of the sixth juan taken from the seven juan edition.³⁹Ma Jixing discusses the Turfan text as the fourthjuan of a seven juan edition, but he does not set out his evidence.⁴⁰ Both Watanabe and Ma refer to the size of the Turfan text given in Kuroda’s reports (i.e. 27 cm in length and 28.5 cms in width).⁴¹ In measuring from the original, I found the longest part of the Turfan text is 27.8 cms in length and 27 cms in width. Kuroda’s measurements are therefore questionable. Furthermore, he also reported the size of the lost Mingyi fupin名醫副品 (Ch1036v),⁴² which is written on the back of the Turfan text and must therefore be the same size, as 23.7 cms in length and 22.7 cms in width. He clearly failed to appreciate that they were the same manuscript. On the other hand, because Kuroda’s photographs are solarized, Watanabe’s calculations are made on the basis of the reported size, and not the size of the section that contains the script. I have therefore calculated again on the basis of a survey of the original manuscript.
We have seen that the preface of the Dunhuang edition actually records that the three juanedition of Bencao jizhu was made up a firstjuan solely containing prefatory material, with the second and the third juan making up the content; we can also see that it itself is a seven juan edition which devotes the first juan to a preface and that all the rest is content. Since they have a common preface, we can determine whether the Turfan remnants of Bencao jizhu came from a threejuan or seven juan edition by testing whether the number of graphs making up the second and the last juanfit into an estimate of the total size of the manuscript. The text written on the Turfan manuscript is 26.4 cm in width and 22.4 cm in length. Each line can fit 20 graphs in large characters, and 28 letters in small characters. Thus, we can estimate that that the Turfan remnant could fit 240 large graphs arranged in 12 lines or 679 small graphs arranged in 24 lines.
According to the reconstruction of Bencao jizhu by Mori Tatsuyuki 森立之 and others, Watanabe estimates that there are 33,000 large graphs and 37,000 small graphs in the main content of Bencao jizhu, excluding the preface.⁴³ By calculating according to this evidence, the main part of the Turfan manuscript originally must have consisted of a combination of 36.3 m of large graphs and 14.5 m of small graphs, i.e. 51 m when combined in an integrated text. For the purposes of determining the length of each scroll, if we estimate the length with no small graphs at all in the text, the manuscript could contain a total of 2,320 lines. Dividing the total number of graphs by two to estimate the number in the second and last juan, each juan would consists of 1,160 lines in 25.5 m. This seems to be about the upper limit for a single juan in scroll form; the Dunhuang edition of Bencao jizhu, for example, which seems to correspond to the preface, consists of 720 lines alone and is about 20 m long. In addition, the longest extant scroll manuscript containing texts that describe medical matters is the Nakarai 半井 edition of Ishinpo醫心方 (Remedies at the Heart of Medicine) which is 24.1 m long and contains a total of 25 juan, in 1,417 lines. On the basis of these measurements it is possible that the Turfan edition was a three juanedition, and the fact that the commentaries are written in small graphs adds weight to this view. Needless to say, because of the unusual length of the second and the third juan, the three juan edition was ultimately revised into sevenjuan by enlarging the graphs in the commentary.
If, as I have argued, the Turfan edition is the three juanedition of Bencao jizhu, there is evidence to show that it is a remnant of the last juan. The text contains a report about four varieties of insects. The 31st to 34th lines of the Dunhuang Bencao jizhu, which contain the table of contents for the three juan edition, place the section about insects in the third juan.⁴⁴Therefore, this is supplementary evidence to conclude that the Turfan descriptions of insect come from the third and last juanof the three juan version. It is a matter of great interest that all the evidence for a three juanedition of Bencao jizhu has been excavated in Japan far to the east of China and in Turfan located on the western frontiers. The excavated materials throw new light on the history of medical exchange between China and Japan and on the distribution networks for the three juanedition of Bencao jizhu.

Conclusion

Re-examining both Chinese and Japanese sources, I have resolved the longstanding debate over the circumstances and dating of the excavated editions of Bencao jizhu. My conclusions are:

1 The Dunhuang text possessed by Ryukoku University was copied in 718.
2 It is clear from references in the Dunhuang texts and wooden strips found from Fujiwara Villa, that there was a three juan edition ofBencao jizhu and that the seven juan edition evolved from it. We can therefore give greater credence to a popular view.
3 The Turfan text held in Staatsbibliothek zu Berlin Preußischer Kulturbesitz Orientabteilung was copied in the early Tang, some time in the seventh century, and contains pieces of the third juan of the threejuan version of Bencao jizhu.
4 The evidence for the existence of a three juan edition circulating to the east and west of China allows us to reassess the scope of medical exchange between China and Japan.

