There is something interesting going on in the apparatus of NA28. But I need a long introduction ...
For any critical edition of the Greek New Testament a decision is made which manuscript to include in the apparatus and which not. Nestle-Aland 28 divides the manuscripts up between ‘consistently cited manuscripts’ and the others. In the introduction to NA28 a clear (!) explanation is given why manuscripts were included as 'consistently cited'. Eighteen manuscripts were included because they have the ‘initial text’ as their closest potential ancestor. There is also a 19th manuscript in this category, minuscule 468, but this one was replaced with minuscule 307 since the Byzantine text was well represented anyway. In addition to these, 88 and 1881 are added for one of the letters only, 33 because it is so interesting, 1448 and 1611 because they are Harkleian, and 642 because it represents a particular Byzantine group. All the papyri are included as well. We all know that for the Catholic Epistles NA28 is dependent on the Editio Critica Maior, 2nd edition (ECM2) and there the continuous witnesses of the first sub-group above (the 18 + 1) are given under the label ‘witnesses that have A as potential ancestor with rank one’.
The term ‘closest potential ancestor’ comes from the Coherence Based Genealogical Method (and if you don’t understand what that is, you should come to the SBL session, which will teach you everything you need to know). What is important for now is that the CBGM is an iterative method, which shapes the overall structure of the witnesses according to the decisions you take during the editing process.
You can of course revisit again and again, but at some points you have to publish something. So the INTF published the Catholic Epistles between 1999 and 2005 (Editio Critica Maior, 1st edition), and the shape of the witness tradition at that particular point was reflected in version one of ‘Genealogical Queries’ (the database of decisions and data underlying the text). Lucky for them, the INTF had the opportunity to use the data collected at that particular point to go over the text of the Catholic Epistles again. So the starting point was ‘Genealogical Queries version one’, which developed in the course of editing the second edition of the Editio Critica Maior into ‘Genealogical Queries version two’.
At the start of working on ECM2, there was a list of witnesses that, at that point, had the initial text as their closest potential ancestor. It is this particular list that is given in the introduction to ECM2 as the witnesses with rank one. However, as a result of editing ECM2, this list changed quite a bit. In Genealogical Queries version two’, which reflects the decisions made for ECM2, three of the witnesses no longer have the initial text (A) as their closest ancestor, 442, 2344, and 2492. Now minuscule 442 has rank 8, 2344 rank 2, and 2492 has sunk all the way down to rank 11. [Update (PJG): 1735 has also dropped out of this category. A has dropped to rank 2 behind 02.]
What does this mean for the justification of what manuscripts to include as ‘consistently cited witnesses’? The group of important manuscripts as found in the apparatus of NA28 is based on the group the editors started out with when working on the text, but is not the group of witnesses that they ended up with. Three of the manuscripts, at least within the method, appeared less important for the text that was produced then initially thought. If the selection of manuscripts to include in the apparatus was made again today, these three manuscripts might well be deselected (unless they were deemed to be interesting for some other reason).
So when do you use ‘Genealogical Queries version one’ and when ‘version two’? If you want to know what the editors started out with for the second edition, you use ‘one’, if you want to know what they ended up with, you use ‘two’. ‘Version two’ reflects the situation at the end of the process. But, counterintuitively, the data from ‘version one’ and not ‘version two’ have been used to draw up the list of ‘witnesses of rank one’ (in ECM2 terminology) which is the same as the first group of ‘consistently cited witnesses’ (NA28 terminology). Current thinking would have left out the three witnesses mentioned earlier.
I needed some patient help from Klaus Wachtel to explain the background and that the pressure of publishing was a big factor in setting up ‘Genealogical Queries version 2’ only after NA28 and ECM2 went to press. And I needed Peter Gurry to point out to me that the list in ECM2 had something to do with the consistently cited witnesses for the Catholic Epistles in NA28. I just hope I have explained all the confusion correctly.
So what was the interesting thing going on in the apparatus of NA28? Well, three of the manuscripts cited in the Catholic Epistles (442, 2344, and 2492) do not have an obvious right to be there.
We don’t make it ourselves easy in this discipline, do we?
For any critical edition of the Greek New Testament a decision is made which manuscript to include in the apparatus and which not. Nestle-Aland 28 divides the manuscripts up between ‘consistently cited manuscripts’ and the others. In the introduction to NA28 a clear (!) explanation is given why manuscripts were included as 'consistently cited'. Eighteen manuscripts were included because they have the ‘initial text’ as their closest potential ancestor. There is also a 19th manuscript in this category, minuscule 468, but this one was replaced with minuscule 307 since the Byzantine text was well represented anyway. In addition to these, 88 and 1881 are added for one of the letters only, 33 because it is so interesting, 1448 and 1611 because they are Harkleian, and 642 because it represents a particular Byzantine group. All the papyri are included as well. We all know that for the Catholic Epistles NA28 is dependent on the Editio Critica Maior, 2nd edition (ECM2) and there the continuous witnesses of the first sub-group above (the 18 + 1) are given under the label ‘witnesses that have A as potential ancestor with rank one’.
