Pliny the Elder states, at the end of his preface to his Natural History, that a list of contents of the work follows. In our modern editions this forms book 1 of the work. The indices to each book are also found sometimes at the start of the book to which they relate. The text of the NH was established by German scholars of the 19th century such as Sillig, Jan, Mayhoff and Detlefsen, and the only article on the transmission of these indices that I could find dates from that period. Although the author plainly does not have all the information he needs, and sometimes is less than clear, it is still quite interesting. I have translated a large portion of that article using Google Translate, and it seems useful to give that extract below. I cannot guarantee exact accuracy, note. (In p.711 the technical detail became too dense for my command of German, but also less relevant to my investigation, and I was obliged to omit from there up to the top of p.716.)
Readers may find my previous post on the manuscripts of Pliny the Elder helpful. To summarise what it said, there are the remains of 5 ancient codices, and then a mass of medieval mss. The medieval mss. are divided into two groups, the vetustiores (=older) and recentiores (=younger). I have not yet found a stemma for any of this.
Let us hear what Detlefsen has to say.
* * * * *
[p.701] 38a. The indices of the Naturalis Historia of Pliny.
(See above p. 284).
In connection with the earlier report on Pliny studies (p.284-337), it may be appropriate to mention a special portion of the NH, which has recently and rightly been receiving special attention, the indices. Long neglected by scholarship, and disfigured by incredible corruptions and interpolations, they have for the first time in our own day been published by Sillig in a form that could satisfy the demands of scholarship. How important they are for Pliny studies was demonstrated by Brunn in his useful work, de auctorum indicibus Plinianis, Bonn 1856, which on the one hand, proved from the NH that the principles of the ordering of the lists of authors arose from use intended, and on the other hand sought to determine what sources these were with more precision. Much that has come out about the results obtained so far has not been researched, although because of Jan’s edition and my own, as well as the recension of Urlich (Jahn’s Jahrb. bd. 71, 256 ff), questions raised by Jan (ibid., vol. 95, 858 f), and against Urlich’s recension of the text by Brunn (ibid., vol. 75, 336 ff) have produced better justification and new light on the subject. But work in this area is far from complete.
For us the work of Pliny is a treasure trove of notes of all kinds, of course depending on the sources of very different value, collected “ex exquisitis auctoribus centum” “lectione voluminum circiter II” (Plin. praef. 17); if we count the authors referenced by him, [p.702] the result is actually 146 Roman and 327 foreign authors. The ultimate goal of the necessary investigation must be to document in detail, what Pliny owes to each of these writers, to deconstruct his mosaic-like text into its component pieces, a process which, on the one hand, by looking at extant writers directly, can work out what has been borrowed, and on the other hand, where this is not possible, can only reach a conclusion through careful analysis. In this connection, Brunn’s greatest service was to first make scholarly use of the indices auctorum; but I believe that the principles established by him require some modification.
Firstly, it seems to me that a careful examination of the NH. itself can show joins in the text not considered by Brunn, which may be exploited to uncover the history of the origins of the text, and which likewise may shed new light on the composition of the indices. But any investigation in this direction requires a sound basis of diplomatic corrected texts of the indices; and this correction must not refer solely to the individual words of the indices, but also on the sequence of lemmata. In that area, I confess that in my own edition I often just followed the statements of Sillig and Jan; it was difficult, almost impossible, to accurately give the many strange and rarely occurring names, particularly of plants, before the same were corrected in the text of the books from the relevant manuscripts. However, I have tried to establish the order of the lemmata from the manuscripts more accurately than my predecessors, and you will often perceive deviations from the latter in my edition. For they had arbitrarily ordered some lemmata so that they exactly corresponded to the order in the text of the relevant books, while the manuscripts gave an entirely different order. In such cases, I have definitely followed the latter, although in some cases I do not know whether carelessness by the copyists, or by Pliny himself, or some other cause should be blamed for the inconsistency. A repeated examination of the manuscript sources led me to ask the following questions.
For the last book of the NH., a special investigation was required to get clarity about its text tradition, different from that of the remainder of the book (see Jahn’s Jahrb. 95, 77 ff). Likewise the data on the tradition of the first book offers special difficulties. Just like the last book, [p.703] it is either not found in most of the older manuscripts, or only partially. It is well-known that the first and last leaves of manuscripts are particularly exposed to the rigours of time, but there is yet a further problem, that critically editing book I is one of the most difficult parts of the task of editing the NH.
