New Testament Papyri Before the 4th Century
Several times in the past years I’ve wished for a good visual representation of what early New Testament manuscripts actually cover. But I’ve never seen one anywhere. I finally decided to try to make one. I also wanted to highlight significant variations and be able to calculate some statistics. Here they are:
Aren’t they beautiful? Now for a bajillion caveats. I knew this was going to be an error prone process, so I wanted to basically do it twice with different sources so I would have a way to cross-check my results. I decided to start with Wikipedia, because they have some really accessible tables and lists of verses on each manuscript that’s hard to find elsewhere. The plan was that once I had tested my methodology with that easier list, I would then follow the same procedure with the INTR New Testament Virtual Manuscript Room, and compare the two so I could correct any typos I made and update Wikipedia if I discovered any errors while I was at it. Unfortunately, I lost interest 3 manuscripts into INTR iteration, and the project has just been sitting for months, and I don’t know if I’ll ever get back to it. So I decided to post what I have. If people are actually interested maybe that will drive me to actually finish the project properly at a later date. And I can take whatever criticisms I get into account when I finish the project. In the meantime, please take all of this with a big grain of salt, there could be a lot of errors, both due to my clumsiness and due to relying mostly on Wikipedia as a source.
I restricted this project to papyrus, because the statement I was fact-checking when I started this project was about papyrus. There are a couple of non-papyrus New Testament manuscripts with dates early enough they could have been included, and I think most people interested in this information don’t care what the manuscript was written on. I might make that update at some point, but from a quick glance the changes don’t look very significant. It would just add a couple more highly fragmentary documents. I included papyri if their earliest generally accepted date was in the right century, using dates generally from Wikipedia, although I did get a bit further checking dates with INTR than I did with the rest of the project. The papyri included are:
2nd Century: 4, 52, 77, 90, 98, 103, 104, 167
3rd Century includes those plus: 1, 5, 9, 12, 13, 15, 16, 18, 20, 22, 23, 27, 28, 29, 30, 32, 39, 40, 45, 46, 47, 48, 49, 53, 64/67, 65, 66, 69, 70, 72, 75, 78, 80, 87, 91, 92, 95, 100, 101, 102, 106, 107, 108, 109, 111, 113, 114, 115, 118, 119, 121, 125, 132, 133
I did my analysis at the verse level. If a manuscript was good quality and I estimated more than half of the words were present, the verses were coloured dark green. If a manuscript was fragmentary and I estimated more than half the words were missing I coloured it light green. In the second iteration I was actually counting the number of words present in each verse to make it less subjective. This was part of what made the project a never-ending grind and resulted in me quitting, so I doubt I’ll finish that even if I do come back to it. But my initial estimates are just estimates, and for the most part they are based on the overall quality of the manuscript, without looking at individual verses. There are definitely verses in generally good quality manuscripts that appear at the end of a page and contain only a couple of words that will be coloured dark green because I didn’t go into that level of detail. If Wikipedia said the document contained a range of verses, my checking is spotty. For example, I’ve got a note that the Wiki on Papyrus 46 says it is missing some parts of 1 and 2 Corinthians in spite of the summary listing all verses as present. Most places where I noticed contradictions I did check the references and figure out which was right, but by this point I was tired and decided I’d figure out which parts when I went through the manuscripts, but never got around to it. I doubt there’s much error the other way around, most of the time if I used light green it’s very unlikely there’s an intact line let alone a verse, and more than once when I was checking whether a manuscript had a variation because the verse was listed in Wikipedia as being present in the manuscript but wasn’t mentioned in any of the discussion around that verse, I found that the verse was represented by only a couple of faded characters on the edge of the page. I never found an instance where a single character from a verse was on the manuscript but it wasn’t listed in Wikipedia. Therefore, I think the number of verses coloured green almost certainly errs on the high side.
Variations presented a more significant problem. How should I judge what is significant? I definitely didn’t want to include all differences, like typos. But many changes that scholars think were done deliberately, to influence doctrine, are small changes in prepositions that would probably seem pretty insignificant to most people (“baptized IN water”, “released us FROM our sins”). After all, deliberate changes had to be subtle in order to not be detected. I definitely don’t want to be arbitrarily deciding what’s important and what’s not.
At minimum, the variation should result in a change in meaning, significant enough that it could be translated to a different language. That should eliminate typos, variations in spelling, changes in word order that don’t alter meaning, repeated lines or missing lines that result in nonsense and are thus obviously just scribal errors, etc. To be translatable, it has to be significant enough to change the meaning. To further narrow it down, I wanted the difference to be something that people actually debated, to make sure I was only including differences that actually matter. Because this is intended to be more about what everyday people care about than scholarly research, I didn’t want to limit this to expert debates. Of course, I quickly realized that was useless. Everything about the Bible is heavily debated, and even things like the number of letters in a word matter to people practicing numerology, so people do debate variations that aren’t even translatable.
