The Ming dynasty Translation Office (Siyiguan) prepared sample letters and vocabularies for studying the languages of its neighbours, resulting a family of texts known as the Huayi yiyu. The Huihui (Muslim), i.e. Persian, component of these translation manuals has received less attention than other sections; e.g., the Gaochang Uighur letters edited by Ligeti, and the works of Lewicki and Haenisch on the Mongolian. One reason for this is that the “Persian” letters are in fact nothing of the sort; they are inauthentic, and written in a garbled Persian, probably the result of translating word-for-word from Chinese sources. Nevertheless the lexical material they contain is not without interest. Liu Yingsheng’s work is a study of three Persian-Chinese glossaries produced within the Muslim Bureau (Huihuiguan) of the Translation Office. The first two texts are zazi “miscellaneous words.” The author began his research on the Beijing Library copy of the Siyiguan edition of the zazi (the longer of two editions), and later studied copies in Europe and Japan. Most textual comparisons are with the Bibliothèque Nationale copy, which is reproduced in facsimile, while the author’s Persian transcriptions and Latin transliterations are based on the Beijing Library copy. (The results can be slightly confusing.) Among the rather bland vocabulary items there are some noteworthy entries: e.g., the ethnonym turki translated into Chinese as “Gaochang” — a rare piece of evidence for the “ethnic” identity of the Gaochang Uighurs. The second zazi text is a supplementary list of words, many of which occur in the “Persian” correspondence, the Huihuiguan laiwen (not included in Liu’s study, for the reasons given above). This text occurs only in the Berlin Staatsbibliothek edition of the Huayiyiyu, otherwise known as the “Hirth Polyglot.” The third text is an early Qing copy of a Huihuiguan yiyu found in the Beijing library. Based on a phags-pa seal which the copyist drew by hand, the work is thought to have been copied from a Yüan glossary once owned by an official of the northern Yüan, and hence pre-date the Ming Translation Office in its composition. Its structure and contents suggest that it was a source for the shorter edition of the Ming zazi. The term yiyu, as the author discusses in his introduction, refers to a style of glossary in which words are given in Chinese transcription, creating greater difficulties in establishing the correct readings. For the convenience of subsequent researches, the author concludes by listing words which he has identified in advance of the earlier work of Honda Minobu (“‘Kaikaikan yakugo’ ni tsuite,” Hokkaido daigaku bungakubu kiyo 11 : 222-150), words which are still in doubt, and a combined lexicon of all three texts. Liu’s introduction discusses the status and study of Persian in China during the Yüan and Ming dynasties, and surveys what is known of the textual history and extant editions of these texts.