I’ve always been interested in the question of when chapter divisions and chapter titles arrived in ancient books. Various articles on the subject have passed through my hands in recent days as I converted photocopies to PDF’s, and again I found them interesting. But in those days the German sources, Birt and Bergk, were inaccessible to me, being large books in large libraries, not to be borrowed and scarcely to be photocopied. I wonder if they have made it onto the web?
Quite by chance I found this material online in V. H. Stanton’s The Gospels as Historical Documents, Cambridge, 1903, p.22f (here).
THE FORM OF ANCIENT BOOKS AS AFFECTING HABITS OF QUOTATION.
I. The only kind of division of the subject-matter which was ever common in Greek and Roman Literature even to the sixth century A.D. was “the book,” in the sense of a portion of a larger work. The book in this sense, as the names for it in Greek and Latin (bi/bloj and bi/blion, volumen, also later and more rarely to/moj) imply, corresponded originally and normally with the contents of a roll. (See Birt, Antike Buckwesen esp. chh. 3, 5 and 7, comparing Bergk, Griechische Literaturgeschichte I. p. 226 f.) For the most part works which could be comprised within a roll of moderate proportions — as for example most of Plato’s Dialogues and even the longer writings of the New Testament could be — had no divisions, and larger works no lesser ones.
Only in the case of works of a few authors do we hear of chapters or headings (kefa/laia, capita, also called ti/tloi) which served to break up the text into portions. The scholiasts and commentators upon Aristotle speak of such in his treatises. In the main this evidence belongs to the third and following centuries A.D.; but the divisions in question may, at least in some instances, have been early introduced and traditionally preserved.
Yet they do not seem to have been employed in all his works. The Constitution of Athens, in the recently recovered papyrus MS. of it, is without them (see Kenyon’s ed. p. xviii.). Moreover, so far as I have observed, the scholiasts and commentators themselves, though they mention chapters when discussing the question how a treatise should be analysed, rarely refer to statements, opinions or words as contained in such and such a chapter. Commonly they give only the philosopher’s name, or the treatise, or book of the treatise, with an indication sometimes that the passage will be found near the beginning, or the end, of a treatise, or book. In writers earlier than the fourth century A.D. this vague mode of reference is, I believe, universal.
Moreover, the works other than those of Aristotle, which were divided into chapters, seem to have been chiefly those which consisted of a series of articles, such as collections of marvellous stories, books on Natural History and Botany, medical, and probably also legal, books. Clement of Alexandria (circ. A.D. 200) also seems to have divided his Miscellanies into chapters. “Let this second Miscellany,” he writes at the close of the second book, “here terminate on account of the length and number of the chapters.”
The only instance of a reference to a numbered chapter appears to be that in Cassiodorus (Lib. Lit. ch. 3, Migne, voL LXX. col. 1204) to “the ninth chapter of the first book of the Antiquities of Josephus.” These numbers may have been inserted in the Latin translation which Cassiodorus himself caused to be made (Div. Lit. ch. 17, Migne, ib. col. 1133). [For the instances given, see Bergk, ib. p. 233, Birt, ib. p. 157.
To the examples of works with headings quoted by these writers, Dioscorides on Plants and Roots may be added, see Palaeographical Society's Publications, I. plate 177. On the other hand, they are both, I believe, in error when they state that Symmachus' copy of Seneca had chapters. The reference to Seneca by Migne (ap. Symm. Ep. x. 27), or some other editor, introduced within a bracket, has, it would seem, been mistaken for part of Symmachus' text. Of the employment of any subdivisions of chapters there is no trace whatever. The word tmh~ma (section) is indeed used, but only as an equivalent for kefa/laion].