Over at the Accordance Blog, Rick Bennett has posted on "The Search for ‘Sacred Names’." In this post Rick summarizes the paper he presented at the SBL in Rome (which I blogged about here).
In his excellent SBL paper, Rick demonstrated how Accordance with the GNT-Pap module - a digitized version of Comfort and Barrett, now extended with some new papyri, 𝔓118-121, and 𝔓123, can be used to search for and analyze the distribution and form of nomina sacra.
I have just been using this module myself when I searched for some nomina sacra, although I used the first version which has been updated since then. Of course I compared the results with the printed edition (and images and other sources). I did spot some errors and limitations along the way, that I have reported to Accordance.
Errors is to be understood here as deviances from the printed version of Comfort and Barrett (2d edition), not as de facto errors (which unfortunately occur too). I found three errors relating to P45: Luke 11:11 (wrong brackets); Acts 11:8 (wrong brackets); Acts 16:18, Comfort and Barrett abbreviates IH XP in the printed version, but the digital version has IH XY. I think these errors occur both in GNTPAP-C and -M. I suspect that the wrong bracket formatting, [ ... [ or ] ... ], occur at several places in the module, not just in the transcription of P45, and I assume that these could be searched for rather easily by folks like Rick Bennett who knows practically everything there is to know about search strings (needless to say that he has helped me a lot with those).
One limitation that still remains in the latest version is the possibility to exclude hits that occur within conjectural text set in square brackets. Rick tells me that the developers will do something about that in a future version. For now one has to sort that out manually. Some other limitations that were included in an earlier version of this post have now been dealt with in the current version of the module (which I don't have for the moment).
In sum, the software is very useful for this kind of research. The fact that one gets hits in the reconstructed passages is a limitation, but in the end, the most important thing is that all the factual hits are there, and it is clearly better to get more hits and exclude irrelevant ones, than to miss some cases.
Update: the above post has been edited since some of the limitations that I brought up are not relevant for the latest version of the module. Thanks to Rick Bennett for updating me :-)