A refutation of Timothy Lim on Galatians 3:16 and Paul's 'literalism', with the testimony of the Constantinople Pentateuch
The evangelical scholar Timothy H. Lim (Holy Scripture in the Qumran Commentaries and Pauline Letters, Oxford, 1997, pp. 63–65) presents Galatians 3:16 as the clearest example of literalism in Paul; he relates it to the rabbinic technique of al tiqrei (אל תקרי) and to the Qumran pesharim; he offers two biblical precedents for זרע in an individual sense; he explains the contradiction with Galatians 3:29 by the plurality of senses in ancient exegesis and by 'corporate solidarity'; and he concludes that objections to this argument are modern anachronisms. The passage now circulates in Spanish as though it closed the discussion: a Spanish-language biblicist spreads it in the service of a programme, that of 'Judaising' Paul. This article maintains the opposite, point by point. Galatians 3:16 is not literalism; it has no parallel in rabbinic or Qumranic literature; its grammatical alternative exists in no text; and every link in Lim's defence gives way before the Hebrew of Genesis, the Greek of Galatians, and the controls constructed in earlier studies (the Josephus control and the three levels of proof). To this is now added a witness whom Lim does not even name: three promises to Avraham (Gen 13:15; 17:8; 22:17) as they were read, in their own Greek, by Greek-speaking Jews who translated from the Hebrew and not from the Septuagint, and whose text, in all three, does not contain the word on which Paul builds his argument. The viewpoint is that of the philologist: the question is not the antiquity of the method, but whether the argument works with the words it has.
I. What Lim maintains, in his own terms
It is fitting to set out Lim in full before refuting him, so that the reader may verify that nothing is attributed to him that he does not say. We cite the Spanish version circulated by the biblicist; the pagination is that of the Oxford edition.
Lim opens with Gal 3:15. He finds it 'curious' that Paul should express the irrevocability of a human testament with notions 'that resemble the terminology of addition and subtraction in Deuteronomy' (Lim, 1997, pp. 63–65). He conjectures, with a 'perhaps', that ἀθετέω and ἐπιδιατάσσομαι deliberately evoke Deut 4:2 and 13:1, whose legal requirement, he says, Paul expects his opponents to obey.
He then arrives at the centre: the argument about the singular of σπέρμα would be 'the clearest example of literalism in Paul' (Lim, 1997, pp. 63–65). Lim admits what the Hebrew does not allow one to deny: the formula 'and to your seed' appears in Gen 12:7; 13:15; 15:18; 17:8 and 24:7, and in Genesis, although singular, it 'refers collectively' to the descendants of Avraham (Lim, 1997, pp. 63–65). Paul, he concedes, insists on the singular sense 'against the more natural interpretation' (Lim, 1997, pp. 63–65). To explain the procedure he evokes al tiqrei and proposes a twist: Paul is not suggesting another reading of the text, but is drawing attention to what is written, and 'perhaps' he is inverting the technique of his opponents, who would have said 'do not read "seed" but "seeds"'.
Lim then looks for support in two cases of individual זרע (Seth, Gen 4:25; Samuel, 1 Sam 1:11) and concludes that Paul chose the singular sense 'for his own purposes', without ignoring the plural (Lim, 1997, pp. 63–65). Next comes the contradiction with 3:29, where 'seed' means all who believe. Lim acknowledges that to the modern reader the move from singular to plural will look like an inconsistency and a trick, but he maintains that 'such criticisms would be out of place in the ancient world' (Lim, 1997, pp. 63–65), because ancient Jewish exegesis read more than one meaning into a single written word. And he adds that 'it is possible' that Paul has in mind the 'corporate solidarity' of believers with Christ. He closes with the general thesis: there are features of literalism in the pesharim and in the Pauline letters, and the idea that the biblical text was modified 'is not a modern notion anachronistically projected' (Lim, 1997, pp. 63–65).
