Text Determination in the "Plucking Grain on the Sabbath" Pericope

(Matthew 12:1-8; Mark 2:23-28; Luke 6:1-5)

by Dr. Jakob Van Bruggen (The Netherlands)[1]

Note: This appendix uses the bwgrkl Greek font of Bible Works.


Between the Authorized Version and the modern English translations there are various differences which go back to differences in the basic text followed. Sometimes Nestle25 (N25) places the siglum p) beside the reading now abandoned, to indicate the opinion that these readings have "probably crept in from one of the other Gospels" (Preface N25 p. 80*). In total N25 mentions six cases of parallelising readings in the Koine-text (K-text) (see Table I). This arouses suspicion about the K-text in general and against the readings at issue in particular. Thus Metzger in his Textual Commentary writes at Matthew 12:4: "Although evfagon is supported by only B and 481, as the non-parallel reading it is more likely to have been altered to evfagen than vice-versa."

Comparison with the Critical Apparatus in Aland's Synopsis Quatuor Evangeliorum (1964) shows that the suspicion about the abandoned readings does not always remain. Twice the Synopsis allows the qualifying designation p) to drop out and in one of these cases it also abandons the hesitation about the correctness of the text originally followed (in Mark 2:26 the square brackets around pwj disappear; in UBS3 they do not reappear). On the other hand a variant not mentioned in N25 is included as a p)-variant to the discredit of the K-text (see Table II).

Number and distribution of the p)-variants

A study of the readings which can be qualified as p) leads to the result that all types of text include such readings.

In the list of possible p)-variants (see Table III) the two readings which were qualified in N25 (but not in the Synopsis) as p)-readings are included. Also included was the variant +deuteroprwtw from Luke 6:1 (omission of this word is very similar to the omission of the words epi Abiaqar arcierewj in Mark 2:26; this last case is qualified as p) by Aland in the Synopsis). Also, the insertion of twn in Luke 6:1 was considered as a p)-variant. So, four other variants were included, besides the nineteen mentioned by Aland in the Synopsis as p)-variants, according to the criteria which Aland evidently had used in the other nineteen cases. Inclusion of these four readings (numbers 2, 11, 12, and 13) is more to the detriment of the K-text than to its advantage; B has only one while the K-text has three of the four variants indicated as p). The distribution of these possible p)-variants and their number per manuscript or group can be found on the left half of Table III.

The complete list of p)-variants must now be closely examined. If of two readings the one is similar to parallel-gospel I and the other to parallel-gospel II, none of the two readings can be qualified as p) (see numbers 10, 17, 23). In the case of number 11, there are special reasons for not using the description p): if BD have the original text, one must admit that at some moment, simultaneously, pwj was inserted in Mark and w`j in Luke. In addition, there are still a number of unclear cases: in numbers 14, 16, and 21, the possible assimilation to another gospel is only present in one manuscript, or it is accompanied by simultaneous dissimilation. When we look at the variants which remain as serious candidates for the title p), it appears that these harmonising readings occur the least in the K-text (see the right half of Table III).

Internal and external criticism

The omission of all the p)-variants would lead to a text which sometimes follows B (numbers 2, 3, and 13) and sometimes the K-text (numbers 4, 12, 19, and 20). The remaining five points of difference between B and the K-text (numbers 10, 11, 17, 21, and 22) could be solved on external grounds. In four of these five cases the Koine-reading is not a specific K-lectio but also occurs in D (10, 21), W (10, 11, 17), or Sinaiticus (11, 17, 21). In these four cases the K-lectio may be chosen. This means that B is followed three times against eight times for the K-text. There is reason to ask whether the K-text should not be followed in number 23 as well: it also is not a specific K-lectio.

In the Synopsis of Aland, B is followed eight times (2, 3, 10, 12, 13, 17, 21, 23) and the K-text four times (4, 11, 19, 20) in the twelve differences between B and K-text now being discussed. In the text of UBS3 the K-text has also been abandoned in number 19 and has been placed between square brackets in number 20. Thus B has been subscribed to nine or ten times and the K-text three or two times. This means that the number of readings in the text of UBS3 qualified as p) increases by two or three (12, 19 and (20)). This increase in the amount of assimilating readings in the text of the United Bible Societies is the result of the abandonment of K-readings under the influence of P4 (this papyrus evidently turned the scale in favor of an altered point of view in Luke 6:3; also in Luke 6:4 the first word w`j is placed between square brackets under the influence of P4). Is assimilation objectionable when it is found in the K-text and not when it occurs in a papyrus?

Or is a reading no longer called assimilating when it appears in a papyrus? This suggestion finds support in the strange fact that the designation p) for the reading efagen in Matthew 6:4 is abandoned in Aland's Synopsis as soon as P70 is also mentioned as a witness to this reading. Such a proceeding raises questions concerning the applicability of the rule of internal criticism that a nonparallel reading deserves preference.

Are p)-variants really p)-variants?

From the presupposition that in the transmission of the text there was a process of assimilation and harmonization, scholars began to distinguish between so-called parallel and nonparallel readings. The question arises whether a framework has not been pressed upon the data (see last paragraph).

