Roger Pearse

I’m still looking at the question of whether ancient books had divisions within a book into “chapters” of some sort, and whether they had tables of chapters at the head of each book, and whether the divisions were numbered, and whether the titles in the tables were in the text or not, and whether any of this was authorial.

The Ecclesiastical History of Eusebius was edited in the SC series at an early stage.  Books 1-4 of the HE appears in 1952 in SC31, edited by G. Bardy, in a volume which contains only a brief introduction, and then the text (reprinted from the GCS, 1909) and translation.  It is all a far cry from the sophisticated volumes we know today, but from such little acorns has such a forest of mighty oaks grown. 

The manuscripts of the HE contain tables of contents at the head of each book.  Bardy writes only (p.vii):

In the manuscripts, following the ancient usage, the table of chapters appears at the head of each book.  But in the text, each chapter is prefixed only with a number in sequential order.

Two more volumes contained the remainder of the HE; but by 1960 it was clearly felt that a proper volume of introductory material should have been required, and Canon Bardy was at work on such when he died.  It finally appeared in 1973.  It contained a section on “books and chapters” on p.101-113.  The first ten pages are devoted to the division into books, made by Eusebius himself.  The remainder consists of assertions about chapters rather than useful discussion.  The conclusion is the same as above.

In SC 206 (1974) J. Sirinelli addresses the same question in his edition of the Praeparatio Evangelica (p.52).

The division into books of the PE is by Eusebius himself.  The author refers several times to this division himself.  Very often he mentions that he is coming to the end of a book, or is beginning one.  We are thus assured that the division into books is indeed his work.

It is worth remarking how much better reasoned this is than the vague assertions of Bardy on the same subject.  After remarking that Eusebius himself says that he is ending a book because it has grown too long, rather than for any reason of design, Sirinelli then continues:

As regards the titles of chapters, it is generally admitted that, for the Ecclesiastical History, Eusebius himself divided the books into chapters and composed the titles for them himself.  On the other hand there have been disagreements concerning the Praeparatio Evangelica.  In his 1628 edition Viger reproduced the titles and the summaries of the books.  Valcknaer in his Diatriba de Aristobulo wanted a more rigorus edition created in which the titles and divisions would be suppressed, in which according to himself Eusebius had no part.  Finally Gaisford himself wrote with decided authority “Lemmata, quae in prioribus editionibus non singulis tantum libris sed et librorum capitibus praefixa orationis nexum saepe perturbant, amovi”, and on his own initiative created a new division, which is currently the basis of reference and was followed by Gifford.

Karl Mras, basing himself on an article by J. Bidez [Revue Critique  d’histoire et de littérature, N.S. 61, 1906, p.506; a review of Gifford], sensibly reintroduced these titles and summaries, which in all appearance are the work of Eusebius himself.  In fact in various ways the titles supply us with indispensable information, not given by the text itself.  This is because, reasonably, the author knew that he had furnished these in the title.  For example we may look at chapter 3 of book IV, and chapter 3 of book X.  The title alone contains the reference to the citation which follows.  We place ourselves alongside the opinion of Mras, therefore, and while retaining the division of Gaisford, we give in the appropriate places the titles of the chapters.

This is not a matter of indifference.  The division of Gaisford is arbitrary, and sometimes unfortunate for the sequence of ideas.  On the contrary the division into chapters given by the manuscripts, far from disturbing the flow of the argument, permits us in some cases to restore with more clarity the sequence of thought by Eusebius.  We will have occasion to refer to this again.

A footnote follows to this last sentence:

But with caution; because, for book I, the situation is complicated by divergence between the manuscripts.  One of them, V, reproduces the titles at the head of the chapters in the body of the text.   In the other manuscripts, at least for the first chapters, the text of the title of the chapter appears only in the summary at the head of each book.

I think we may infer from this that the chapter divisions are marked and numbered even in V, but it is a shame that this is not made clearer.  However I suspect all this is derived from Mras.  In SC369, on p.34 we find the remarkable statement:

The chapters indicated in Arabic numerals are those of the Mras edition; reference is always to these.  No modern edition takes account of the ancient division into chapters (with titles) which derives from the Greek manuscripts.

This problem — that the witness of the manuscripts is not published in modern critical texts — renders it very difficult to acquire the necessary information about how ancient texts were divided.

UPDATE: I have found Bidez’ review online.  One remark is interesting in an otherwise not very useful review:

… Gifford was wrong not to place the titles at the head of each chapter.  Sometimes these titles are the only fact we have on the provenance of an extract (e.g. book XV, ch. 17, for a chapter taken entirely from Numenius).

Revue critique d histoire et de littérature / publiée sous la direction de MM. P. Meyer, Ch. Morel, G. Paris, H. Zotenberg
Source: Bibliothèque nationale de France

UPDATE2: I find that I have a copy of volume 1 of Mras’ edition also.On p.viii of the foreword he discusses chapter titles.  I give the German word used to facilitate searching.  The  volume references are to Mras own edition in the GCS.

