As divided by the manuscripts, the extant letter corpus comprises thirty-six books, or a little more than that: one book and evidently part of a second consisting of letters to Marcus Brutus, three books of letters to Quintus, sixteen of letters to Atticus, and sixteen of letters to various persons which acquired the collective designation Epistulae ad Familiares in the sixteenth century. But citations in earlier sources suggest that these sixteen books did not originally circulate as a set, but were separately identified by the lead correspondent in each book (for example, “liber epistularum M. Ciceronis ad L. Plancum,” Gell. NA 1.22.19, “liber M. Tullii epistularum ad Servium Sulpicium,” Gell. NA 12.13.21). See Peter (1901, 54–57) and Büchner (1939, 1195).
There is also evidence for several books of letters that were published but have perished. They are attested by about one hundred citations in mostly late classical sources, especially Nonius Marcellus. (The citations are conveniently gathered in Watt [1958, 152–75].) Since the sources sometimes specify the number of the book from which a quote is taken, it is possible to estimate the minimum number of books that are missing. Thus there appear to have been a total of at least thirty-eight: two books each containing letters to Axius, Cornelius Nepos, Licinius Calvus, and Marcus junior; at least three each to Caesar, Octavian, and Pansa; four to Pompey; nine to Hirtius; and eight to Marcus Brutus (in addition to the book that is extant), for a total greater than the 36 books that have survived. For details, see Büchner (1939, 1199– 1206) and Weyssenhoff (1966 and 1970). However, I do not subscribe to Weyssenhoff ‘s hypothesis that letters attested only as fragments never had an existence as circulating published texts. She believes that either Tiro, or a grammarian who had access to Cicero’s unpublished letter files in the century after his death, borrowed from them in writings of his own, and that all later testimonia derive from these excerpts.