Three of those sentences deserve to be cited in full, because the mechanism can be seen in them. On al tiqrei and the inversion of the opponents: 'Perhaps Paul is once again inverting the exegetical technique of his opponents: whereas they say "do not read 'seed' (as is written in the text of Genesis) but 'seeds'", Paul replies by pointing out that the text in fact says "and to (your) seed" and not "and to seeds"' (Lim, 1997, pp. 63–65). On the shift from singular to plural: 'The fact that sperma can be both singular and plural accords with the nature of much ancient Jewish exegesis, which interprets more than one meaning of the same written word. The literal meaning is only one of the several meanings of the biblical text' (ibid.). And on the conclusion: 'attention to the biblical text is often confused with a figurative, allegorical or non-literal interpretation, where the meaning is more important than the very words used' (ibid.). The last sentence is the most revealing. Lim defines literalism as attention to 'the very words used'; but the word Paul needs, 'seeds', is not one of the words used by the text. It is the sophism in miniature: attention to a word that the text does not contain is called 'literal'.
Observe the architecture. Of the six steps of the reasoning, three rest on a 'perhaps' or an 'it is possible': the echo of Deuteronomy, the inversion of the opponents and corporate solidarity. The other three (the literalism, the two precedents, the plurality of senses) are assertions that the following sections examine. And the function of the passage is declared by Lim himself: the promise is fulfilled in a single heir, Yeshu, 'and not in the many descendants who observe the law' (Lim, 1997, pp. 63–65). That sentence should be retained. The argument is not an exercise in grammar; it is an instrument for excluding from the promise the descendants who keep the Torah.
II. Lim's propositions and where each one fails
III. The 'literalism' that is not
Lim calls Gal 3:16 literalism. The label is mistaken for three reasons. First: a chapter later, in the same letter, Paul classifies his own reading of Genesis with another word:
ἅτινά ἐστιν ἀλληγορούμενα ('which things are said allegorically', Gal 4:24).
– P46, Folio 084v, University of Michigan Library Inv. Nr. 6238, recto.
The author does not speak of literalism; he speaks of allegory. Second: the literal is what is read in its context, and the context of the promises that Lim himself enumerates (Gen 12:7; 13:15; 15:18; 17:8; 24:7) is collective in every case. In 13:16 the seed is like the dust of the earth, and is counted as dust is counted: וְשַׂמְתִּי אֶת זַרְעֲךָ כַּעֲפַר הָאָרֶץ… גַּם זַרְעֲךָ יִמָּנֶה ('And I will make your seed as the dust of the earth… your seed also shall be counted'). In 15:5 it is said while looking at the stars: כֹּה יִהְיֶה זַרְעֶךָ ('So shall your seed be'). In 15:13, in the same chapter as 15:18, the singular noun governs plural verbs and pronouns: גֵּר יִהְיֶה זַרְעֲךָ בְּאֶרֶץ לֹא לָהֶם וַעֲבָדוּם וְעִנּוּ אֹתָם ('your seed shall be a stranger in a land that is not theirs, and they shall serve them, and they shall afflict them'). And in 17:7–12 the covenant 'between me and you and your seed after you, throughout their generations' (לְדֹרֹתָם) is sealed with the circumcision of 'every male' (כָּל זָכָר): a seed that is circumcised by generations is not an individual. Lim concedes this, as we saw, but treats it as a matter of local colour and not as what it is: a fundamental objection to his own label. Third: what Paul does is not to read the text, but to isolate a grammatical form from the context that defines it and to extract from it a referent. That is not literalism; it is atomism. And Lim himself acknowledges that Paul chooses the singular sense 'for his own purposes'. An argument that comes from the purpose and not from the text is not literal.
IV. The alternative that does not exist
The argument of Gal 3:16 sets 'to the seeds' (τοῖς σπέρμασιν) against 'to your seed'.
– Codex Vaticanus, folio 1490a.