3.  Eusebius not only prefixed the books with tables of contents (“Inhaltsangaben”), but also intended the headings (“Überschriften”) for the chapters in the manuscripts.  J. Bidez has rightly complained in his review of the Gifford edition that the editors since Gaisford have omitted these headings.  There is hard evidence that these originate with Eusebius:  that the third chapter of the fourth book is from a work of Diogenianus we learn neither from the text, nor the table of contents (“Inhaltsverzeichnis”) of the fourth book, but only from the chapter heading (“Kapitelüberschrift”) (Vol. 1 p. 169, 21); the title of the work of Porphyry — and the number of the book — quoted in the third chapter of the tenth book,  is only given in the chapter heading (Vol. 1, p.561, 12f.); book 11, chapter 30 begins Πάλιν Μωσέως καὶ τούτους; this τούτους is incomprehensible without the preceding chapter heading Περὶ τῶν κατ’ οὐρανὸν φωστήρων; likewise chapter 32 (vol. 2, p.68, 15) Καὶ περὶ τούτου is incomprehensible without the chapter heading Περὶ τῆς ἀλλοιώσεως καὶ μεταβολῆς τοῦ κόσμου; XV 5, 1 (vol. 2, p.355, 17)  πρὸς τοῦ δηλωθέντος — who is meant here we discover only from the chapter heading.  The author cited and his work are listed only in the headings of the chapters or sections in the following cases: IX 14,3 (vol. 1 p. 500, 9f.); X 10 (vol. I p. 591,6): only at the end (p.595, 18) is Ταῦτα μὲν ὁ Ἀφρικανός named (without the title of Africanus’ work, however; the title is missing also in the table of contents of the book); XIV, 7 (vol. 2, p.303, 11f.) : in the table of contents only the name of the author is given *;  XIV, 22 (vol. 2, p.320, 13) gives the name of the work, Philebos (the table of contents of the book says only Ἀπὸ τοῦ Πλάτωνος — Plato); XV 14 (vol. 2 p.378, 17f.): in the table of contents both the name of the author and the work are absent; likewise XV 17 (vol.2, p.381, 9).  As we can see, the more accurate information is in the chapter headings, as is natural; the author first provides for each chapter the appropriate indication of contents; gathering these into tables of contents at the start of the book is then a copyist task.  This explains some small differences (although they are never contradictions).  Of course it is Eusebius who has ordered that these collections should be placed at the head of each book.

* Do not be deceived by the Κεφαλαίων καταγραφή of Gaisford, Dindorf and Gifford; they present a mishmash from the tables of contents and the chapter headings.

There is a lot of solid information in there.  One thing that I do not see, tho, is discussion of whether these symptoms could be accounted for by damage to the inherently fragile tables of contents, rather than by the priority of the material embedded in the text.

Yesterday I translated what the Sources Chretiennes volumes containing books 3, 4 and 5 of Irenaeus Adversus Haereses had to say about the tables of contents (or argumenta) in the manuscripts.  Chapter titles and divisions are also discussed.

Book 1 is covered in SC263, p.30-1.

The manuscripts C and V do not contain the Tabula capitulorum, suggesting that their archetype had lost this item.  This accident, probably caused by the age of the manuscript, but which took place in an era impossible to determine, will astonish none who are aware of the ease with which the first page of an old book  may deteriorate and then be lost.  Thus this does not have the significance that the intentional mutilation of the end of book 5 must have in the family A Q S ε (cf. SC 152, p.30).

But another problem arises in the interior of this second family itself.  Codex A is in disorder when compared to its relatives Q S ε: these begin with the tabula itself, preceded by a title which announces them: incipiunt tabula …, and are followed by the Praefatio of Irenaeus.  The Arundelianus reverses the order; first the Praefatio, then the Tabula.  Which is original?  In all the evidence, from Q S ε as well as from the manuscripts of the other books, the Tabula precedes the Praefatio, and this is so in C V as well.  

Why this reversal in A?

First we may remark that this manuscript — alone of those known to us — begins with the Prologue of Florus (title: Prologus) without any attribution, without the divisions marked by Pitra and Harvey who edited it.  But this page of introduction to the work of Irenaeus could not have disordered what follows, any more than when the copyist of Q, finding it in his exemplar at the Grand Chartreuse (which also had the Prologue of Florus), put it to one side and gave the rest in the exact order.  This is, therefore, an accident particular to A.

The author then uses this information to classify manuscripts, and on p.35 returns to “the capitula in the tabula“, saying that there are 35 and analysing the variants in the manuscripts, which he finds show disorder, which he believes is due to the Greek original.  Then he discusses the insertion at a later date of these into the body of the text.

This location [of the title] is unvarying from one manuscript to another, which should not surprise us because, in general, once a position for a title and its portions in red, and the amount of space to be left, and the large initial, was established in the text, then it doesn’t move an inch from copy to copy to copy, except where the scribe has a positive and pressing reason to change the arrangement in the copy before him.

