During his visit to the Sa skya Monastery in 1936, the famous Rgthula Safikrtyhyana photographed a palm-leaf manuscript consisting of altogether 108 folios, which he described as containing an Arthavini~caya-dharmaparyfiya. 1 In fact he noticed a discrepancy between this manuscript and two others of a text with the same title, 2 which he could also examine and photograph (nos. 47 and 87 on his list, both from the Iqor monastery), but he apparently found no time to assess this puzzling difference. Evidently relying on the colophon of the last folio, he was mislead just as a Tibetan reader had been before him, who had written (don) rnam par hes pa'i chos kyi rnam grafts (= Arthavini~cayadharmaparyfiya) in Dbu-med characters on the title page, the syllable don being broken off with part of this leaf.
Indeed, the colophon has arthavini~cayo ndma dharmaparydyah, samdptah., and the size and the script of this folio look so similar to those of the foregoing 107 leaves, that, on first sight, they all seem to belong to one and the same manuscript. 3 On closer examination, however, they have to be separated for two formal reasons. First, the folios are consecutively numbered from 1 to 107, while the last folio bears the page number 10, and second, folios 1 to 107 are divided by two punch holes into three columns of text, while the last leaf has two punch holes as well, but with only the left one dividing all the lines of the text. Accordingly, only the last folio belongs to a manuscript of the Arthavini~cayasfitra, preserving text from mahdpur~alaks..ana 29 (it begins with the word kegatd) 4 until the end of the sfitra (10b5).
All the remaining folios, however, form part of a manuscript of the Samfidhirfijasfitra, and once this is recognized, it becomes possible to read some very faint aksaras in the middle of the first page as "Candrapradipa", the alternate title of the sfitra. Next to the Gilgit manuscript which goes back to the 6th century, 5 it appears to be the second oldest manuscript known so far, since the script can, with a great deal of caution, be dated to the 1 lth century. Judging by the script, the five palm leaves of a Samhdhirfijasfitra manuscript listed by Hara Prasad gfistri and microfilmed by the Nepal-German Manuscript Preservation Project under reel no. A 38/86 are definitely younger, and the same probably holds true for the 25 palm leaves in the Tokyo University Library, 7 which I have not seen. Regrettably, the