It is probable that oral tradition was the main source from which al-Hujwírí derived the materials for his work. Of extant treatises on Ṣúfiism he mentions by name only the _Kitáb al-Luma`_ by Abú Naṣr al-Sarráj, who died in 377 or 378 A.H. This book is written in Arabic and is the oldest specimen of its class. Through the kindness of Mr. A. G. Ellis, who has recently acquired the sole copy that is at present known to Orientalists, I have been able to verify the reading of a passage quoted by al-Hujwírí (p. 341), and to assure myself that he was well acquainted with his predecessor’s work. The arrangement of the _Kashf al-Maḥjúb_ is partially based on that of the _Kitáb al-Luma`_, the two books resemble each other in their general plan, and some details of the former are evidently borrowed from the latter. Al-Hujwírí refers in his notice of Ma`rúf al-Karkhí (p. 114) to the biographies of Ṣúfís compiled by Abú `Abd al-Raḥmán al-Sulamí and Abu ´l-Qásim al-Qushayrí. Although he does not give the titles, he is presumably referring to Sulamí’s _ṭabaqát Al-ṣúfiyya_ and Qushayrí’s _Risála_.[15] The _Kashf al-Maḥjúb_ contains a Persian rendering of some passages in the _Risála_ of Qushayrí, with whom al-Hujwírí seems to have been personally acquainted. A citation from `Abdalláh Anṣárí occurs on p. 26. Footnote 15: Cf., however, p. 114, note. Manuscripts of the _Kashf al-Maḥjúb_ are preserved in several European libraries.[16] It has been lithographed at Lahore, and Professor Schukovski of St. Petersburg is now, as I understand, engaged in preparing a critical text. The Lahore edition is inaccurate, especially in the spelling of names, but most of its mistakes are easy to emend, and the text agrees closely with two MSS. in the Library of the India Office (Nos. 1773 and 1774 in Ethé’s _Catalogue_), with which I have compared it. I have also consulted a good MS. in the British Museum (Rieu’s _Catalogue_, i, 342). The following abbreviations are used: L. to denote the Lahore edition, =I.= to denote the India Office MS. 1773 (early seventeenth century), =J.= to denote the India Office MS. 1774 (late seventeenth century), and =B.= to denote the British Museum MS. Or. 219 (early seventeenth century). In my translation I have, of course, corrected the Lahore text where necessary. While the doubtful passages are few in number, there are, I confess, many places in which a considerable effort is required in order to grasp the author’s meaning and follow his argument. The logic of a Persian Ṣúfí must sometimes appear to European readers curiously illogical. Other obstacles might have been removed by means of annotation, but this expedient, if adopted consistently, would have swollen the volume to a formidable size. Footnote 16: See Ethé’s _Cat. of the Persian MSS. in the India Office Library_, i, col. 970, where other MSS. are mentioned, and Blochet, _Cat. des manuscrits persans de la Bibliothèque Nationale_, i, 261 (No. 401).