But that pair does not exist as a reading option. In Hebrew, זֶרַע has no plural meaning 'descendants': the only biblical plural is that of sown seed (זַרְעֵיכֶם, 1 Sam 8:15), and the Mishnah had to coin זַרְעִיּוֹת ('descendants') in order to reason as it does in Sanhedrin 4:5, where the plural is indeed written: דְּמֵי אָחִיךָ צֹעֲקִים (Bereshit 4:10), 'it does not say "the blood of your brother" but "the bloods of your brother": his blood and the blood of his descendants' (דָּמוֹ וְדַם זַרְעִיּוֹתָיו). The Sage reasons from a written form; Paul, from a form that no text contains. In the LXX, for its part, there is no σπέρματα for 'descendants' in any of the promises to Avraham. The only σπέρματα that the LXX offers is that of sown seed: in 1 Kgdms 8:15 (1 Sam 8:15 in the Hebrew text), τὰ σπέρματα ὑμῶν renders זַרְעֵיכֶם, the crops that the king will tithe; and the Mishnaic order of Zera‘im, 'Seeds', is so called because of crops. The plural 'lineages' that Paul needs in order to say 'it does not say σπέρματα' is, as far as I can see, free Greek and not biblical Greek. Paul disputes with a plural that nobody has read.
The simplest proof that the opposition 'seeds / seed' does not exist in Hebrew is provided by those who have had to write it in Hebrew. Two Hebrew versions of the New Testament render Gal 3:16, and both must do the same thing with the same word. The first is that of Franz Delitzsch, in the eleventh edition (1892), whose title page declares the origin of the text:
ספרי הברית החדשה נעתקים מלשון יון ללשון עברית על ידי החכם פראפעסאר פראנץ דעליטש ז"ל ('Books of the New Covenant, translated from the Greek language into the Hebrew language by the sage Professor Franz Delitzsch, of blessed memory'; eleventh edition, 1892).
The title says it in so many words: the Hebrew text is a text translated from Greek. This is how he renders Gal 3:16. It reads, on p. 338:
וְהִנֵּה לְאַבְרָהָם נֶאֶמְרוּ הַהַבְטָחוֹת וּלְזַרְעוֹ וְלֹא־אָמַר וְלִזְרָעֶיךָ כְּאִלּוּ לָרַבִּים אֶלָּא כְּאִלּוּ לְיָחִיד וְלְזַרְעֲךָ וְהוּא הַמָּשִׁיחַ
'And behold, to Avraham the promises (havtaḥot) were spoken, and to his seed (zar‘o); and he did not say "and to your seeds" (lizra‘ekha), as if to many, but as if to one, "and to your seed" (lezar‘akha); and he is the Mashiach.'
The second is that of the Bible Society in Israel, prepared at the beginning of the 1970s by an ecumenical team of scholars (Joseph Atzmon, Yohanan Elihai, Bob Lindsay, Gabriel Grossman, Miriam and Yohanan Ronen and Yann de Waard). It reads, on p. 241:
וְהִנֵּה הַהַבְטָחוֹת נֶאֶמְרוּ לְאַבְרָהָם, וּלְזַרְעוֹ. לֹא נֶאֱמַר "לִזְרָעֶיךָ," כְּמוֹ עַל רַבִּים, אֶלָּא "לְזַרְעֲךָ," כְּמוֹ עַל יָחִיד, וְהוּא הַמָּשִׁיחַ.
'And behold, the promises (havtaḥot) were spoken to Avraham and to his seed (zar‘o). It was not said "to your seeds" ("lizra‘ekha"), as of many, but "to your seed" ("lezar‘akha"), as of one, and he is the Mashiach.'