However I now have far richer materials than before, provided by Sillig’s edition, and although it is now possible for me to determine the basic features of the transmission, I do not conceal from myself that further study on the manuscript tradition is desirable. From Sillig’s preface, and from his notes on book I, an insight into the outline of the tradition can only be obtained with effort. In many places he obscured or misreported the truth, and only after lengthy collating of the manuscripts have I succeeded in discovering the key points in this part of the NH to which attention should be directed. But very often the opportunity was lacking, and there was not enough time to examine the manuscripts a second time more precisely. However I believe that the most important thing is to make known the preliminary results of my work, because the details are of some importance for the understanding of the transmission of the text and for the correction of it.
Sillig states in his first note on book 1 that, until the Hardouin edition of 1685, the indices as printed were simply erroneous, interpolated and spurious; and that Hardouin’s edition also suffered from careless distortions, which have caused confusion ever since. He himself says, “Hinc neque Dalecampii neque Harduini vel Broterii editionibus in hoc indice respectis ego editionem eius, si ita dicere licet, principem feci, cuius haec ratio fuit, ut singulorum librorum indices e codicibus, qui eos continent et quorum sigla cuique libro apposui, ederem.”
From this anyone without a personal knowledge of the manuscripts of Pliny would suppose that the problem is fixed. It cannot be denied that Sillig’s edition of the indices marks a very significant degree of progress over all previous editions. But as remarked, he has overlooked some serious problems. He further states: “In universum vero tenendum est in aliis codd. (Tbd = cod. Tolet. e d of my edition) hunc indicem suo loco legi et proinde primum naturalis historiae esse, in aliis (BVAp; the latter = O of my edition) ipsis libris, in aliis (Ra = RE in my ed.) et hoc loco omnibus libris et rursus singulis praemitti, ut bis in iis exstet, unde suspicio oritur paulo post Plinium exstitisse librarium, qui in lectorum suorum commodum indicem Plinianum [p.704] bis scriberet, semel suo loco, tunc in initio singulorum librorum.”
The three classes of manuscripts, which Sillig puts forward here are correctly distinguished, but not all the manuscripts that he cites have been classified correctly, as we shall soon see.
This study will concentrate on those manuscripts which I, in my edition, and in the article on p.284 f., on the mss. of Pliny, have named the “younger” manuscripts. The main manuscripts of this group are the most complete that have come down to us, and the details of the transmission of the indices can be seen most clearly in them. There is no doubt that in their archetype, X‘, the indices were not only united into a book which preceded the others, but also that each individual book had its respective index placed at the front of it. This is what we find in mss. RF, and substantially in E, where book I is at least partially preserved, and we may presume that the same was originally true in D+V+G, where book I is now absent while the respective indices stand before the individual books, and the books themselves are numbered from 2-37. (I should add that what I will show here is occasionally in conflict with my statement above, p.288 ff, that F is a direct copy of the single manuscript D+V+G, now divided into three pieces. I have now received full collations of F and V from Leiden University, thanks to the kindness of Prof. Pluygers and Dr. du Rieu). Some significant exceptions should be noted, however.
In EF, the index for book II is not repeated at the start of the book, and no doubt it will have been the same in D and the other cited mss., where the start of book II is not preserved. The omission of this was no doubt because it had seemed superfluous to the scribe, who had already copied this index at the beginning of book I, to repeat it again before he had written even a word from the actual text of the NH. No doubt in the archetypus X‘ the same situation was to be seen. But the index to book III is also missing in ED and its copy F, perhaps because here the repetition still seemed superfluous; whether it is absent in R, I have not noted, and Sillig gives no information on this. R agrees with ED, so this omission must have been found in X‘. Of the other peculiar omissions, in E, we will talk after we have discussed the other two classes of manuscript given by Sillig.
In both families, in the one in whose manuscripts the indices are purely collected in book I, as well as the one whose members [p.705] completely omitted book I and where the indices are given only in the individual books, are, compared to the other younger codices, only relatively later. To the scribes of these mss., it seemed a pointless effort to write each index twice, and they therefore soon left out the duplicate, and soon the first book.