Notes

1 This chapter is a revised version of a paper given at the 94th conference of the History of Japanese Medicine (Kanazawa, 15 May 1993), later published in Nihon ishigaku zasshi 日本醫史學雜誌 (Journal of the Japan Society of Medical History) 1993.39–1, 26–28.

2 Shang Zhijun 尚志鈞, Lin Ganliang 林乾良 and Zheng Jinsheng 鄭金生 (1989),Lidai Zhongyao Wenjian Jinghua歷代中藥文獻精華, Kexue Jishu Wenjian Chupanshe, Beijing, 162.

3 Ibid., 162.

4 Kuroda Genji 黑田源次 (1935) ‘Puroshia gakushiin shozou chuou ajia shutsudo ihosho yonshu 普魯西學士院所藏中央亞細亞出土醫方書四種’ in Shinagaku 支那學 vol. 7. no. 4, 633–665.

5 Mayanagi Makoto 眞柳誠 (1994) ‘Tonkou bon [Honzou Shichu] 敦煌本『本草集注』’ inKampo no rinshou 漢方の臨牀 vol. 41–12, 1522–1524. http://www.hum.ibaraki.ac.jp/hum/mayanagi/paper04/shiryoukan/me079.html

6 Okanishi Tameto 岡西爲人 (1977) Honzou gaisetsu 本草概説, Sougensha, Osaka, 53.

7 Akahori Akira 赤堀昭 (1997) ‘Tonkou bon [Honzou Shichu] kaisetsu 敦煌本『本草集注』解説’ in Ueyama Daishun ed., Ryukoku daigaku zenpon sousho 16 tonkou shahon honzou shichu jyoroku/Bikyu Gan chukai bon龍谷大學善本叢書16　敦煌寫本本草集注序錄・比丘含注戒本 (Facsimile Series of Rare Texts in the Library of Ryukoku University 16, The Bencao jizhu xulu and The Biqiu hanzhu jieben from Dun-huang with Plates and Comments), Houzoukan, Kyoto, 220–231.

8 Fujieda Akira 藤枝晃 (1997) ‘Shahon Kaidai 寫本解題’ in Ueyama Daishun ed.,Ryukoku daigaku zenpon sousho 16 tonkou shahon honzou shichu jyoroku/Bikyu Gan chukai bon 龍谷大學善本叢書16　敦煌寫本本草集注序錄・比丘含注戒本(Facsimile Series of Rare Texts in the Library of Ryukoku University 16, The Bencao jizhu xulu and The Biqiu hanzhu jieben from Dun-huang with Plates and Comments), Houzoukan, Kyoto, 207–219.

9 Ma Jixing 馬繼興 (1988) Dunhuang guyiji kaoshi 敦煌古醫籍考釋 (Annotations on the Ancient Medical Texts from the Dunhuang Cave Library), Jiangxi kexue jishu chubanshe, Nanchang, 337.

10 Liao Maoxin 梁茂新 (1983) ‘[Bencaojing Jizhu] xieben niandai kaoyi ≪本草經集注≫寫本年代考異’ in Zhonghua yishi zazhi中華醫史雜誌 vol. 13–3, 181–182.

11 Sakurai Kensuke 櫻井謙介 and Kobayashi Seiichi 小林清市 (1997) ‘[Honzou Shichu] kanren shiryo koui 『本草集注』關連資料考異’ in Ueyama Daishun ed., Ryukoku daigaku zenpon sousho 16 tonkou shahon honzou shichu jyoroku/Bikyu Gan chukai bon 龍谷大學善本叢書16　敦煌寫本本草集注序錄・比丘含注戒本(Facsimile Series of Rare Texts in the Library of Ryukoku University 16, The Bencao jizhu xulu and The Biqiu hanzhu jieben from Dun-huang with Plates and Comments), Houzoukan, Kyoto, 232–239.

12 Akahori (1997), 220–231.

13 Takahashi Shintaro 高橋眞太郎 (1943) ‘Shinno honzoukei ni tsuite 神農本草經に就いて’ in Nihon ishigaku zashi日本醫史學雜誌 vol. 1320, 325–342.

14 Okanishi Tameto 岡西爲人 (1944) ‘Shinno honzoukei ni tsuite wo yomu 「神農本草經に就いて」を讀む’ in Nihon ishigaku zashi 日本醫史學雜誌 vol. 1323, 1–13.

15 Watanabe Kouzou 渡邊幸三 (1951) ‘Tou Koukei no honzou ni taisuru bunkengakuteki kousatsu 陶弘景の本草に對する文獻學的考察’ in Touhou gakuhou 東方學報 vol. 20, 195–222.

16 Liao Yuqun 廖育群 (1922) ‘Tao Hongjing Bencao zhuzuozhong zhuwenti de kaocha 陶弘景本草著作中諸問題的考察’ in Zhonghua yishi zazhi 中華醫史雜誌 vol. 22–2, 74–78.