What is decisive lies in a single word. Both versions write לִזְרָעֶיךָ, 'to your seeds': a plural of זֶרַע with a suffix, which the translator has to fabricate in order to make Paul say what he says. That form is not a quotation: the Hebrew of Genesis has לְזַרְעֲךָ in 13:15 and in 17:8, and there is not in the whole Bible a verse that says 'to your seeds' in the sense of 'to your descendants'; the only plural of זרע that biblical Hebrew knows is that of sown seed (זַרְעֵיכֶם, 1 Sam 8:15). The two Hebraists, without intending it, do what retroversion imposes: they put into Moshe's mouth a word that Moshe did not write. Delitzsch moreover leaves it without quotation marks, so that the Hebrew reader receives it as if it were the text of the Torah; the version of the 1970s, more careful, places it in quotation marks, and thereby makes visible that it is a quotation that does not exist.
The nuances of the two versions show to what extent this is an operation of translation. Delitzsch gives the Greek comparison (ὡς ἐπὶ πολλῶν… ὡς ἐφ᾽ ἑνός) the form כְּאִלּוּ… כְּאִלּוּ, 'as if… as if', a particle of the Hebrew of the Sages that gives the sentence an air of derashah; the later version prefers כְּמוֹ עַל, closer to the Greek, and puts the 'he did not say' into the passive (לֹא נֶאֱמַר, 'it was not said'). And both translate Χριστός as הַמָּשִׁיחַ, 'the Mashiach', with the article: it is the usual translation, but it has an effect, because it turns a Greek name into the title of which Jewish expectation speaks, and makes the identification pass for a messianic reading of זרע, when, as far as I can see, none of the Jewish sources examined here identifies with the Mashiach the seed of the promises to Avraham.
There, in miniature, is what it means to 'Judaise' Paul by translation. The title page is presented with the conventions of the Jewish book (החכם, the date by the minor reckoning, לפ"ק); the Hebrew is that of the Bible and of the Sages; and the result is that the reader who does not have the Greek before him believes himself to be facing an exegete of the house of study arguing on the letter of Genesis. But the only road by which that argument can reach Hebrew is the one its translators had to construct: inventing the plural. The 'Hebrew Paul' is not a discovery of philology; it is a product of the version. And the version, with its own title page, confesses where it comes from: nit‘aqim mi-lashon Yavan li-lshon ‘Ivrit, 'translated from the Greek language into Hebrew'.
Lim suggests that the formula recalls al tiqrei ('do not read… but…'). Al tiqrei proposes another reading of the consonantal text (another vocalisation, another letter: אל תקרי בניך אלא בוניך, Berakhot 64a),
And it is attested in the tannaitic midrashim, in the Yerushalmi and in the Bavli, that is, in corpora later than 70; as far as I can see, neither Qumran nor Philo nor Josephus knows it as a formula. Paul proposes no alternative reading of the text: he proposes an alternative that the text cannot produce. The analogy would require him to say 'do not read זרע but זרעים', and that, in Hebrew, neither exists nor yields the sense sought. Lim tops this with a conjecture: that Paul 'inverts' the technique of his opponents, who would have said 'do not read seed but seeds'. No text attributes that argument to anyone. It is what John Barclay called mirror-reading (JSNT 31, 1987): reconstructing the adversary from the author's reply and then presenting him as proof of the reply.
Before measuring Paul against this formula, it is well to know what the Sages who used it did with it. A witness of the first rank is R. Zvi Hirsch Chajes, who devotes chapter 25 of his Mevo HaTalmud to the function of אל תקרי and gathers the cases from the Palestinian baraitot, from the Yerushalmi and from the Bavli. He is of interest for two reasons: for the list of examples itself, which shows the material with which the formula works, and for the closing discussion on whether the derashah of al tiqra is a true reading or a mere support (asmakhta) for an already established norm. The passage is cited and translated.