The indices are only contained in book I, in the following descendants of E (see above p 299 ff.): Vat. 1954, Borbon. V. A. I and V. A. 2, Angelicus or Passionaeus, Paris. 6798, 6800, 6802, 6803, Taurin. CDLXv/vi, Luxemburg.; — in the following, probably associated with the archetype X3 (see p 303): Borbon. V. A. 4 and Leopoldo-Laurent. CLXV; — in the following derived from F (or D+G+V) (see p. 289 ff.): Tolet., Paris. 6797, 6799, Vat. 1953, Vindob. CCXXXV; — in those probably related to R: Vat.-Palat. 1559 (see p. 295); — finally in the following manuscripts of uncertain lineage: Vat. 1951, 1952, 1955, 1956/7, 3533, Vat.-Ottob. 1593/4, Chigianus, Barberin. 758, 2303, Laurent.-S. Crucis XX sin. 1, Parmensis H. H. 1. 62, Ambros. E. 24 inf.
By contrast, in the following the indices are found only in the individual books: in the closely related to E: Vindob. a (= w in Sillig), which we will discuss in more detail below, and Vat.-Urb. 245 (see p.302 f.); — in the offspring of F (see p. 289 ff.): Laurent.-Slaglosianus, Laurent. LXXXII, 3 und 4; — and in the manuscripts of uncertain origin: Vat. 1950, Barberin. 2503, Borbon. VA. 3, Marcianus CCLXVI,
Pollingensis (see Sillig’s praef. p. XXI, Jan, observ. p. 11).
From this list we learn that the copyists already, from saec. XII onwards, regularly omitted one or the other copy of the indices which were duplicated in the older codices. Certainly by far the majority of the younger manuscripts not mentioned belong to one of these two classes. The youngest manuscript that I have recorded which provides the indices in both places is F, the eleventh century copy of D + G + V.
Let us return to the study of codex E by comparing it with the transmission of its nearest relative, Vindob. a. We have already remarked that in E, book I was present; but the first and last leaves have been ruined with damp, are very old, and only partially readable. Sillig makes the same observation in his editorial note.
Instead let us look at Paris. 6796 A (e in my edition), the oldest direct copy of E. In this, the duplicate indices are missing from before books II and III, as already stated. That before book IV is present, those before books V and VI are missing again, the one before book VII is present, although [p.706] the first half has fallen out together with the close of book VI, and been replaced by parts of the second volume added from a codex belonging to one of the vetustiores manuscript group (see. p.298). The indices are again missing before books VIII-XIII, that before book XIV is present, they are missing before books XV-XIX, and there is none before book XX either, but it is present before book XXI, and is there introduced with the following words: Libro XXI continentur; but it is not the index which belongs to this book, but rather the one for book XX. The beginning of book XXII is now missing in E, and with it most likely its index. That before book XXIII is present. That before book XXIV is missing together with the start of the book. Finally all are present for books XXV-XXXII, and the remaining books, whether they had indices or not, are no longer present in the manuscript. (Sillig’s description of the codex is so very poor). Apparently the presence of absence of an index before the individual books of E is random. This naturally leads to the supposition that the original intention of he who laid out the manuscript was to write the indices only in book I, but that some of the copyists who succeeded him in the work added, here and there, an index to individual books because they found it in the original; so that for book XXII would not have been present, as originally intended. Of course the arrival of the index for book XXI [in front of book XXII] must have happened in some other way.
Let us compare this with the tradition in codex a, the next nearest relation to E, which likewise is descended from archetype X3, but is not descended from E itself (see p.303, and Rhein. Mus. 15, 380 ff.). From this relationship it follows that peculiarities in the tradition of the indices which are found in both must be attributed to that archetype. One such feature is that, although the scribe of a completely omitted the indices in book I, he still often numbered the other books from 1-36, although not consistently. Likewise in a before book XXI is found the index corresponding to book XX. This is also found in E, but is then followed by a second index, really belonging to book XXI, but not the same one found in book I, but another, probably compiled by the copyist himself, or copied here from elsewhere. It seems to be compiled from the text, like the chapter titles found in many other manuscripts which do not agree in wording with the lemmata of the genuine indices. The peculiar displacement of the index of book XX must certainly be traced back to the archetype, X3. Moreover a has the correct index before every book, except that it has the index to book XX twice and that of book XXI not at all in the manuscript. Furthermore, for books XXII, XXVII, XXX, and XXXI, [p.707] after the genuine indices there are shorter items of the same type, like that for book XXI, and finally before book XXXII only the front half of the genuine index is present, followed by a shorter one of the same type, at the end of which is placed the list of authors used for the book, absent from the other shorter indices.
