The examination of oral and written practices of knowledge transmission during the first four centuries of Islam has proved extremely productive (Sezgin 1967-84; Schoeler 2006). A subset of this field, the reconstruction of lost works, has led scholars in various disciplines to substantially revise the early history of religious movements or redate the advent of literary genres (inter alia: H. Modarressi 2003, E. Landau-Tasseron 2004, D. Stewart 2008), and allowed others to demonstrate the significant influence of previous works on the organization of major literary or exegetical works (Sezgin 1956, Fleischhammer 1974, S. Günther 2009). The present paper argues that the positive identification of biographical works written by and devoted to Sufis from the first half of the 4th/10th century necessitates a reconsideration of the early written systematization of Sufism, with particular reference to the Baghdad school of Sufism. Major studies on the systematization of Sufi biography date the advent of the tradition to the early 5th/11th century, based primarily upon extant prosopographical sources (“group biographies”) such as al-Sulamī’s Ṭabaqāt and al-Iṣfahānī’s Ḥilyat al-awliyāʾ (henceforth HA; Cooperson 2000, Mojaddedi 2001). Contrary to this characterization, both Massignon (1975 [1922]) and Sezgin have identified no fewer than seven lost prosopographical works of Sufi biography composed during the first half of the 4th/10th century, mostly by Baghdad-based Sufis; these findings have been restated by Böwering (1980), Knysh (2000), and Karamustafa (2007). Mojaddedi, conversely, has stated these works are misidentified and should instead be described as “notebooks” (2001, p. 56-61). Until the present, no study has attempted to determine to what extent these works have been preserved and can be reconstructed. To test Mojaddedi’s claims about the works previously identified as Sufi prosopographies, I have compiled the notations of transmission between al-Iṣfahānī and his immediate informants from the final three volumes of HA, the largest extant work of Sufi prosopography. The dataset comprises 4,615 accounts from 291 biographical notices. Parsing these data in a non-relational database reveals consistent patterns of usage indicating both the persistent citation of substantial literary works which can be considered books, and that the organization of HA was heavily dependent on them. Consequently, the advent of the written biographical tradition of Sufism can be dated 50-70 years earlier than previously argued, and it is hypothesized that the reconstruction of these works will reveal substantial differences in emphasis between the earlier and later traditions.