Wednesday, May 31, 2017

Missing Witnesses in Textual Flow Diagrams

After about 3½ years of studying the CBGM, I continue to learn new things about it. Sometimes I discover answers to question I was not even asking. This happened this week while reproducing the textual flow diagrams in my thesis. Because these diagrams connect each witness with their closest potential ancestor (within the constraints set by the user), someone smarter than me might  wonder what happens when a witness’s closest potential ancestor is not extant at the point of variation being studied.

Well, I found the answer. When this happens, any descendants are simply left out of the diagram, even if they are extant at this point.

Let’s give an example. If we look at the textual flow diagrm for the whole Catholic Epistles, we see that 104 is the closest potential ancestor for 1838 and 459. 459, in turn, is the closest ancestor for 1842.

104 and its descendants in the predominant textual flow for the Catholic Epistles
So, what happens when 104 is not extant? Its three descendants will disappear whenever the diagram is set to connect each witness with its closest potential ancestor (a connectivity of 1). In such cases 1838, 459, and 1842 are not displayed. For example, here is 2 John 1.2/2-6 with a connectivity of 1: You will not find the four witnesses I just mentioned.

At 2 John 1.2/2-6 with connectivity set to 1, 104 and its descendants are missing
But if you change the connectivity to absolute, they all reappear because they can be connected to other potential ancestors which, unlike 104, are extant at this point of variation.

At 2 John 1.2/2-6 with connectivity set to absolute, the descendants of 104 reappear
I think there are probably better ways to handle these situations, such as making the non-extant ancestors gray or circling them with a dashed border. Alternately, the descendants that have no ancestor available could just free float in the diagram.

I suspect this was something that was just overlooked in the original programming. Perhaps the situation will be treated differently in Acts. In any case, it’s worth knowing in case you’re using a connectivity of one and wondering where some of your witnesses have gone.


  1. Hi Peter,
    I don't understand your last sentence.
    Was is not the case, that all witnesses would reappear when using absolute connectivity?

    1. Conrad, you're exactly right. Typo on my part. I've fixed it.

  2. I raised this very issue on the CBGM list on 18 Feb 2013 with relation to minuscules 319 and 400 at 1Pt 1:1/8. I quote from Klaus's answer:
    "the explanation is that with connectivity 1 the script is looking for potential ancestors with rank 1 only and does not display any result, if such a potential ancestor is not found with one of the variants of the passage in question. The rule is followed here, but it is followed too strictly. You were the first to see this problem, thank you for the note. I think the best way to go about it would be to display the next available potential ancestor with the next highest rank number in such a case. For 1Pt 1:1/8a, 400/2 would remain in the graph then, and the user could be sure in such instances that the first potential ancestors is not extant at the passage. We are going to implement this for CBGM2 which will be released as soon as ECM2 for the Catholic Letters is available."

    It may not have been implemented then.

    1. Thanks, Dirk. Figures that you discovered this years before I did. ;)

      It definitely wasn't implemented in the CBGM 2 because that's what I was using for the diagrams here. Maybe in Acts.

  3. Peter Gurry,

    It seems like a lot of folks think that the CBGM is a took for making text-critical decisions. So, while you're posting about it:

    How many text-critical decisions, or what percentage of text-critical decisions, are already made in the course of map-construction using the CBGM? (Feel free to use the CBGM's use in the Catholic Epistles for a real-life example.)

  4. James, my books are on a truck right now so I can’t give precise numbers. The editors use pre-genealogical coherence at the beginning and make the majority of decisions with this type of data. Then they revisit various decisions multiple times with the help of genealogical coherence. In the end they leave maybe 100 undecided. That’s out of 3,000+ in the Catholics.

  5. In Acts the editors made decisions in 7000+ passages based on pre-genealogical evidence and left ca 400 open after that first phase. These and other passages were revisited in successive phases as genealogical evidence accumulated. Klaus Wachtel presented on the procedure at the SNTS in Szeged, a paper to which I responded. I can upload my powerpoint to our resource page. In the new ECM Acts we will of course have more details

  6. I should add that I illustrate this first phase in my articl on Mark 1:1 (we love that passage) in the TC journal