In the end I decided that the reason people care about this is that people want to know how different those manuscripts are from what they read in their Bible today. The most popular Bible translation today, the NIV, was done by scholars researching these manuscripts, and it has footnotes that denote differences between manuscripts where there is debate about the original reading. So there aren’t really any differences there. But a lot of churches, especially in the US, use the King James Version (KJV), which was written before many of these early manuscripts were discovered, and with many technological disadvantages. That’s the Bible translation still in use today that has the most differences with these early manuscripts, so I decided that should be the basis of my comparison. Any place where the translators of modern Bible versions argued on the basis of one of these manuscripts that the KJV translation should be modified I marked as a significant variation. I also did a Google search to see if I could find people debating the difference to try to determine if it was significant, but I don’t think there was a single verse where I didn’t find debate. If the NIV differs from the KJV in any way, people on the internet are mad about it.
I never found a great source to reference for these variations. My Google searches for people mad about Bible translations led to a lot of lists made by random angry people, so I mostly went through those and verified which ones were justified with manuscripts on my list. Most of these sources I don’t even want to name or link to because believing that the KJV is superior seems to correlate with a lot of nasty racism. I was going to try to obtain a variorum to reference for the final version, or perhaps compare the Nestle-Aland to the Textus Receptus, but this project will never be completed if I have to add “learn ancient Greek” to my todo list. If anyone has any suggestions for an impartial way to find variations please let me know.
The variations were coloured orange if they occurred in all known manuscripts before the 4th century, and yellow if some matched and some did not. If an entire verse was missing it was red. By missing I don’t mean that a section was torn out of the papyrus, I mean that the verse should have been in the preserved part of the manuscript but was not, like when Matthew 21:43 is immediately followed by 21:45 with no sign of verse 44. If a verse is missing in some but not all manuscripts it is pink (but I only found one instance of that).
In total, 3396 of the 7956 verses (43%) in the KJV New Testament were coloured dark green (substantially present in at least one manuscript), while 1014 (13%) are light green (partially present in at least one manuscript) for a total of 57% of verses at least partially represented. Note that this doesn’t necessarily mean 57% of the New Testament, as the percentage of words represented would be lower, and the percentage of characters even lower than that, so it very much depends what you’re counting. This is also using the earliest generally accepted dates for the manuscripts, if I had chosen a date in the middle of the range there would be fewer manuscripts included. As you can see from the PDFs, the verses represented are not spread evenly throughout the New Testament, and some books are far better represented than others. There were 148 significant variations identified, 23 of which were verses missing in their entirety. That’s about 3% of the total verses.
Note that a verse being green doesn’t mean that the manuscript confirms the KJV text. Note how most of the orange appears in dark green sections. In most cases, where a controversial variation exists in later documents, the manuscripts I marked light green text do not contain enough of the verse to confirm which variation was in that manuscript. Many of those big light green chunks in the image are where we have a vertical strip from a page of papyrus. That strip might just be a word or two, or even just a few characters wide. Therefore it can’t really be used to argue for either variant.
Also, because I was focusing on manuscript-inspired differences between modern versions and the KJV, I did not colour any of the changes in a manuscript that are agreed by scholars to probably not be the original. That hides a lot of the variations that exist in these manuscripts. In particular, most of the green in Matthew, Mark and Acts comes from a single document, Papyrus 45, which is known for kind of summarizing the text instead of copying it. There are a ton of variants in there. For example, in Mark 6:40 he omits the “by hundreds and by fifties”, and in John 11:25 “and the life” is omitted. Papyrus 45 is the only early source for some of these verses, and I have no doubt that these would be controversial if they were thought to be real. Because they are contained to this manuscript, and scholars agree this was probably a characteristic of this scribe and not a reflection of the original text, I left them light green. But that most certainly doesn’t mean that this manuscript confirms the KJV text.
I’m still not really sure how to address this issue. I know that if I write that 3% of the verses have significant differences, that’s going to be turned around to say that 97% of the verses confirm the KJV text. In reality, probably about a quarter of those verses are either insufficient to confirm or contradict the KJV text or have substantial variations that scholars just don’t think are original. I would estimate the reality is about 3% have intentional differences, 72% are substantially the same except for obvious innocent mistakes, and 25% are either too low quality to tell, or intentionally different but unlikely to be original. But because my methodology didn’t record differences not thought to be original I have no real data on that.
Anyway, it is what it is. If you think it could be useful to you give me some feedback and I’ll see if I can find the energy to finish it sometime.
I I decided to include the missing verses in the numerator (because they’re obviously big variants) but not the denominator (because the verses aren’t in the manuscript). This would have made things a bit weird mathematically if the number of missing verses was significant. But it’s negligible, so it doesn’t really matter.