传世的敦煌《汉书》唐写本有十一件,它们有着较高的辑佚、校勘、文字学、音韵学、版本学等方面的学术价值。敦煌《汉书》唐写本的校勘价值远非吐鲁番《汉书》唐写本所能媲美。利用敦煌《汉书》唐写本,可校中华本《汉书》中存在的讹脱衍倒等问题。

TheHistoryofHanDynasty writtenonpaperinTangDynastyinDunhuanghaselev-enmanuscriptssofar,whichhaveahigheracademicvalueincompilation,collation,philology,pho-nology,versionandotheraspects.Thisarticlefocusesonthevalueofitscollation.Comparedwith threepiecesof“HanShu”writtenonpaperinTangDynastyinTurpan,collationvalueof“HanShu”writtenonpaperinTangDynastyinDunhuangismuchbetterthan“HanShu”writtenonpaperin TangDynastyinTurpan.Itcanbeusedforcollatingsomeproblemssuchasincorrectcharacters,la-cunacharacters,charactersaddedincorrectlyandupside-downcharactersorsentences.

中国20世纪初前后敦煌文献出现之前的古写本多由日本流入，日本学者或以为《浙藏敦煌文献》中有古笔了仲题跋言为最澄大师经卷者即来自日本。本文据最澄原作比较，确定此非最澄书，并以为也非日人古写本，而是我国唐时人书。古笔是二百多年前的人，不可能在1902年题跋，故可认为是假冒。原有藏家题跋也只字不提最澄与古笔，可证其收藏时并未见所谓的古笔题跋，当为后人造假。

Most manuscripts in China before the early 20th century came from Japan. Therefore, some Japanese scholars think that, the manuscript supposed to be Saicho’s with Kohitsu Ryouchuu’s epilogue among the Dunhuang Manuscrips Collected in Zhejiang Province also came from Japan. By comparing it with Saicho’s works, this paper confirms that it was neither Saicho’s manuscript nor an ancient Japanese manuscript, but a Chinese manuscript of the Tang dynasty. Kohitsu Ryouchuu lived two hundred years ago, and could not have written the epilogue in 1902. The epilogue of its original keeper mentioned nothing about Saicho and Kohitsu Ryouchuu, and this indicates that the original keeper did not see Kohitsu Ryouchuu’s epilogue when he got the manuscript, which might have been added later.

探究《古小说钩沉》之文献出处，不难发现鲁迅在辑校《列异传》、《幽明录》、《玄中记》等三种小说时，征引了当时新发现不久的“敦煌石室所出唐写本类书残卷”，这种类书残卷，正是被伯希和掳走之后又被罗振玉抄录的《修文殿御览》。考察《敦煌类书》、《〈古小说钩沉〉手稿》、《鲁迅手稿全集》等资料，我们一方面可以看到敦煌遗书对于唐前古小说文献的保存之功，另一方面不由惊叹鲁迅在收藏和利用敦煌文献方面的兴趣与卓识。

Exploring the sources of the Guxiaoshuo Goucheng ( Compilation of Ancient Novels) obviously suggests that when Lu Xun was compiling and correcting the three kinds of novels named Lieyi Zhuan, Youming Lu, and Xuanzhong Ji, he quoted “various manuscripts of Tang dynasty reference books” then newly discovered from the Dunhuang Library Cave, namely, the fragments taken away by Pelliot and then copied by Luo Zhenyu as Xiuwendian Yulan. An investigation of Dunhuang Reference Books, Draft of Guxiaoshuo Goucheng as well as the Full Collection of Lu Xun’s Handwritten Drafts not only demonstrates Dunhuang documents’ contribution to the preservation of ancient literature, but also indicates Lu Xun’s interest and insight in collecting and using Dunhuang documents.

透过对现存敦煌文殊作品的观察，考量其所在位置，归纳作品构成形式与图像元素的发展概况。初唐时，除第220窟外，都见有法华经变作品，说明与法华经关系较密切。盛唐时，狮子图像与初唐一样是最明确的特征，也见有一组后来常见的大菩萨与俗装众随侍以及昆仑奴，但未形成固定模式。昆仑奴的出现，说明佛教造像除以佛经为文本外，亦纳入社会流行的题材。中唐时，与法华经变同时出现的情形减少，见有五台山图的文殊或在背景中加入更多山水云彩表现，应与五台山文殊信仰有关。晚唐时，无论其所在位置或作品组合要素，皆延续中唐发展。五代宋时，增加于阗王驾狮子与善财童子，或佛陀波利与大圣老人的形象。依敦煌写本资料，可知是与中原往来密切有关。

This paper presents a summary of the development of the composition and iconographic elements of the Manjusri images at Dunhuang by studying the extant works and their positions. In the Early Tang period, the image of Manjusri appeared together with the Lotus Sutra illustration except in cave 220, suggesting a close relationship between Manjusri and the Lotus Sutra illustration. In the High Tang dynasty, the lion was the most specific character of the Manjusri images just as it was in the Early Tang period. There was also a group of great bodhisattvas with their followers in secular clothes and the slave of Kunlun, which was not stylized then but common in later periods. The appearance of the slave of Kunlun indicates that Buddhism not only based its images on Buddhist scriptures, but also borrowed popular themes from social life. In the Middle Tang period, the paintings with both Manjusri and the Lotus Sutra illustration decreased, and the appearance of Manjusri together with the map