A common methodological assumption is that the number of preserved manuscripts is a reliable indicative of a book’s popularity. Also common is the recourse to levels of popularity in determining the relative importance of each work in the so-called medieval Jewish “philosophical canon”. This paper argues that in the study of Jewish medieval philosophy the quantitative method is misleading. In the medieval world of Islam the double liminality of Jewish philosophers—as Jews, and as philosophers—determined the books they read, those they had in their possession, those they openly cited, and the differences between these categories. The manuscript evidence must therefore be complemented by additional information, gleaned from other sources. Rather than a fixed canon, the end-result should give us a reader-sensitive library: what did people read, who were the readers of particular books, and what books influenced their writings.