The term ‘closest potential ancestor’ comes from the Coherence Based Genealogical Method (and if you don’t understand what that is, you should come to the SBL session, which will teach you everything you need to know). What is important for now is that the CBGM is an iterative method, which shapes the overall structure of the witnesses according to the decisions you take during the editing process.
You can of course revisit again and again, but at some points you have to publish something. So the INTF published the Catholic Epistles between 1999 and 2005 (Editio Critica Maior, 1st edition), and the shape of the witness tradition at that particular point was reflected in version one of ‘Genealogical Queries’ (the database of decisions and data underlying the text). Lucky for them, the INTF had the opportunity to use the data collected at that particular point to go over the text of the Catholic Epistles again. So the starting point was ‘Genealogical Queries version one’, which developed in the course of editing the second edition of the Editio Critica Maior into ‘Genealogical Queries version two’.
At the start of working on ECM2, there was a list of witnesses that, at that point, had the initial text as their closest potential ancestor. It is this particular list that is given in the introduction to ECM2 as the witnesses with rank one. However, as a result of editing ECM2, this list changed quite a bit. In Genealogical Queries version two’, which reflects the decisions made for ECM2, three of the witnesses no longer have the initial text (A) as their closest ancestor, 442, 2344, and 2492. Now minuscule 442 has rank 8, 2344 rank 2, and 2492 has sunk all the way down to rank 11. [Update (PJG): 1735 has also dropped out of this category. A has dropped to rank 2 behind 02.]
What does this mean for the justification of what manuscripts to include as ‘consistently cited witnesses’? The group of important manuscripts as found in the apparatus of NA28 is based on the group the editors started out with when working on the text, but is not the group of witnesses that they ended up with. Three of the manuscripts, at least within the method, appeared less important for the text that was produced then initially thought. If the selection of manuscripts to include in the apparatus was made again today, these three manuscripts might well be deselected (unless they were deemed to be interesting for some other reason).
So when do you use ‘Genealogical Queries version one’ and when ‘version two’? If you want to know what the editors started out with for the second edition, you use ‘one’, if you want to know what they ended up with, you use ‘two’. ‘Version two’ reflects the situation at the end of the process. But, counterintuitively, the data from ‘version one’ and not ‘version two’ have been used to draw up the list of ‘witnesses of rank one’ (in ECM2 terminology) which is the same as the first group of ‘consistently cited witnesses’ (NA28 terminology). Current thinking would have left out the three witnesses mentioned earlier.
I needed some patient help from Klaus Wachtel to explain the background and that the pressure of publishing was a big factor in setting up ‘Genealogical Queries version 2’ only after NA28 and ECM2 went to press. And I needed Peter Gurry to point out to me that the list in ECM2 had something to do with the consistently cited witnesses for the Catholic Epistles in NA28. I just hope I have explained all the confusion correctly.
So what was the interesting thing going on in the apparatus of NA28? Well, three of the manuscripts cited in the Catholic Epistles (442, 2344, and 2492) do not have an obvious right to be there.
We don’t make it ourselves easy in this discipline, do we?
Two thoughts on that:
ReplyDelete1) This is a nice illustration of the iterative process allowing you to revise your decisions once you (seem to) have better judgment available.
2) It would be good to know whether the different lists of witnesses may be reflective of different sets of data. As the ECM Catholic Epistles came in batches the dataset grew over time (from James to finally include all seven). This could have contributed to revise earlier decisions on variant passages which in turn might have caused a different overall picture for the position of some manuscripts.
In light of Ulrich's second thought: Is any record of decision-making kept in this regard? It would be nice if some of those issues were transparent.
ReplyDeleteI second Andy's remark, related to which is the (perhaps surreal?) thought of a textual commentary...
ReplyDeleteThe local stemmata give a record of every decision INTF made for CBGM1&2. That's about as transparent as they've ever been at one level. Of course, there's no commentary except for scattered discussions in the literature.
ReplyDelete"Well, three of the manuscripts cited in the Catholic Epistles (442, 2344, and 2492) do not have an obvious right to be there."
ReplyDeleteCare to elaborate on that?
