Wednesday, January 27, 2021

Darrell Post on Family 1 in John: Five New Core Members

In May last year we published a guest post by Darrell Post on the newly registered minuscule 2957.

Darrell is a graduate of Virginia Beach Theological Seminary, Here below we publish another guest post in which he identifies five new core members of Family 1 in John.




The 2011 dissertation A Textual Study of Family 1 in the Gospel of John, by Alison Welsby, presented a complete textual analysis of John’s gospel among 17 manuscripts suspected of being related to Kirsopp Lake’s famous Family 1. Welsby included a proposed family stemma and reconstructed text of the Gospel of John according to Family 1. Her completed research placed the 17 manuscripts into the following groups:

Core Group: 1, 565, 884, 1582, and 2193

Venice Group: 118, 205, 209, 2713, and 2886 (formerly 205abs)

Subgroup: 22, 1192, 1210, 1278, and 2372

Miscellaneous: 131

Not a member of Family 1: 872

In her introduction, Welsby explained her selection of the 17 codices to study. 1, 22, 118, 205, 209, 1192, 1210, and 1582 were included based on the confirmation of Amy Anderson's work on Family 1 in Matthew. 131, 872, 1278 and 2193 were also selected because Anderson had found them to be weak Family 1 members in Matthew. Then Welsby added 565, 884, and 2372 as these were not included by Anderson, but were indicated by the Text und Textwert volumes as possible members of Family 1 in John's gospel. Finally, 2886 was also included in its own right, coming out from under the shadow of being formerly known as 205abs.

Having included additions to Family 1 found in Text und Textwert, Welsby’s work gives the impression that all available witnesses to the text of Family 1 in John were included in her dissertation, leaving no others with a GA number as of the year 2011. However, it seems the manuscript clusters tool available on the INTF internet page ( was not available to Welsby back in 2011. A quick search using this tool selecting GA-1, John, and Strict Grouping, reveals a much longer list of possible Family 1 members. 138, 357, 994, and 2575 are all indicated as related to Family 1, but none were included in Welsby’s dissertation. These four additional manuscripts also include a commentary in an alternating text format, and they all include the PA in its traditional location, 7:53-8:11 (other core Family 1 members place the PA after the end of the gospel). Further use of the INTF’s Clusters tool also suggest 809 and 2702 as more distant relations to Family 1.

Welsby’s subgroup, 22, 1192, 1210, 1278, and 2372 appears to have several additional members including 19, 149, 660, 697, 791, 924, 1005, and 1365. Most of these additional cluster relationships were also observed by Bruce Morrill in his 2012 dissertation on John 18 (see GA-138 on page 147).

I have also confirmed each of these proposed additions to Family 1 in John’s gospel as part of a project to collate manuscripts of John chapter 11. 138, 357, 994, and 2575 all belong to Welsby’s core group, with these four most closely related to 884 in John 11. I also found Welsby’s subgroup to be two related clusters, both with only a minimal connection to Family 1 itself.

Finally, 2517 was found to preserve only 12 leaves of John’s gospel, and only a small portion of chapter 11. However, this portion was found to strongly match 1, 565, 1582 and 2193. Bruce Morrill found the same results for 2517 in its extant portion of John 18.

The first table below shows the percentage of agreement in John 11 among the core members and the Venice Group, while the second table shows the agreement in John 11 among the subgroup.









There are completed IGNTP transcriptions of 138, 357, and 2575 available on the INTF manuscript workspace page. These three along with 994 and 2517 need to be fully studied and integrated with the work of Welsby toward better understanding the Family 1 text of John’s gospel.


  1. The catena group (138, 357, 884, 994, 2575) has previously been remarked upon by D.C. Parker (in Klaus Wachtel's Festschrift) – it should go where 884 does in Welsby, and her mention of the large number of singular readings in 884 should be interpreted accordingly (i.e. 884 is not a particularly good member of the catena group). Amusingly enough, Parker and I independently came to the conclusion that 2575 is a direct copy of 994. Sickenberger also notes Milan D.282 inf. (part of a four volume set, two of which have been catalogued as 2579) as a direct copy of 138. A curious feature about the catena group is that the text becomes f1 about two-thirds of the way through Luke – one thing I'd like to try and establish is if there is a change in the catena type at this point.

    If we're adding f1 witnesses from the catena group, we also need to include 2684, which does not have a catena, but seems to be very close to 994 from midway through John 14 to the end of John.

    2517, 809, and 2702 are interesting. I know little about any of them.

    As for the extra 22 group members (and indeed the division into stronger and weaker subgroups), many of these are already known from McReynolds and Wisse's CPM work on Luke and Omanson's on Mark.

    1. James, thank you for all that great info. There is also the supplement to 1784. Bruce Morrill found it to align with Family 1's Venice group in chapter 18. I found in chapter 11 that the supplement began in verse 20, and it closely followed the Venice group until verse 51 where it reverted back to closely following the MT. Somewhere between chapter 12 and 18 it apparently reverts back to the Venice Group.

  2. For what it's worth, I applied non-negative matrix factorization (a machine learning technique) to Bruce Morrill's collation of John 18 a long time ago, and it isolated the following manuscripts as members of the same group (from highest group mixture coefficient to lowest): GA 1, 1582, 357, 138, 565, 209, 994, 2713, 2575, 1784, 2684, 205, 2886, 2702, and 884. Unfortunately, even if I increase the number of clusters to identify to 30, the procedure doesn't break this cluster down into the A, B, and Venice subgroups of Family 1. The other two subgroups (containing GA 22 and 1005) get classified differently.

    1. Joey, that's interesting. I don't know anything about the calculation of the technique you mentioned. All I can speak to is the work I completed in chapter 11 and explain my methodology. I excluded things like minor spelling variations due to confusing vowel sounds, corrections made by the first hand, and obvious nonsense readings like starting a word at the bottom of a page and then starting it over at the top of the next page. I collated from John 11:1 through 12:2, a total of 991 words in the 2005 R-P MT text. Then I added up the variation units and did the math to get a percentage of agreement, as shown in the tables above.

      Also, I was not impressed with the data from John 11 for the subgroups (22, 1005, etc), as most of the readings they shared with F1 were also found among other manuscript clusters (thus far I have collated John 11 in about 550 manuscripts). Of course the data may well be different outside of John 11.

  3. Darrell,
    You've got collations of John 11 in about 550 MSS??! Is that available?

    1. James, yes, if you wish to provide a private email address, we could discuss further. Thanks.