So I’m debating whether to develop a project down the line dealing with text criticism (external evidence at least), and I want to put this idea out there and see if anyone thinks it might be fruitful.
Basically the problem has to do with the dating of manuscripts. First, dates inherently include margins of error, whether from carbon-dating, paleography, or other variables. These may be wide, hidden in that “circa.” Second, not all scholars working directly on dating the manuscripts agree with each other on how to narrow things down. One thinks the handwriting looks second century, another fourth. The use of the nomina sacra points one way, the binding on the papyrus another. It seems like the general solution is to pick the most convincing from among arguments which, without the manuscripts in hand (and the proper training), it may be difficult to judge. If we are qualified, it may be the case that the handwriting leans one way and the binding another — we should not dismiss the variance suggested by the data we have.
Granted, if the datings range from 150 to 400, some scholars may be bound to be right, and others wrong. But there is no way of deciding that without more data. It becomes an arbitrary choice between Type I and Type II errors.
And when we pick the dating we prefer, we tend to artificially narrow it down. A dating of 175-225 in the original study becomes ca. 200, which then becomes rigid as it is cited and cited again.
But think about two manuscripts with datings of 175-225, but with different readings. You might find scholars who say, “Both readings are equally as old.” But by the estimates, there could be 50 years or more separating the two. And think about the Church Fathers writing in those years: Irenaeus, Clement of Alexandria, Origen, Tertullian… These were theologically and scripturally active decades where fights were being fought over the authority of Scripture on a number of important points. The critical change could easily have happened within that 50-year span.
Is there not a more comprehensive, nuanced way to look at the transmission of readings in the various manuscripts that testify to them?
What I’m thinking of working on is a way to aggregate the various studies done on manuscripts to get a more fluid idea about how prevalent a reading was at a given time. Here is one really, really sloppily done example.
John 1:34 gives three different readings, in which John the Baptist testifies that Jesus is:
- God’s Son (ὁ υἱὸς τοῦ θεοῦ)
- God’s Chosen One (ὁ ἐκλεκτὸς τοῦ θεοῦ)
- Chosen Son (electus filius)
Each reading has its defenders, and no particular internal argument has won the day. Even English versions give different readings: CEV, KJV, NASB, and NRSV give “Son of God,” while NIV and NLT have “God’s Chosen One.” One appears in p66, the other (apparently) in p75, both of them early.
I took 17 early manuscripts (dated no later than the 7th century), and (again very sloppily) aggregated the various datings for each manuscript. The more diverse conclusions, or the fewer studies, the wider the margins of error. The result I got was this:
Does this make sense to people?
I should highlight that these levels of prevalence have margins of error that are wider the earlier on we look, so this would have to be refined to show that. But despite the similar dating of the earliest manuscripts for “Son” and “Chosen,” “Son” aggregates higher early on. One thing I found interesting is that “Chosen” seems to increase until it is almost 45% of the evidence around 270, when “Son” gets as low as 53%. It’s at this point, when the two readings are close to 50/50, that the conflation, “Chosen Son,” begins to rise. This would seem to suggest that Chosen Son is indeed a conflation and not the original reading, created (or gaining popularity) in response to the likelihood of seeing manuscripts of both types.
Does that sound like a fruitful way to look at this?
Other things to look at potentially:
- Refining this data, drawing in later manuscripts as well and addressing margins of error in these estimates.
- Comparing, for example, the Western text to the population or to another group of texts.
- Using this to test readings against turning point dates (i.e. manuscripts we are, say, 90% sure are prior to 270 or 350 with mss we are 90% sure are later — is there a significant difference in their reading of John 1:34? And what does this tell us?).
I’d love to hear anyone’s thoughts!