I apologise for the length of time between my examination of the external evidence and the internal evidence of Romans 8:28. Feel free to go back and refresh your memory as to the previous post. I changed all the Greek to Times New Roman so it should show up fine but there are no accents. The original manuscripts had none and as we are doing TC I see no reason why we need them 🙂 Let’s get to work!
The first thing we can do is to include P46 with the others as a witness to τον θεον παντα συνεργει ο θεος εις αγαθον. Although the change from παν to παντα or vice versa may explain the variant it is likely an accidental scribal error. If ο θεος is original it would cause συνεργει to be transitive and παντα would be the accusative direct object. If we do not have ο θεος, παντα possibly becomes an accusative of reference according to Wallace (203-4). He says that this is rare in Koine Greek and such identification should only be employed as a last resort. This type would be used to qualify a statement that would otherwise typically not be true. He notes that Cranfield dismisses the likelihood of παντα as an accusative of reference assuming the longer reading and agrees with him (204 n.93). He does use it as an example of a substantival use of the accusative (180). Wallace dismisses the longer reading and leaves us with two probable options: either ‘he works all things together for good’ or ‘all things work together for good.’ In the first instance the subject is embedded in the verb and ‘God’ is clearly implied (as in v29). In the second instance, παντα becomes the subject of an intransitive verb” (181). This discussion is needed because in order to get inside the head of the scribe we need to understand how he could have variously construed the original and changed it.
While it is obvious that ο θεος makes συνεργει transitive, it is unclear whether we should take God as the implied subject in those manuscripts where is missing. If the shorter text is original it is probable that a scribe added ο θεος in order to make this obvious. On the other hand because God immediately precedes παντα it may have seemed repetitious and a scribe dropped the second noun, leaving the subject to be supplied. However, it is not immediately obvious that the subject is God and therefore a scribe could have added it. We have come full circle only to find that we cannot be conclusive. F. F. Bruce states that the addition “makes the construction excessively heavy.” Therefore this would seem to be the more difficult reading.
Bruce also prefers the NEB translation which saw the subject of ‘works’ as that of the preceding clause, ‘the Spirit.’ However, as William Hendrickson points out this would make Jesus the Son of Holy Spirit. This would be a good time to note the eight possibilities that exist for this verse as noted by Cranfield:
1) To accept the longer reading and explain παντα as an accusative of respect (‘in all things’, ‘in all respects’).
2) To accept the longer reading and explain συνεργει as used transitively and παντα as its object (so for instance, Sanday and Headlam translate: ‘cause all things to work’; while the RV margin gives: ‘worketh all things with them’).
3) To accept the shorter reading and supply ο θεος, explaining παντα as in (1).
4) To accept the shorter reading and supply ο θεος, explaining συνεργει and παντα as in (2).
5) To accept the shorter reading and take παντα as the subject of συνεργει.
6) To accept the shorter reading and understand the subject of the verb to be the same as the subject of the last verb of verse 27, namely το πνευμα, explaining πνευμα as in (1).
7) As in (6), but explaining συνεργει and παντα as in (2).
8 ) To accept the shorter reading with the emendation of παντα to πνευμα or το πνευμα.
If we decide on the shorter reading we must first decide whether the referent of the verb is found within the sentence or whether it is to be supplied from context. If it is to be found within we accept the traditional translation: “all things” work together for good. Gordon Fee lists several reasons for rejecting this based on internal grounds. First, Paul never uses παντα as the subject of an active verb. Secondly, παντα almost always precedes the verb when it appears as the object of a personal verb. Thirdly, συνεργει appears twice more in Paul and in both instances the verb has a personal subject. I would then agree with Fee in rejecting the notion that “all things” is the subject of “works.” We thus reject option (5). We can readily reject option (8) as being implausible as it would be hard for a scribe to eliminate the Spirit from a passage either accidentally or deliberately. If one does emend the text the resulting reading almost demands its emendation back to παντα. We are then left with two longer and four shorter readings. In tackling the shorter readings we must deal with the identity of the supplied subject. Readings (3) and (4) take it to be God and (6) and (7) to be the Spirit. Fee argues for taking το πνευμα as the subject on the strength of the context. The Spirit has been the subject of the argument since 8:1 and he is clearly the grammatical subject of 8:27. The strength of this argument fails because the unexpressed subject of the following οτι clause has to be God, not the Spirit. The condition expressed in Romans 8:28 is such “in order that those he foreknew he foreordained to be conformed to the image of his Son, that he might be the firstborn among many brothers” (8:29). We must therefore question whether this transition did not occur previously beginning with the statement “according to God’s will.”
Fee also mentions the increase of συν- compounds in the argument beginning at 8:16. “The Spirit bears witness together with (συνμμαρτυρει) our spirits” and “the Spirit assists (συναντιλαμβανεται) us in our weaknesses” (8:26). He thus sees συνεργει as another instance in which Paul is referring to the joint working of the Spirit with us. BDAG here gives the meaning of συνεργει as “assist (or work with) someone to attain something or bring something about.” Theologically, God always works in us and with us through his Spirit and thus Holy Spirit has to be the subject of the verb. This argument only works if we restrain “all things” to the inner reality of the persons who love God. In other words, God makes their reality such that ‘all things’ external glorify him in their lives regardless of the ordinary understanding of good. It could not be used to argue that God uses events in themselves for that good.
I must reject the Spirit as being the subject based on context. The Spirit’s work is done in accordance with the will of God (8:27b) and as a consequence this is a beneficial thing because God works in all things to bring good to those who love him. It is the will of God that determines the Spirit’s actions and it is the will of God that causes all things to work. We thus reject options (6) and (7) and are left with deciding whether ο θεος was added in order to clarify the subject or whether it was removed from an already crowded sentence. If the longer reading was original it is hard to conceive of a scribe deliberately removing God from the sentence unless he thought the meaning was clear. It would be more logical for him to substitute autoß than to remove ο θεος. However, it is possible that a scribe removed ο θεος believing that συνεργει would supply the subject. Rodgers mentions one small precedent for this at Romans 1:28 where ο θεος is omitted by א* A 0172*. I agree with Rodgers that this may not be an accidental omission but reflect a scribal tendency. However, the evidence does not appear strong enough to say that this was the case here especially given the fact that A includes ο θεος at 8:28. Furthermore, in 1:28, the removal of God did not create the same kind of ambiguity that is apparent here.
It seems most likely that the shorter reading was original and a scribe added ο θεος in order to make the subject clear. We are thus decided on the shorter reading and are left with options (3) and (4). The decision between these two will be left for the conclusion, as the decision is broader than internal textual evidence.