ודע כי גם בברייתות א״י ובתלמוד ירושלמי מצינו, סמיכות כאלו הנאמרים בלשון אל תקרי, רק לפעמים מבואר ממש לשון א״ת, ולפעמים סתמו הדבר אולם הוא ענין הנלמד מעצמו, דכוונו לדרשא זו, כמו במשנה דקנין תורה אמרינן אל תקרא חרות אלא [חירות], וכן הברייתא ת״ח מרבים שלום בעולם א״ת בניך אלא בוניך. וע׳ מכילתא פ׳ בא דלמדו מקרא ושמרתם את המצות, מצוה שבא לידך אל תחמיצנה, וע״כ דדרשו דרך א״ת המצות אלא [המצוות], וכן בירושלמי פ״ק דברכות, מלאכים אין להם קפיצה שנאמר קריבת על חד מן קמיא קימיא קרינין שהן עומדין, וכן בירושלמי ברכות ר״פ אין עומדין בהדרת קודש בחרדת קודש, וע׳ ירושלמי פ״ד דשביעית, נפלים יש להם חלק לעולם הבא דכתיב ונצורי ישראל להשיב ונצירי ישראל להשוב. הפירוש דבלא וא״ו רק ביו״ד משמע יציר כל מי שנוצר מישראל, אז בהעדר הוא״ו וגם היו״ד משמע נצר. היינו כל ענף מישראל ישיבו, ובעל יפ״מ לא ידע לפרש, וע׳ ירושלמי פ״ג דנזיר אין שחיטה מחוורת דבר תורה, דכתיב וישטחו להם שטוח, אל תקרי וישטחו אלא וישחטו להם שחוט, והיינו בחילוף אותיות, כמו כבש כשב תחת חתת וכדומה, וכן הוא בספרי פ׳ בהעלותך מובא (יומא ע״ה ע״ב), אכן על שיטת הרמב״ם דפירש דעקר דרשה א״ת, אינו רק דרך אסמכתא לדבר תורה, למען יהיה נסמך הדרש לכתבי קודש ולא יבא לשכחו, א״כ איך יפרנס קו׳ הש״ס (ברכות ל׳ ע״ב) והא בהדרת כתיב, וכן מה דמקשה הש״ס (ר״ה י״ג ע״ב) אל תקרי לשלש אלא לשליש, ושדה שנגמרה שליש, מיקרי גמר לענין מעשרות, ומקשה הש״ס והרי לשלש כתיב. ולשיטת הרמב״ם הרי אין כאן דרשא גמורא רק אסמכתא בעלמא למען יהיה נסמך למקרא, ומה איכפת לן אם נאמר בהדרת או נאמר לשלש סוף סוף יכולין לסמוך דרך זה לסימן בעלמא. וכבר תמהתי ע״ז בהגהותי לש״ס וצ״ע:
(Translation: And know that also in the baraitot of Eretz Israel and in the Jerusalem Talmud we find such supports (semikhot) expressed in the idiom al tiqrei; only that sometimes the al tiqrei idiom is explicitly stated, and sometimes the matter is left unstated, although it is a matter learned of itself, for they intended this derashah. Thus, in the Mishnah on the acquisition of Torah we say: 'Do not read ḥarut but [ḥerut]'; and likewise the baraita: 'Disciples of the Sages increase peace in the world: do not read banayikh ("your sons") but bonayikh ("your builders")'.
See also the Mekhilta, parashat Bo, where from the verse 'And you shall guard the matzot' (ushmartem et ha-matzot) they learned: 'A commandment that comes to your hand, do not let it leaven'; and so, necessarily, they expounded it by way of al tiqrei: do not read ha-matzot but [ha-mitzvot] ('the commandments').
And likewise in the Yerushalmi, the first chapter of Berakhot: 'Angels have no qefitzah (lit. "leaping", that is, flexing of the joints), as it is said: qeribet ‘al ḥad min qamaya ("I approached one of those who stood there", Dan 7:16); we read qayyamaya, for they stand.'
And likewise in the Yerushalmi of Berakhot, at the beginning of the chapter Ein omedin: one does not stand [in prayer] in hadrat qodesh ("the splendour of holiness") but in ḥardat qodesh ("the trembling of holiness").