This relationship of E and a to one another, and to the other manuscripts and to the editions, is instructive in many ways for the history of the transmission of the indices. From it we may conclude the following.
The matching testimony of all the older manuscripts of the younger group shows, that in their common archetype, X1, the indices were found both as book I, and before the individual books, except before book II and probably book III. This arrangement persisted in archetype X2, a descendant of X1 and its older descendants D+G+V and R.
In contrast, the second descendant of X1, archetypus X3, from which E and a are descended, was different. Book I remained in its place, but already a number of the indices before individual books had perhaps been omitted, and the index for book XX was placed before book XXI. More of these indices fell out in E.
In contrast, the writer of a followed a different principle, left out book I, and placed indices once more before the individual books, although he was unable to provide the proper index before b. XXI. For this purpose he placed new indices before some books, and these are probably the ones which then gradually displaced the real ones, and finally until Hardouin’s edition even took their place in book I. Unfortunately, I can find no manuscripts other than a in which they may be found. More information about such, and in particular the tradition of the indices in the manuscripts more closely related to E and a (see p. 298 ff.) would certainly allow a clearer insight into these complex relationships.
Let us now consider the tradition in the manuscripts belonging to the older group (see p. 306 ff.).
Very little can be said with certainty or probability, because from all of them we have only fragments, but none of them contain book I of the NH. That Pliny himself prefixed the indices as book I to the following books is certain from his own words in the dedicatory letter to Titus 33, and the running count of the other books from II-XXXVII in all the older manuscripts demonstrates that this arrangement is authentic. One might therefore consider it certain that likewise in AMOB, the fragments of which have a similar numeration, that book I was originally present. [p.708] This is probably correct for the others, but for A it can be shown with certainty that book I was always missing.
Codex A is composed of thirty sequentially ordered pages:
(“quaternionzeichen = “quaternion numbers”; “gezeichnet” = “labelled”)
The quaternion marks, which are written in the original hand, show that only two quaternions are missing from the beginning of the manuscript. If we calculate what these contained, based on what survives, it follows that the manuscript must have begun with book 2, and did not contain book 1, nor probably even the letter to Titus. By contrast the index for book 2 probably was present, just like the indices for books 3-5, and the same arrangement can be assumed for the following, now missing, books. Can this arrangement of the codex be attributed to the emendator, Junius Laurentius? Note that we cannot suppose that this led to the similar arrangement in codex a and the other younger mss, since there is no other trace of a connection between the two.
In the following manuscripts of the older group, other than A, indices are preserved before the individual books. In M they are preserved in fragmentary form before books XI-XV. In O (p in Sillig), the same is true before book XXXIV. In B they are preserved before the last six books.
Finally, it seems that some of the variants in book I, and in the indices before individual books, derive, as already noted, from a second hand in the [p.709] codex underlying ERD+G+VF (see p.306 f.), and in E the first half of the index to book VII comes entirely from this source. In respect of the tradition of the indices in book I, however, we remain mostly in the dark for this large group of manuscripts.
So we have determined the scope and nature of the sources from which the text of the indices is to be taken. This leaves the difficult question of how to use them, what their value is, compared to each other. In this area Sillig and von Jan have gone completely in the wrong direction. The latter says only in his praefatio to vol. 1, p. iv: “In emendando libro I, qui continet ceterorum indicem, ut Silligius nihil fere recepi quin inveniretur in exemplaribus manuscriptis, sed e codicibus qui hunc librum continent ille magis secutus est Riccardianum (R) et Parisiensem primum (E), ego Toletanum et Parisiensem secundum (d), ut qui magis consentirent cum ipsis ceterorum librorum verbis.” Both lack insight based on an in-depth investigation into the relationship of these manuscripts to one another and to the other sources. The questions to be asked are as follows: if the indices are found twice in the base manuscripts of the younger group, what is the relationship between the two transmissions? Are there interpolations or lacunae in one or the other of the divergent traditions? And how do they relate to the text of the older group? For the sake of brevity, I shall refer in what follows, when discussing the text of the indices which are placed before the individual books, using the bracketed siglum of the codex. E.g. I shall use R to refer to the text of the index in book I, and (R) for the text of the index repeated at the start of the relevant book.
