Some of the extant manuscripts of Ockham's Dialogus have dates written into them somewhere (some of which may not be reliable), others can be dated only by reference to their handwriting style, the style of their artwork, the way in which pages have been ruled, the treatment of catch words, etc. Since such methods, even in combination, can not give precise dates, and since the period of time in which all these manuscripts must have been written is comparatively short (about 150 years), it seems unlikely that our manuscripts could be put into chronological order with any high degree of assurance.
In any case, even if we could be certain that manuscript X is earlier than manuscript Y, there could be no presumption that X is closer than Y to the original. In some places and at some times reproduction may have been very rapid. The medieval book trade was capable of rapid multiplication of manuscripts to meet peak need. For example, different parts of a manuscript in pieces could be copied by different scribes simultaneously (see Destrez, "La pecia. . ."). The Weimar manuscript seems to be the result of such a process -- different hands have copied different sections. At other times and in other places the rate of reproduction may have been much slower. Thus manuscript X, although it is earlier, could be descended through many copyings, whereas Y could , for all we know, be a direct copy of the original, or at least a copy of an early copy.
For some sample sections we have made thorough collations of all the available manuscripts, in an attempt to discern the relationships among the manuscripts. Although it might be possible to present these relationships as a stemma, it would be a mistake to suppose that this is anything more than a rough guide to the history of the text. In the Middle Ages scribes and users of manuscripts made serious efforts to correct their texts by comparison with other manuscripts and by conjectural emendation. (See, e.g., essays 8 and 11 in M.A. and R.H. Rouse, Authentic Witnesses: Approaches to Medieval Texts and Manuscripts, University of Notre Dame Press.) Unfortunately, they did not record which comparisons they had made or which changes were conjectural. Borrowings from other manuscripts ("contamination") make it impossible to draw up a true stemma. (See M. Bénevot, The Tradition of Manuscripts, Oxford, 1961.) Our manuscript show many signs of medieval editorial activity. Several of our manuscripts, namely Pa, Pe, Mw, and possibly Mz, belonged to Simon de Plumetot (see G. Ouy, "Simon de Plumetot (1371-1443) et sa bibliothèque" in Miscellanea codicologica F. Masai Dicata MCMLXXIX, ed. P. Cockshaw, M.-C. Garrand and P. Jodogne, Gand, Editions Story-Scientia S.P.R.L, vol. 2, pp. 353-81). Simon wrote into the margins of his manuscripts many readings taken from other manuscripts. It is possible that some of the other extant manuscripts are copied from his corrected manuscripts, though we cannot recognise them (such a copy would be difficult to distinguish from a copy of the manuscript from which he took his corrections). A number of other manuscripts, including Fr, Es, Rg, Re, and Na, contain notable marginal corrections. Any attempt at drawing up a stemma will postulate a considerable number of lost manuscripts; what we have are the outer twigs of a fairly large tree.
Even if we could draw up a stemma, it would not be very useful in establishing the text. If it could be shown that a certain manuscript was simply a copy of another extant manuscript, with no readings derived from any other source, it would be justifiable to disregard it. But to show this both MSS would have to be collated extensively, so there would be little economy of effort. Among our manuscripts some pairs are so close that one could be a copy of the other, but it is difficult or impossible to know which is the exemplar and which is the copy. If manuscript X contains a string of words of some length (it is impossible to be precise) found also in most other manuscripts but not in manuscript Y, then we can safely infer that X is not (simply) a copy of Y - since it is not likely that a guess would supply the very words found in the other manuscripts (provided the passage is not a quotation from a well-known source or an obvious implication of the context). Thus it is possible in some cases to show that one member of a pair cannot be a simple copy of the other. But it is never possible to show that they cannot both be copies of a common exemplar not extant.
A stemma constructed purely by collation, and without any external evidence, therefore cannot justify disregard of some of the manuscripts as being mere copies of other extant manuscripts. It might be thought that a stemma can at least justify disregard of the worst manuscripts, that is, those that belong to the most corrupt branch. But even the worst manuscripts may contain some readings genuinely derived from the original. Corruption generally occurs for some reason (e.g. the occurrence of the same word a few lines apart may result in the omission of the intervening text), and the reason may well have operated on several occasions in different lines of transmission. It may be that the correct reading has been lost in many branches of the stemma but preserved in the manuscript that is generally least satisfactory.
The chief use of grouping manuscripts according to resemblances reflecting common ancestry is to make sure that the readings adopted are supported by manuscripts belonging to as many branches of the tradition as possible. As far as Part 1 is concerned, we are lucky (and this is a contingent feature of the set of manuscripts we are dealing with) in that there is a fair consensus among the extant manuscripts for almost all the text we have collated, a consensus which is satisfactory in grammar and sense. Even luckier, there is one group of manuscripts most of which nearly always show the same reading as most of the others, i.e. manuscripts which are almost always part of the consensus. On most of the relatively few occasions when these manuscripts are in the minority, their reading makes better grammar and sense than the most common reading, and it is generally possible to see how the common reading could be a corruption of their reading. It seems reasonable to infer that this group is a consistently accurate witness to the common text lying behind all the extant manuscripts. We regard this as the best group of manuscripts, not because they satisfy any external criterion (such as date or provenance), but simply because they almost always show a satisfactory text found also in most of the other manuscripts. If the first draft of a corrected text is based on these manuscripts, reading the other manuscripts will not much improve it.
It might be suspected that our "best" manuscripts are too good to be true. Perhaps they are descended from some late-medieval "clean-up" of the text, based perhaps partly on comparison of manuscripts and partly on conjecture. (The clean-up must have happened before Simon de Plumetot's death, since Pa contains in the margin some of the readings characteristic of the "best" group.) Perhaps our best manuscripts show a satisfactory text because the medieval editor's notions of sense and grammar correspond with our own. Such an hypothesis cannot be ruled out. From the readings of the extant manuscripts alone, without any external evidence, it is impossible to make any sure inferences about the history of the text. Perhaps our "best" manuscripts do result from a clean-up, or perhaps, on the other hand, they are copies descended with few intermediaries from an ancestor or ancestors that faithfully preserved the original text. Between these possibilities (and there are others) we have no means of choosing. However, since the text of our best manuscripts for Part 1 is generally the text supported by most of the other groups, it would make a little if any practical difference if we were to disregard the "best" group. A text based on the consensus of the other manuscripts would be very close to the text of the "best" group. If that text is the work of a medieval editor, we cannot improve on what he did.
We are assuming that it is indeed an editor's function to re-establish the text as the author intended it to be. Some say that an editor should simply print one manuscript, or all the manuscripts, and not presume to make corrections for which there can be no certainty. But if an editor is not justified in choosing among the variants, the reader is not either. Certainty is not possible, but scepticism is not the only alternative. C.S. Peirce and other philosophers recommended "fallibilism", the assertion of what at the moment seems true, on the understanding that it may not be true and that the assertion may have to be revised. An edition cannot be "authoritative" or "definitive", but it ought to aim at reconstructing what the author wrote or intended to write. The author's intention is the "regulative ideal" shaping and giving meaning to editorial activity, though it is not something we can ever be certain we have attained.
Return to Relations among manuscripts and editions