See also the Yerushalmi, the fourth chapter of Shevi‘it: 'Untimely births (nefalim) have a share in the world to come, for it is written: u-netzurei Yisrael lehashiv (Isa 49:6) and [also] u-netzirei Yisrael lehashov.' The explanation is that, without a vav and with only a yod, [the word] means yatzir, that is, anyone who was formed (nitzar) from Israel; and in the absence of both the vav and the yod, it means netzer ('branch'). That is: every branch from Israel they shall restore. The author of the Yefe Mareh did not know how to explain it.
See also the Yerushalmi, the third chapter of Nazir: 'Slaughter is not clear [as a requirement] of the Torah, for it is written vayishteḥu lahem shatoaḥ ("and they spread them out for themselves a spreading", Num 11:32); do not read vayishteḥu but vayishḥatu lahem shaḥoṭ ("and they slaughtered for themselves a slaughtering")'. This is a permutation of letters, as in kevesh / keshev ("lamb"), taḥat / ḥatat and similar cases. So also in the Sifrei, parashat Beha‘alotekha, cited in Yoma 75b.
Now, according to the view of Maimonides, who explained that the essence of the derashah of al tiqra is no more than a means of asmakhta (support) for a matter already established in the Torah, so that the teaching may be supported by the sacred writings and not come to be forgotten, how then will one resolve the objection of the Talmud (Berakhot 30b): 'But behadrat is written'? And likewise what the Talmud objects (Rosh HaShanah 13b) concerning 'do not read lishlosh ("to three") but lishlish ("to a third")', that is, that a field that has reached a third [of its ripening] is deemed 'finished' (gamar) with regard to tithes; and the Talmud objects: 'But lishlosh is written.' According to Maimonides, indeed, there is here no true derashah, but a mere asmakhta so that [the norm] may be supported by the verse. What does it matter to us, then, whether it is said behadrat or lishlosh? In the end one can support [the teaching] in this way as a mere mnemonic sign (siman). I have already expressed my perplexity about this in my glosses to the Talmud, and it requires further study (tzarikh ‘iyyun).)
(R. Z. H. Chajes, Mevo HaTalmud, ch. 25.)
Note on the text. In the Hebrew we have used, the pairs ḥarut… ḥarut and ha-matzot… ha-matzot are graphically identical, which appears to be due to the loss of a yod or a vav in the edition or in the transcription. The readings required by the sense are restored in square brackets (חירות, Avot 6:2; המצוות, Mekhilta of R. Yishmael, Bo, 9). The same may occur in the citation of Isa 49:6, whose variant is explained in the text itself (ketiv and qere), so that the spelling may depend on the edition the author used.
Chajes's list teaches, first of all, with what material the formula works. The cases are four mechanisms and one constant. In ḥarut / ḥerut and banayikh / bonayikh the vocalisation of the same consonants is changed; in Dan 7:16 an Aramaic word is read differently (qeribet… qamaya is read qayyamaya); in Isa 49:6 the alternative is the ketiv against the qere, with or without yod and vav; in Num 11:32 the letters are permuted (וישטחו / וישחטו). In none of them does the formula introduce a form that the consonantal text cannot give. The alternative reading always comes from the written letters, by vocalisation, by division or by permutation. That is the reason why the formula 'do not read X but Y' makes sense: Y is contained in what is written. The plural σπέρματα that Paul opposes to 'your seed' does not emerge from the letters of זַרְעֲךָ by any route: it is not another vocalisation, nor another division, nor a permutation; it would be another word, זרעים, which the text does not bring.
It will be objected that Chajes also records cases in which the formula is omitted and 'the matter is deduced of itself'. But these cases are implicit because the alternative reading is recoverable from the letters and the hearer reconstructs it: in the Mekhilta, from ha-matzot one passes to ha-mitzvot without the 'do not read' being pronounced, because the only difference is a vav. The omitted formula needs a recoverable alternative. Paul's does not have one.