The translator of the tabula can hardly, in my opinion, have also inserted them in the text.  He translated with discernment, and would not have tolerated the awkward disagreements between many of the titles and their content.  Be that as it may, the capitula were inserted very early, before the separation of the mss. into families, and before the suppression of the final millenarist passage.

Their distribution in the text does not reflect their appropriate place.  The scribe who took the initiative or who was responsible for it — the rubricator — was guided by two principles; to follow the order of the tabula, and to use proper names as a guide to position.  Otherwise, in passages not equipped with proper names, it seemed easy to him to read the text to find the coincidence of language.  In this way he sought to achieve an appropriate division of the text.

The result of this approach is lamentable, as the rubricator has added, to the disorder to the tabula, his own mistakes in placing the titles.  The first six titles have been placed correctly.  From no. 7, where the disorder starts, to no.19, the capitula have been placed by guesswork, and careful observation permits us to see the mistaken logic that resulted in the place of insertion.  Nos.20-32 are fine, apart from two accidents, i.e. the inversion of 32 and 33 and 30 being placed somewhat early, because of the presence of various proper names.  From 32 to the end each insertion is late.

He then tabulates the chapter divisions and says that he is not going to use them in his own text.  He also tells us that the tabula are numbered in C and V, and in A and S.  In Q the tabula are in capitals and unnumbered from 1-12, but thereafter in normal bookhand and with numbers.

SC293 deals with book 2 of Irenaeus, and once again has a lengthy section on tabula and capitula, p.51-69.  It is very welcome to see so much attention paid to these items, so often ignored, and also to the chapter divisions.  Would that all modern editions did likewise!  Much of it is detail of variants which is not of special interest here.

In book 2 all the mss. have a tabula at the head of the book, as they should; a numbered list in C and V, unnumbered in the other mss.[=AQS]  The case of Q with its Greek numbers is peculiar and we will deal with it in a moment. … It is an accident only that the numbers are missing in AQS. …

On p.56 he discusses the Greek numbers in manuscript Q.  I will abbreviate heavily here.

We have left to one side a curious phenomenon which we do not have enough evidence to discuss properly: the Greek numbering in ms. Q.  We will all the same describe them better than has been done so far.  Pitra made the attempt, and Loofs later, after him.  But both were trying to transcribe into the characters of the print-shop some very malformed Greek signs by a hand that Pitra described as “maleferiata”.  The printed outcome  was not very successful.

Were these numbers written by the copyist of the rest of the text, or added later by someone wishing to display his knowledge of Greek?  Because they were written afterwards, and in single session.  But there is no doubt; the writing is that of the copyist.  In fact in the course of book 2 the copyist had to write in Greek those portions of the text left untranslated by the Latin translator (21,39; 22, 177).  However, as far as we can judge, while there was more application in those passages, the same incompetence is visible there also.  The κ, ε, and θ show the same ductus.  We shall not deceive ourselves if we attribute the numbers of the Tabula and the Greek lines of text to the same copyist.

But if so, it is necessary to accept that the [lost] manuscript of the Grand Chartreuse, from which Q was copied, also had this Greek numbering.  Why this ms, and not the others?  Is it handed down, or the result of some philhellenism along the way?  We think the latter, without following Pitra and supposing the intervention of Florus himself.  But we wonder whether this explains the absence of numbering in the Tabula in the family AQS.  Sirmond tells us that the Carthusianus contained the Preface of Florus, and A does also; but A does not reproduce this numbering, while it was reproducing in a secondary line of transmission of which only Q has come down to us.  S leaves them out.  …

It seems that the copyist of Q found in his model some Roman numbers.  In fact he has reproduced in the same column where they align with the Greek numbers the Roman number every 10th number. …

The capitula in the text.

In C and V the capitula are an integral part of the text, copied with their number, without discontinuity between the chapters.  But AQS either have no capitula or, where AQ seem to have them, they are not the original ones. 

The omission of the capitula in A is accidental, ancient, and inexplicable.  A later hand copied them into the margin with their number.  But since the same hand also added in the margin glosses from the Erasmus edition of 1526, these late capitula are based on the [artificial] ones of Erasmus.

Like A, Q bears an adventitious division and numeration.  In the continuous and regular text, a later hand has marked paragraphs in arbitrary places with a large paragraph sign, with a corresponding number in the margin, in large Roman numbers.  There are no titles.  This later division into paragraphs corresponds astonishingly with that of Erasmus, but not perfectly.  …

The late numbering of Q disguise another, sporadic and little remarked, but which seems to be in the hand of the copyist, in the body of the text.  [Some numbers in tiny letters have been written over with new numbers by the late hand].  However this primitive numbering corresponds exactly to that in CV. … [S also has the same numbering as CV, in tiny letters].

Thus there is no doubt.  Since Q and S, mss of the second family, agree with the first family, this is proof that the division of CV is that of the authentic transmission.