Thursday, September 15, 2022

Comparing New Testament Manuscripts


Recently I had some reason to compare a number of manuscripts in Acts and was happy to be able to use the CBGM tools for this, in particular the ‘Comparison of Witnesses’. The reason for this post is to point out some issues that are perhaps at first sight less intuitive. This is not criticism of the way the data set is constructed, as the actual manuscripts are always messier than what can be captured in a database.

1. Beware of comparing manuscripts to ‘A’, the reconstructed initial text. Why? Because at every place where ‘A’ has a ‘split reading’ it does not show up in the actual comparison; ‘A’ is treated as if it presents no reading here. Since the number of split readings equals around 2% of the total variant readings, there are some potential issues.

For example in Acts 1:6, we have ‘A’ and ‘01’ compared with ‘Show Agreements’ activated, and we see that the split reading at Acts 1:6/10 is simply not there.

2.    Equal does not always mean that things are the same. This caught me out when I was comparing P50 – arguably a difficult manuscript to encode – and 01. At three places differences between these two are given as ‘equal’ (‘=’) in the list (see below) despite a difference between the two: at Acts 10:28/22 (P50: ανδρι ιουδαιου; 01 ανδρι ιουδαιω), Acts 10:29/17 (P50: ουν; 01: -), and Acts 10:30/50 (P50: μου; 01: εμου).

In the actual apparatus of the ECM it becomes clear why these three differences are treated as being the same. At the first two places the original hand of P50 is deemed to have made an obvious scribal error which was corrected by the original hand, at the third place 01 is deemed to give a mere orthographic variant. This is all perfectly justifiable, but it is good to be aware of this when relying on the ‘Comparison of Witnesses’ to give you all differences. You would not get this extra information from simply relying on the list of Agreements/Differences.

Not equal does not always mean that you will see the difference. The example to illustrate this comes actually from the ECM apparatus, which gives here more detailed information than the comparison of witnesses does. Nothing comes up in the comparison at Acts 10:30/10, yet the ECM tells us that P50 has a scribal error in the word απο (recorded as P50f) namely οπο. Again, I can see the rationale why such difference does not show up in the Comparison of Witnesses. However, there is some inconsistency in recording corrections by the original hand on itself in P50 (as the first two examples in 2. above). At Acts 10:30/18 the first hand corrects τη to ταυτη, which – in analogy to 10:28/20 and 10:29/17 – should probably be recorded or show up somewhere. But this time it is not found as a 'difference that is equal' or as a remark in the apparatus of the ECM.

4. Despite best intentions, the apparatus at ECM can still beat me. Again this has to do P50, and in particular how P50 is treated at the place of the split reading at Acts 10:28/34-38 in the ECM. The top line of the split reading has the support of P50C*V (so the probable reading of the original hand correcting itself). But then for the complete absence of this passage (absence, so not an omission) we find P50(C)*, and what this might mean beats me, though I trust there is a perfectly logical explanation.

As so often in scholarship, it is only by using a tool that we learn about its strengths and limitations. I am still impressed how much the data gathering and sharing by the INTF has enabled progress and deeper understanding of the textual tradition of the New Testament, yet data are never as hard as we want them to be. 


  1. I have some thoughts on P50.

    1. So, is it an amulet, and what's the evidence for that?