More instructive still is the discussion of principle. For Maimonides, according to Chajes, the derashah of al tiqra is asmakhta: a mnemonic support for an already established norm, and the norm precedes the reading. Chajes objects with the Talmud's own questions: if it were only that, why does the Talmud reply 'but behadrat is written' (Berakhot 30b) and 'but lishlosh is written' (Rosh HaShanah 13b)? The reply shows that the Sages treated the written form as a constraint to which the derashah had to respond. The two positions agree on what matters here: nobody says that the text 'does not say' what it says. Whoever practises al tiqrei starts from what is written, announces that he is reading something else and acknowledges, when reminded, that what is written is different. Paul presents his reading as what the text says (οὐ λέγει) and not as a second, announced reading of the letters, and he does so without the form that he claims to contrast existing.
Two methodological precisions complete the picture. The first: according to the usages that Chajes records and that are confirmed in the editions, אַל תִּקְרֵי is the predominant form in the Babylonian Talmud, whereas אַל תִּקְרָא is more frequent in the tannaitic and Palestinian sources (the midrashim and the Yerushalmi); the label al tiqrei with which Lim designates the technique corresponds, then, to the form of the latest corpus. The second: the classic example of the formula in the Mishnah, that of Avot 6:2, as well as Pirkei deRabbi Eliezer 46:3 and Avot deRabbi Natan 2:3, cannot date the technique before Paul. Chapter 6 of Avot (the baraita of Kinyan Torah) is a late addition: it does not appear in Codex Kaufmann (MS A 50) and, among the manuscripts that have been collated, only Munich MS 95 attests it.
Munich Manuscript 95, folio 1134a, Avot 6:2: אל תקרי
The remaining cases in Chajes are baraitot, Yerushalmi and Bavli, all corpora later than 70. It is not claimed that the procedure did not exist earlier; what is claimed is that Lim, in invoking it as the background to a first-century letter, invokes later literature without demonstrating that the procedure so named was already in use. The analogy that was supposed to legitimate Paul ends up turning against him: the technique most bound to the letter among those used by the Sages always needs letters, and Gal 3:16 has none.
V. Seth and Samuel
Lim adduces two cases of individual זרע: Seth (Gen 4:25) and Samuel (1 Sam 1:11). They must be looked at in their text. The text we are commenting on writes זרע אחד ('one seed'); the Hebrew of Gen 4:25 has זֶרַע אַחֵר, 'another seed', which the LXX renders σπέρμα ἕτερον, and which the verse itself ties to a person with a complement: תַּחַת הֶבֶל, 'in place of Abel'. The case of Seth does not show that זרע, without more, means 'one': it shows that it designates a descendant when the context fixes it with 'another' and 'in place of'. The second case is worse for the thesis: זֶרַע אֲנָשִׁים (1 Sam 1:11) is 'male offspring', a generic with a plural complement, not a numerical singular, and the LXX understands it so (σπέρμα ἀνδρῶν). Neither resembles the promises to Avraham, whose context both belie (section III). The Sages also read an individual into זרע, and how they did so is instructive. On Gen 4:25, 'for God has appointed me another seed' (כִּי שָׁת לִי אֱלֹהִים זֶרַע אַחֵר), a midrash attributed to R. Huna reads: 'another seed': seed that comes from another place; and who is it? The King Mashiach (Bereshit Rabbah 23:5; and, in parallel, Ruth Rabbah). Even here the Sage individualises a זרע, but he does so by hanging on a written word, אַחֵר, 'another', and not on the number of the word. It is exactly the difference that Lim needs to erase: the Sage starts from a mark that is in the text; Paul starts from one that is not.
