BOOK REVIEWS/COMPTES RENDUS 211 Finally, no one can deny that it was legitimate for the “Byzantines” to use the ethnonym ^Rvma”ow and that there is a direct administrative continuity from the ancient kings of Rome down to 1453. However, calling the “Byzantines” “Romans” today may create confusion. For there were other “Romans” in the Middle Ages: the inhabitants of Rome itself, the subjects of the Western Roman emperor (often called thus in the sources), and, finally, the Orthodox Christians in Syria and Palestine (who still use the same designation for themselves, though everybody else calls them “Arab Christians,” which is as incorrect as labeling the Romaioi “Byzantines”). The Romaioi themselves recognized this confusion and in their discussions with the Roman Church about the Union of the Churches they were referred to as the Graikoi (also in Byzantine and non-official texts), and the Catholics as the Romans. Thus, unless English and other languages allow for or invent some kind of differentiation, the use of the terms Romaioi and Romanı́a or “Romean land” seems more legitimate than “Romans,” “Romanland,” or “Roman empire .” University of Cologne Christos Malatras Language and Authority in DE LINGUA LATINA: Varro's Guide to Being Roman. By Diana Spencer. Madison, WI: University of Wisconsin Press. 2019. Pp. 387. Diana Spencer reads Varro’s masterwork on the Latin language as an authoritative handbook-cum-“narrativized memory” (51) on what being “Roman” means for Varro and his readers in the 40s b.c.e. Like the works of Varro himself, this book ranges widely, and the polysemy of its many threads and of the writing itself can be both inspiring and frustrating. But the result is an original explication, which opens up a productive way of reading De lingua Latina and continues the reinvigoration of studia Varroniana over the last decade. Perhaps this book’s signature contribution is the case it makes for the value of what Spencer calls “through-reading.” Approaching De lingua Latina as an “aspiring hypertext” (4), “through-reading” entails reading the extant six books (of an original twenty-five) as an intentionally literary narrative, which activates a radical intratextuality and intertextuality (i.e., with other Latin and Greek texts, as well as all sorts of other material, visual, and oral memes from Roman, Italian, and Mediterranean contexts). In practice, “through-reading” involves construing, and interpreting the significance of, various series of etymologies and other linguistic phenomena, some of which Varro himself explicitly flags in the design of his text, others of which Spencer discerns as larger “arcs” despite their being separated by numerous chapters or books or being embedded in different contexts . One implication of “through-reading” is that it has the potential to destabilize fixity of meaning in any one place. Hence, different etymologies for the same word that other scholars might deem as “disparate,” “contradictory,” or just plain “wrong” turn out, for Spencer, to be somehow complementary, supplementary, or otherwise productive—if only by intentionally provoking readers to reflect on res Latinae and res Romanae. Above all, “through-reading” seeks to unpack the internal logic underlying “Varro’s cumulative and associative discourse enrichment” (55; cf. “cultural poetics,” 6). And “through-reading” tends to be maximalist in its interpretations, teasing out meaning and significance from virtually every phenomenon present within the text, as well as numerous “silences”— variously described as “loud” (e.g., 56), “meaningful,” or “intentional” (e.g., 226). The 212 PHOENIX result is a study that, in explicating the diffuse and sometimes enigmatic treatment of linguistic phenomena found in De lingua Latina, replicates that diffuseness at the same time as it seeks to explicate—often with a fair amount of jargon and with varying degrees of speculation—the possible underlying cultural meaning that, to whatever extent, must have been immanent for ancient readers. This lengthy book comprises an introduction, eight chapters, and a conclusion, with endnotes, bibliography, and an index of subjects (but not, unfortunately, of passages). A prefatory nineteen-page outline (“A Roadmap for a Ruinous Text”) primes readers on De lingua Latina’s structure and scope. De lingua Latina is textually fraught, and Spencer uses Kent’s 1951 Loeb edition with deviations treated in the endnotes; the...