Contamination?

A Codex Stemming from a Single Tradition or

A Copy Based on More Than One Exemplar?

A Study of Codex Salmanca, BU, Bibl. General Historica 2322

For this study of the possible contamination of Salamanca BU 2322 (S2322), we have taken advantage of the fact that a substantial part of book three appears twice, at the beginning and end of this witness to Richard Rufus of Cornwall’s Dissertatio in Metaphysicam Aristotelis.  We wanted to see whether there was any evidence that

either the original scribe or a subsequent corrector had corrected the first copy on fol. 72r-75r (S) by reference to a different manuscript.  Does S2322 witness a single tradition or was its witness contaminated by reference to another text tradition?

The second copy on fol. 130r-132r (Z), unlike the rest of the manuscript, shows no sign that anyone checked the text after it was copied making corrections. Probably, this is because the person supervising the preparation of the text noticed that it was a duplicate, writing on the top of 130r: “Ista duo folia quae sequuntur cum media plana sunt super II Metaphysicorum et ad litteram posita sunt in principio operis et sic superfluunt."

For the purposes of our study, we first eliminated variants from a single manuscript other than S and Z and also individual indications of correction unrelated to any other codex.

We then divided the remaining variants according to their significance for the question whether S2322 was copied from a single manuscript.  Listed separately are variants in seven different categories, beginning with the variants that suggest that S and Z have the same source and closing with those that might indicate corruption.  The variants are also displayed here graphically in the context of the complete text -- both with and without insignificant variants described below in Category VII.

The variants in the first four categories are shared by both S and Z.

            Category I (green) includes cases where S and Z have a shared error.

            Category II (blue) is for cases where S and Z share the original reading

                                    but it  is not witnessed by any other manuscript.

            Category III (yellow) contains readings that SZ share with another manuscript.

            Category IV (orange) covers passages where SZ share an error with another witness.

Categories V and VI contain passages where S and Z differ from each other.

            Category V (violet) is for individual reading in S or Z, not shared by another codex. 

            Category VI (red) indicates a different reading of S or Z, shared by another codex.

Category VII includes variants that we think should probably be ignored because they are not significant for our study (grey).

            We have listed separately the following variants:

1.         ergo/igitur, because scribes use these words interchangeably.

2.         iste/ille etc., because they are often used interchangeably and the abbreviation         marks used by scribes are ambiguous.

3.         a missing 'etc.', since scribes seldom regard them as significant.

4.         missing contraction signs and missing abbreviation marks for missing -m or -n, -ur,            since these are signs of scribal carelessness rather than differences in the exemplar.

For our study the variants of category VI (9 variants) are of special importance. For, if S shares an error with another manuscript, this might mean that S was corrected from another manuscript  addition to the primary exemplar from which S2322 was copied. 

S2333 might have borrowed a reading from another tradition, either the tradition from which codex N stems or V’s tradition.  And, since it is possible, though not likely, that Z had been corrected, we included such readings in Z as well.

Of special importance are significant errors that S or Z share with N or V that are not likely to be made independently from a model. If, however, we look at the variants in category VI, we see that there are many fewer variants (9) in this category than the others.  Most of them come from Z (6) where we have no particular reason to suspect corruption, and most errors seem insignificant.  Most are transpositions (3) or omissions (3) that could easily have arisen independently. The two alternate readings look like cases where perhaps a stroke had been inadvertently omitted – namely, “numeralibus” for “numerablibus” and “ponit” for “proponit.”  One might have been important, namely the `signficant’ rather than `significent’ that Z shares with N.  A single anomaly, however, can indicate only coincidence.

   Categories II (122 variants) and III (26 variants) don't indicate contamination. They are cases where S and Z share a correct reading either only with each other or with another manuscript. The correct reading comes in all likelihood from the original exemplar and not come from another tradition.

   Category I (246 variants), which contains passages where SZ share a mistake not found in other manuscripts, strongly suggests that SZ were copies from the same model with the same mistakes. The fact that this category contains the most variants is an indication that they were both copied and corrected from the same exemplar.

   Category IV (45 variants) indicates that SZ were copied from a model in the same tradition as another manuscript.

   Category V (151 variants) contains individual reading of S or Z. Because they are individual readings and are not shared by other manuscripts, they do not indicate corruption.

   As with all stemmata studies, we cannot exclude the possibility that correct readings were arrived at independently either from another manuscript or by an intelligent copyist.  Suspicion is most likely to arise where the manuscript has been corrected – as when a reading appears above the line or in the margin.  In our study, there were five instances of corrections, three marginal corrections (mg.), and no supralinear corrections.

            There are 5 instances where the copist of S or Z corrected a word into another word. These corrections can easily be explained by assuming that the copist made (or thought he made) a small mistake while copying, noticed it and made the correction. Such corrections need not indicate corruption. Since, for example, `duabus’ modifies `rationibus’, it could easily have been corrected without reference to another witness.

721 si nihil]  vel forsan simul V, simul (in corr. S) SZ^y

776 Contingit] Continget XN S, Continget in corr. Z^y

773 quod in corr. S, quod superscript r or cut l Z^r

781 enim] in in corr. S^r

787 Quorum in corr. Z, add. sed exp. una Z, add. ult. forma Z^r

1134 duabus corr. ex: duobus S, duobus Z^r

            Marginal corrections in S are as follows:

95 secundae lac. SZ^g, add. (mg.) secunda S^r

286 moverunt lac. S, ma/7un/1t mg. S, manerent (ma/7u/1nt) Z^g

880 ipsum] numerum Z, lac. S, add. (mg.) num (?) S^g

            In  second and third cases S has a lacuna in the text and a reading is supplied in the margin. This can be explained by assuming that the scribe of S was unsure about the reading in the exemplar, and later he or the corrector put it in the margin. Because the words in the margin of S are the same as the text in Z, these marginalia do not indicate contamination. Rather, since both S and Z have the same mistake, these cases again indicate that S and Z used the same model, since they were clearly both uncertain what to do and would probably have supplied another reading if a different exemplar had been available.

            In the first case the copists of both S and Z inserted a lacuna. Only the copist of S wrote a word in the margin. There is no evidence that he got the word secunda from a second model. The word `secunda’ is a mistake that might have been in S and Z’s exemplar, but it was so difficult to read that both the S and the Z scribe had difficulty with it.

            So there is no evidence for corruption in the marginal corrections of S. Like the uncorrected text, the marginal readings seem to indicate that S only used one exemplar. For, assuming that the scribe of S used two models, why would he write manerent for moverunt and numerum for ipsum, if he had a second manuscript, when the latter readings are clearly preferable? 

            In addition to variants, there are other considerations that suggest a shared exemplar.  One of these is odd shared abbreviations or spellings, such as the following:

200 igitur] Gⁱ  SZ

   Conclusion: The shared errors of SZ, demonstrate that S and Z were copied from the same model. There are only a few passages where S or Z agree in error with other manuscripts, and they are mostly insignificant. There is, therefore, no evidence for corruption. And we can conclude that S2322 is a reliable witness in a single text tradition.