First let us identify some key features of the text that will make it possible for us to learn about these questions. The most important of them are the following. The lemmata at the conclusion of book XXXIII from 52 quando repositoriis on, together with the index auctorum missing in one tradition; instead of these are the lemmata of book XXXIII §.19 nobilitates ex aere, together with the index for this book, repeated, except for the words on p.65, 33 of my edition, CCLVII. Ex iis ad canis morsuus until p. 66, 1 contra lymphationes. Summa: res. This is found in ERF (see Sillig’s praef. p. xiii; ind. 33 n. 53; 34 n. 15) and so it was, no doubt, in D, the original of F; the defect must also have been present in the archetype X‘. (What Sillig gives in the index auctorum on book XXXIII for FR is only a demonstration of his carelessness (see especially n. 71); it refers entirely to the wrongly repeated index for book XXXIV, [p.710] and in this area is the cause of a significant interpolation in index XXXIII, which we will discuss below). From Sillig’s notes we learn further, that this alteration is not found in (B) (V) d and Toletanus; nor is it found in (F), the copy of (V), nor in (a). We cannot say anything about the situation in (E), because the manuscript today ends at book XXXII, 135; but an investigation of manuscripts derived from it (see p.299 ff.) would perhaps permit conclusions about the original tradition. In (R), however, the situation is as in (B)(V), although Sillig says nothing about it, as this source is first used from the index to book XXXIV (see Sillig ind. 34 n. 15).
— So this change is found only in the text of book I in the younger group of manuscripts, except not in B in the index before book XXXIII. The origin of this seems to be explicable in the following way. The lemma at book XXXIII, 52, quando lances immodicae factae, must have been at the end of a leaf or a page in an archetype prior to X‘. The copyist of the latter must have skipped a leaf or a page, and continued with the next leaf which began with nobilitates ex aere, and continued happily until the directory of authors at the end of book XXXIV. Since he believed that he had copied the index of book XXXIII, he began with the beginning of index 34, and carried on, without ever detecting and correcting the error made earlier, so that the latter half of index 33 is missing, and that of index 34 is duplicate. The size of the supposed leaf or page is 34 lines of Jan’s edition, so it is difficult to draw any other conclusion. It remains only to suppose that the origin of the error was in the same codex in which the materials in books II-V were first switched (see p.288 and Rhein. Mus. 15, p.369), however the evidence for this is difficult to produce, because the indices were probably written intermittently and at longer intervals than the rest of the text, so this and that must be calculated according to different methods.
It should be noted that the change is not found in D and Toletanus (see Sillig’s notes), nor from my information in Vindob. CCXXXV. These three manuscripts derive, as I think I have probably said (see p. 289 ff.), from F (and I would very much like to know whether they too have the features of F described there), so we must assume that the copyist, who took the deviant arrangement from F, only gave the text of the indices in book I (see above), either compiled from those standing at the front of the individual books, [p.711] related to the originals, or else at least corrected the error in book I from the existing index before book XXXIII. We will refer below to other similar cases, from which we may obtain some clarity.
Now we must discuss the interpolation of the index auctorum of book XXXIII mentioned earlier. In book I of our best manuscripts, as noted, it is not present, so we are dependent on the tradition of (B)(V)(a) alone, and maybe also d Toletanus; because Sillig’s citation of RE has already been shown to be incorrect.
[snip technical details pp.711-716]
Let us briefly put together the general results of this investigation. It has been found that the principal basis for the text of the indices is found in the manuscripts of the younger group, which are only present in the older group in rare cases. In the archetype of these, as well as in book I, every index, with the exception of book II and sometimes even book III, was repeated in front of the book to which it related. This arrangement must be very old, because individual errors in their archetype, X1, show that they were present in an even older manuscript, or possibly even be derived from Pliny himself and his publication of the NH. The value of both texts is about equal; they complement and correct each other. Little can be said here about the manuscripts of the older group. Even the copyist of codex A omitted book I completely. In this group the indices before b. XXXIV and XXXVI have been interpolated, and that interpolation is already present in book I in the oldest archetype of the younger group. In the archetype X3, derived from X1, the arrangement was changed so that the individual indices before each book were omitted. This approach was followed both in E, which was descended from X3, and in many more recent mss, while in a, and some others, book I was omitted instead, as in A. To these belong d and Toletanus, whose text of book I, from combining the duplicate texts of their originals, …. Their value is, compared to other sources, only secondary and often doubtful.