Example (number 2):

The reading efagen seems to be a complete formal assimilation to Mark and Luke. As an assimilating reading it must, however, be of a later date. How can it then appear in P70? Aland, Synopsis, now abandons the sign p) with this reading. But surely the nature of readings does not change when they occur in papyri? Aland's omission of the designation p) makes us ask for a different kind of approach to these and other readings. Now the apparent nonparallel reading efagon can be described as internal (inside Matthew) assimilating. It gives a better association with the words directly preceding it (epeinasen kai oi met ,auton), and the words directly following it (autw ) ) ) ) oude toij met ,auton). However, while the reading efagen does correspond formally with Mark and Luke, it does not do so materially, for by means of this reading the emphasis lies more on David's deed ("they that were with him" now stands in the shadow). It appears to be characteristic and specific for Matthew that he places the emphasis in this pericope more on (David and) Christ personally (cf. Matt. 12:5-7, but not in Mark and Luke). The notion of "giving to those who were with David" is absent in Matthew. This implies that the reading efagon can be described as an alteration of the text with the purpose of improving it philologically, but with the effect that it becomes more vague in content and more assimilated to Mark and Luke, where the eating is explicitly related to Jesus and the disciples, to David and the people in his company.

A new look at the readings which have by now been selected as real p)-variants leads to the conclusion that these variants can also be explained without the p)-model.

1. A number of readings can be interpreted as the making explicit of a preposition (number 1), subject (number 5), article (number 13) or prepositional object (number 15); in all these cases it is clear from the context that these are meant. Regarded as p)-variants these readings would be secondary, but as explicit-making readings they can be authentic: (1) because of the more "Semitic" character; (2) because of the circumstance that with the dictation of a text the omission of apparent dispensable details is more acceptable than insertion by the writers.

2. In one case the thought of mutual influence between the Gospels seems acceptable because this influence is reciprocal (numbers 7 and 18 vice versa).

3. In a reasonably large number of cases variants can be regarded as the result of philological improvement of the text:

Numbers 2 (see discussion above).

3 (Mark-Luke: ou`j ouvk evxestin Mt: o` ouvk evxon ''''''''''''''''''''hvn).

4 (diaporenesqai in connection with dia twn sporimwn).

9 (omission of a difficult, apparently incorrect, and also dispensable element).

12 (like 9).

19 (o`te in this case better Greek than o`pote).

20 (omission of ovntej as the removal of needless redundacy).

22 (the sentence does not run well; this is solved by the omission of labwn

(Sinaiticus D W), or by the omission of w`j (P4BD), cf. number 11).

A review of the distribution of the readings placed in the right column of Table III as real p)-variants according to the evaluation now offered in three groups leads to the conclusion that the ph)-variants (philological improvement variants) are totally absent in the K-text (see Table IV).

The application of the p)-criterion led to confusion; the application of the ph)-model leads to the coincidence of internal criteria and external data. The absence of ph)-variants in the K-text gives us occasion to grant this text our trust.


With the K-text as a basis it is possible to explain the divergent variants (more explicit, assimilating, philological improvement), but it is not possible to explain from the B-text or the D-text how a K-text originated (especially where the readings in the K-text are definitely not philological improvements).

If N25 gives the impression that a non-K-text can be taken as a basis, this is due to the fact that N25 neither follows B completely nor D; if all the variants from B rejected as p) had been maintained in the text, no model could be developed for the explanation of the variant-groups.

Example: If in Luke 6:4 w`j/'pwj was originally absent and labwn was to be found there, then it would be inexplicable why pwj was added (via assimilation to Mark?) without a simultaneous omission of the consequently difficult labwn also by assimilation to Mark. If the K-text is followed then the embellishment of the somewhat paratactically jerky elaben kai to labwn calls for measures regarding w`j.

Another example is the choice of the reading o`te by UBS3 in Luke 6:3. How can one explain that this reading would ever at a later time be substituted by the reading o`pote which deviates from Matthew and Mark and is poorer Greek?


For the pericopes under consideration the K-text is evidently the most recommendable. This conclusion may not automatically be transferred to other pericopes. Yet, in applying the model presently in use for the selection and evaluation of readings elsewhere, it does induce one to test it critically against the totality of a literary text-unit, as well as the variation of readings occurring within. Atomistic treatment of variants is not the same as text-determination.

[1]Jakob van Bruggen is professor of New Testament exegesis at the Reformed Theological College in Kampen, The Netherlands. The material in this appendix comes from an unpublished lecture first given in Dutch. With the exception of a few stylistic changes, the translation into English sent to me by Dr. van Bruggen is reproduced verbatim and with his permission.

Critics of the first edition have pointed out, appropriately, that my discussion of "harmonization" was weak. I include this appendix as a partial response to such criticisms. I believe it justifies two conclusions: Many apparent harmonizations may reasonably be interpreted in other ways, and the "Alexandrian" and "Western" texts may be said to be just as guilty of harmonizing as the "Byzantine" text.

Previous Appendix
Table of Contents
Next Appendix