As a non-specialist in the field, but someone who has tried to read as much as possible on CBGM, I continue to struggle with both the goal(s) and implementation of it. This article appears to make the CBGM a step in the process of evaluating variant readings which still are subject to the other methods historically applied to determine the original reading, even to the extent that such further evaluation is used to go back and reconfigure the closest potential ancestors. Yet, in practice, based on NA28 it appears that Muenster followed the CBGM religiously, even when it revealed that the closet relative to the initial text was a single late Greek manuscript or even if it showed the closet relative did not include any Greek manuscripts. Admittedly, I am foggy on the whole process but what you describe here sounds different than what Muenster describes.
ReplyDeleteHopefully, experts, such as those on this site have a full grasp of the CBGM, and one day will feel the desire to explain it.
I am committed to using the best representative of the original/initial text from the pulpit, and am glad for the NA/GNT text, and believe the best tools available should be used. The CBGM may be one, but it appears that us non-specialist may never know.
Tim
I will gladly be corrected by those who know better here, but let me take a swing at this. (Maybe it takes an amateur to give an overview for other amateurs.) In my attempts to use the CBGM with the data available to me, I've found that it breaks a lot of the assumptions we have been taught as students. It's hard to get a grip on it without breaking them ourselves. And yet it's not radically different from the process you can find on display in the old Text und Textwert series. It's just finally being used radically: from the root.
DeleteWe are obsessed with manuscripts. With witnesses. But the picture the extant witnesses give us is not representative of the whole. It doesn't do what we hoped it might when we tried to divide it into families and pit them against one another to see which was closest and most original. The situation is much more lacunose, and much more parallel, than that. There are readings we have a surplus of manuscript copies for, that are truly in the minority and late; and there are readings we have a deficit of witnesses to, that we know were prevalent early and lost that status through historical processes—including readings that may only be preserved in non-Greek versions, of which the study has historically been meager.
The older process worked from both ends to the middle, in a way: from manuscripts at one end, and from variants at the other. The early Moderns with a growing wealth of archaeological data were continuing a basically Erasmian process, heavily determined by preference among manuscripts, polling the manuscripts decide the variants. This involved high subjective bias toward certain manuscripts within a limited field of options. In order to make that process more reliable, with a still-growing volume of data at their disposal, the Alands (as key figures operating in the middle of the process) pushed this closer to the "variants first" side. They used strategic polling of variants to correct certain biases of manuscript preference. This is where the opposition to the late Byzantine Koine forms finds solid ground: the majority was routinely found to be both late and editorial. So we got the Alands' rating system of Alexandrian, common, Byzantine, and unique readings per manuscript per canon, and we shuffled our manuscript preferences accordingly. But the Nestle-Aland text that resulted is still quite dependent on manuscript preferences, even as those were pushed closer to variant data to justify and correct them. We had more easily-rationalized preferences.
The CBGM is designed to push this process even closer to the variant-first side of things. It should do exactly what you complain that it does when the INTF uses it "religiously," by which you mean, in an effort to be consistent about their real priorities. It reveals possibilities we had previously ruled out. And it should do that, while doing exactly what Dirk shows it doing: ruling out possibilities we had previously considered. On the one hand, the most plausibly initial text of a variant may come from any manuscript, even a late one, and it should always point us beyond itself because that manuscript is not original. Preservation in Greek of close relatives is not a genealogical criterion; so much that we want hasn't been preserved at all! On the other hand, it should also allow us to rule out witnesses on the basis of iterative use, because the more we refine our picture of the variants, the better a picture we get of the witnesses that contain them. Manuscript priorities are going to fluctuate in both directions.
Tim, I'll let you know once my SBL paper is available. I am even working on an 'Idiot's Guide' to the CBGM, but am not sure yet if that will be worth my while.
ReplyDeleteOn the other hand, I find Richard Laurence's 200-year-old comments regarding Griesbach's system quite applicable, mutatis mutandis, to the current CBGM scenario:
ReplyDelete“It is natural therefore to expect, that every novel mode of ascertaining the validity of a reading will be at first received with caution, and long watched with jealousy. And notwithstanding the ability which has been displayed in support of Griesbach’s theory, notwithstanding the high tone which it has assumed in the literary world, I must confess, that it is far from producing in my own mind complete conviction.
“I shall . . . only be understood as urging the propriety of circumspection upon the points of the practical conception and application of Griesbach’s particular hypothesis . . . : but as it is extremely liable to be misconceived as well as misapplied; is so intricate in its construction; is so difficult to be detailed with precision, or even to be made out in its subordinate arrangements; and is so readily convertible to party purposes; surely we should again and again contemplate it, and that in every possible point of view, before we consent to admit the conclusions which have been deduced from it into general currency.”
Richard Laurence, Remarks upon the Systematical Classification of Manuscripts adopted by Griesbach in his Edition of the Greek Testament. Oxford: At the University Press for the Author, 1814, pp. 6-7.