The Gospel of Johnphrase by phrase, as eleven Old Latin manuscripts most nearly agree

What a pathway is

These manuscripts disagree. A variome keeps every reading at every place where the manuscripts differ, together with the words they share. A pathway takes one reading at each of those places.

The variome used here is built from eleven manuscripts of John, as the International Greek New Testament Project has transcribed them. It holds about 10^{^3,808} pathways. The complete collection is not stored. Any single pathway can be generated when it is asked for, from a short recipe, and every pathway has a number. This volume’s number has 3,799 digits.

How this pathway was chosen

By a rule, applied mechanically. The manuscripts differ in 3,323 stretches, each running between words they all share. In each stretch the rule takes a whole phrase that one manuscript reads: the phrase that agrees, place by place, with the most readings of the others there.

A manuscript counts only where its first scribe’s reading survives whole, and every letter of it was read without doubt. Where two phrases tie, the rule takes the one more manuscripts read, then the one whose oldest manuscript is oldest.

Some manuscripts leave out whole verses. Whether such a verse is printed is decided first, by counting the manuscripts that have it against those that omit it. Only 7:53 is left out. Five manuscripts omit it and five have it, and the tie goes to the side with the oldest manuscript, which omits it.

The rule and its amendments

Each version of the rule was written down before its text was generated. The rule was revised twice, each time in response to the result before. Every version was checked by a second AI model, from a different developer, which reviewed the rule, the code that applies it and the claims made for it before its text was computed. Every earlier result is published below.

1. As first written, the rule chose each place on its own: the reading most manuscripts give there. The result joined words from different phrasings. At 1:9 it printed lux uerum, a feminine noun with a neuter adjective. It also printed half of a passage some manuscripts omit whole: at 8:2, iterum uenit in et populus ad eum. The review stopped it: 122 of its joins between neighbouring words stood together in no manuscript.
2. The first amendment chose whole phrases instead. That fixed the hybrids, but it accepted that manuscripts omitting a passage together could outweigh others that have it and word it differently. The result printed the story of the woman taken in adultery in pieces: 8:1–2, 8:4 and 8:10–11, without the rest.
3. The second amendment reversed that accepted consequence, because of that result. Whether such a verse is printed became a matter of counting manuscripts, not of how they word it. It is the last amendment this volume’s selection will have.

What it claims, and what it does not

• Every phrase, between the words the manuscripts share, is one that a manuscript reads.
• Not every join between phrases is. Of its neighbouring words, 3 pairs and 24 runs of three stand together in no manuscript. Each was reviewed and is described in the volume.
• The text was assembled by the rule above. It makes no claim to reconstruct the original text, or to represent a type of text scholars have established. Its complete text matches none of the eleven manuscripts.
• The eleven manuscripts are a selection, not every Old Latin manuscript of John. They are not independent of one another: several are largely Vulgate texts, and the rule counts each manuscript once.
• It reads most often with Codex Colbertinus, at 82.8 % of the places where that manuscript survives whole, and it matches none of the eleven.

Checks before printing

Lining up the corresponding words of the manuscripts (the alignment) is itself a choice. So the variome was aligned two more ways, and under the same rule 127 of the 879 verses read differently.

The review of the joins that no manuscript has is complete. The review of the 127 verses affected by the other alignments is in progress. Printing will wait until it is complete.

The supplement

The printed volume cites this page for the results its rule replaced. Each is given in full, with its audits, exactly as computed. The texts list each verse’s reference and words, spelled as the chosen manuscripts spell them.

Citing this volume

A pathway’s number, the version of its variome and a checksum of its text (a digital fingerprint used to check it) point to exactly one text. The volume prints all three, and its number in full.

The Gospel of John, phrase by phrase, as eleven Old Latin manuscripts most nearly agree. Variome, Single Volumes, I. Variome john-vl11-4, pathway 136239200728…618240933890 (3,799 digits). Text SHA-256 deb12450a99c6906369f57b8146791c73c96fe564d8ca7ca58cee6765a9d9c33.

Sources

Transcriptions: Vetus Latina Iohannes, edited by P. H. Burton, J. Balserak, H. A. G. Houghton and D. C. Parker (International Greek New Testament Project, CC BY 4.0). The eleven manuscripts are a, aur, b, c, d, e, f, ff2, l, q and r1. Their dates are those in H. A. G. Houghton, The Latin New Testament (Oxford, 2016). The Clementine Vulgate, the standard printed Latin Bible (here from Latin Wikisource), is used only to start lining up corresponding words. It never counts as a manuscript.