17 Okanishi (1977), 53.

18 Tao Hongjing 陶弘景 (500), Zhouhou beijifang 肘後備急方 (Emergency Prescriptions to Keep up your Sleeve), Renmin weisheng chubanshe, Beijing, 1982, 7.

19 Sakurai Kensuke 櫻井謙介 (1997) ‘[Honzou Shichu] joroku shakumon『本草集注』序錄釋文’ in Ueyama Daishun ed., Ryukoku daigaku zenpon sousho 16 tonkou shahon honzou shichu jyoroku/Bikyu Gan chukai bon龍谷大學善本叢書16　敦煌寫本本草集注序錄・比丘含注戒本(Facsimile Series of Rare Texts in the Library of Ryukoku University 16, The Bencao jizhu xulu and The Biqiu hanzhu jieben from Dun-huang with Plates and Comments), Houzoukan, Kyoto, 255.

20 Sakurai (1997), 243.

21 Ibid.

22 Morohashi Tsuguji 諸橋轍次 (1986), Dai Kanwa jiten 大漢和辭典, Taishukan shoten, Tokyo, 3066.

23 Sakurai (1997), 289.

24 Wei Zheng 魏徴 (580–643) (compiled 629 – 636), Suishu 隋書, Zhonghua shuju, Beijing, 1973, 1040 –1041.

25 Fujiwara no Tsugutada 藤原繼縄 (1935) Shoku Nihongi 續日本紀, Yoshikawa koubunkan, Tokyo, 524.

26 Fujiwara no Sukeyo 藤原佐世 (1996) Nihonkoku Genzaisho Mokuroku 日本國見在書目錄, Meicho Kankoukai, Tokyo, 80–81.

27 Liu Xu 劉昫 (887–946) (compiled 940–945), Jiu Tangshu 舊唐書, Zhonghua shuju, Beijing, 1975, 2048.

28 Ou Yangxiu 歐陽脩 (1007–1072) (compiled 1043–1060), Xin Tangshu 新唐書 (New History of the Tang), Zhonghua shuju, Beijing, 1975, 1567.

29 Shinmura Taku 新村拓, Kodai iryou kanjinsei no kenkyu 古代醫療官人制の研究, Housei Daigaku shuppankyoku, Tokyo, 1983, 74–81.

30 Nara Ken Kyoiku Iinkai 奈良県縣教育委員會, Fujiwarakyu(naraken shiseki meishou tennenbutu chousa houkoku dai 25 satu) 藤原宮(奈良縣史跡名勝天然物調査報告第25冊)vol. 72, 74, Naraken kyouiku iinkai, Nara, 1969.

31 http://www.nabunken.go.jp/open/mokkan/mokkan2.html

32 Nihon Ishi Gakkai 日本醫史學會, Zuroku Nihon iji bunka shiryoushusei 圖錄日本醫事文化史料集成vol. 1, Sanichi shobou, Tokyo, 1977, 28.

33 Mayanagi Makoto 眞柳誠 (1993) ‘Genzon saiko no Chugoku honzou – Torufan shutsudo no [Honzou Shichiu] 現存最古の中國本草－トルファン出土の『本草集注』’ in Kampo no rinsho 漢方の臨牀 vol. 40–8, 1082–1084. http://www.hum.ibaraki.ac.jp/hum/mayanagi/paper04/shiryoukan/me064.html

34 Kuroda (1935), 633–665.

35 Kuroda (1935), 633–665.

36 Ma (1988), 384.

37 Watanabe Kouzou 渡邊幸三 (1954) ‘Chuou Ajia shutsudo honzou shichiu zankan ni taisuru bunkengaku teki kenkyu 中央亞細亞出土本草集注殘簡に對する文獻學的研究’ inNihon touyou igaku kaishi日本東洋醫學會誌 vol. 5–4, 35–43.

38 Professor Touno Haruyuki 東野治之 gave an expert opinion privately, judging that it seems likely to have been written as early as the end of the Sui dynasty and seemingly in the early Tang, although there is also evidence of the old calligraphic style of the Six Dynasties.

39 Mayanagi (1993), 1082–1084.

40 Ma (1988), 384.

41 Kuroda (1935), 633–665.

42 Kosoto Hiroshi 小曾戸洋, Mayanagi Makoto 眞柳誠 (1993) ‘Torufan Shutsudo no Ihousho - Cho Bunchu no Yihou トルファン出土の醫方書－張文仲の遺方’ in Kampo no Rinshou漢方の臨牀 vol. 40–9, 1218–1220.

43 Okanishi (1944), 1–13.

44 Sakurai (1997), 242–243.