In a 2008 article in JBL, Don Garlington shared the fact that he wrestled with the idea that if the preposition e0k in the phrase e0c e1rgwn no/mou, in texts such as Galatians 2:16, is understood as a reference to “origin” (i.e., coming out of as by production or causation) then the Reformed concept that Paul is negating a legalism that believes justification can come from good works may have merit to it.
In mulling the question over yet another time, I had to concede that if Paul’s use of ἐκ retains a strict and exclusive sense of “origin,” then the traditional rendering of justification “by (means of) the works of the law” might just have some merit. If, in fact, justification originates in “the works of the law,” then said “works” could be construed as the “legal basis” of right standing before God. By conviction, I assume a stance within the camp of the so-called New Perspective on Paul. But still, I had to ponder matters afresh.
Hence, Garlington, as a proponent of the NP, proceeds to argue that the preposition should be taken as a partitive that refers to the Jewish group that belongs to the sphere or orb of “Torah-works.” It is this group that Paul is supposedly arguing against, and declaring that one needs to belong to the group within the sphere of “Christic faith.”
The problem with his argument is that the preposition e0k does not allow for a concept of sphere in the way that he is using it. He seems to be using it in a way that the object of the preposition merely indicates a sphere within which a dissimilar entity can dwell within that sphere. In other words, he seems to be saying that e0k + op = something outside the delimitations of op. He states
In point of fact, Paul uses prepositions carefully. Characteristically, he employs διά, ἐν, and ἐκ. The first (διά) speaks of means: one is justified by the instrumentality of faith in Christ. The second (ἐν) denotes locality or sphere: Paul avows that justification is to not be located or found within the parameters of the ancient covenant people. Inherent in the third (ἐκ) is the notion of source or origin. But in this vicinity of the letter, to be demonstrated presently, ἐκ takes on the nuance of belonging: to be “from” a realm means to belong to it—the partisan use.
Instead, however, the partitive indicates that one of a number of like entities is being singled out. In other words, if I say that X is not of Y in the partitive sense, I would be saying that X is not one of a larger group of Y. In this case, righteousness is X and Y is the “works of the Torah.” There is no group of humans in view, just the activities of Torah. What this means is that if one were to take the e0k in Galatians here as a partitive, it would not mean that this refers to a group that is “within the sphere of Torah-works,” but rather that righteousness is not a work that belongs to the works of the law. In other words, the numbered entity is the “works of the Torah,” not “a group that performs the works of the Torah.” Hence, a partitive would single out one likened entity from that group, i.e., a work from the group of larger works (here denying that righteousness is one of those works). It would not refer to a person who was a part of the works of the Torah. That’s a complete misunderstanding of the partitive. In order to achieve that interpretation, one must prove that e1rgwn no/mou refers to a people group rather than actions/observances and that “righteousness” is also a people group (i.e., a part of the whole). To argue such, however, is to bend the words to fit a preconceived ideology, which is, ironically, a fallacy that NP proponents often attribute to their opponents. The phrase “works of the Torah” is not a people group but a group of deeds; and righteousness is not a single part of a larger group of works. Hence, a partitive would only identify a deed from among the group of deeds. Ergo, the partitive use of e0k does not save the NP interpretation here, and Garlington’s dilemma is unresolved. If, in fact, e0k is taken either as a reference to “origin” or as a partitive reference, both would indicate that the traditional Reformed understanding of the phrase has more weight to it. In fact, if the partitive use is in view, then this would indicate that there is not even a single work of the Torah that can produce righteousness, and this might have some force to it since Paul is here addressing a single work of the Torah. However, the partitive use is a bit forced here, and the traditional understanding of e0k here as that of “origin” (i.e., “out of,” “from.” with the concept of production by a source) is to be preferred, since righteousness is a state of one’s being or a sum of the entirety of one’s works, not a single work within a larger group of works that one performs. Hence, the traditional translation as “not from/by works of the Law” stands as the best option out of the two.
Of course, I’m sure the NP movement will survive just fine without this argument. It just won’t survive using it as its primary justification for moving away from the traditional understanding of the phrase.
 Don Garlington, “Paul’s ‘Partisan e0k’ and the Question of Justification in Galatians,” JBL 127 (2008) 567.
 Ibid., 573.
 Garlington (Ibid., 574) does not go so far as to reinterpret righteousness here in such a way, but he does appeal to Gal 3:7, 9 to bolster the claim that this refers to a people group, or sphere, in which one may dwell. The problem is that the article in 3:7 and 9 refers to uioi “sons” and if the ek is partitive in those passages, it would mean that the sons were a part of the larger group of faith. That makes no sense. Sons and faith are not similar entities. Instead, understanding the preposition as one of origin is best here. Hence, “the sons who are produced by faith, these same are the sons of Abraham . . . so then the sons who are produced by faith are blessed together with Abraham, the one who had faith.”