VI. The διαθήκη that Hebrew does not have
Before arriving at 3:16, Lim finds in 3:15 an echo of Deuteronomy. Paul writes:
ὅμως ἀνθρώπου κεκυρωμένην διαθήκην οὐδεὶς ἀθετεῖ ἢ ἐπιδιατάσσεται ('nevertheless, no one annuls or adds provisions to a human testament that has been ratified'),
– Codex Vaticanus, folio 1490a.
and that, it is said, would evoke Deut 4:2 and 13:1, 'you shall not add nor take away'. The echo is not in the words. The Greek Deuteronomy says οὐ προσθήσετε… καὶ οὐκ ἀφελεῖτε (4:2) and οὐ προσθήσεις… οὐδὲ ἀφελεῖς (13:1), and there is in it neither ἀθετέω nor ἐπιδιατάσσομαι; and in Paul's verse there is no 'take away' either. The Pauline verbs are vocabulary of Hellenistic testamentary law: κυρόω is to ratify a testament, ἀθετέω to annul it and ἐπιδιατάσσομαι, a hapax of the New Testament, to add provisions to one already made, as the lexica record. What Paul evokes is not Deuteronomy, but a scene from a Greek notary's office.
And that scene sustains the argument of 3:15–17 by one word: διαθήκη. In Greek legal usage it is the testament; the LXX chose διαθήκη to render בְּרִית, 'covenant', and Paul plays on both senses within three verses: the irrevocable human testament (3:15) and the covenant with Avraham (3:17). The Hebrew בְּרִית is not a testament (that word is צַוָּאָה), and the Mishnah, to say 'testament', borrows the Greek διαθήκη (דיתיקי). The argument cannot be made in Hebrew. The Letter to the Hebrews plays the same game (ὅπου γὰρ διαθήκη, θάνατον ἀνάγκη φέρεσθαι τοῦ διαθεμένου, Heb 9:16, 'where there is a testament, the death of the testator must be established'), which is to be expected of a tradition that reasons in Greek. In four consecutive verses (3:15–18), the reasoning depends on three Greek words, διαθήκη, ἀθετέω and σπέρμα, whose interplay does not survive retroversion.
Two further observations. The first is an irony: if Paul truly invoked the 'you shall not add nor take away', his own reading would add to the text a restriction, the number 'one', that Genesis does not write. The second is that the analogy starts from a unilateral and irrevocable will, and the בְּרִית of Genesis is not read in that way: it is a relationship with obligations. Gen 17 seals it with circumcision 'throughout their generations', and Lim lists among the promises Gen 26:3–4, whose next verse gives the reason for the covenant:
עֵקֶב אֲשֶׁר שָׁמַע אַבְרָהָם בְּקֹלִי וַיִּשְׁמֹר מִשְׁמַרְתִּי מִצְוֺתַי חֻקּוֹתַי וְתוֹרֹתָי (Gen 26:5, 'because Avraham obeyed My voice and kept My charge, My commandments, My statutes and My teachings').
The promise that Paul declares prior to, and independent of, the Torah comes, in Genesis itself, bound to the keeping of the Torah.
VII. Galatians 3:29 and the two defences
Thirteen verses after 3:16: ος εστιν χς, the one, becomes ει δε ϋμεις χυ αρα του αβρααμ σπερμα εστε ('if you are Christ's, then you are Avraham's seed'), the many.
– Codex Vaticanus, folio 1490c.
The demonstration disappears: if σπέρμα admits the collective of believers, 'to your seed' does not exclude plurality, and if it does not admit it, 3:29 cannot stand. Lim offers two defences.
The first: ancient exegesis admitted several senses of one and the same word, so that the incoherence is seen only by the modern reader. But the multiple senses of the pesharim and of the midrashim are added as layers, whereas what Paul does is to use one sense in order to exclude another ('not… but') and, afterwards, the other in order to include. Moreover, the principle that the Torah has seventy faces (שבעים פנים לתורה) is attested in late sources (Bemidbar Rabbah 13:15) and not in Paul's time; and even then the Sages retained a brake, אין מקרא יוצא מידי פשוטו ('a verse does not go out from its plain sense', Yevamot 24a), and the plain sense of זרע, in each of the promises, is collective. Whoever invokes 'the ancient world' has to show a comparable ancient argument: one that derives an individual referent from a collective singular without any textual mark (a superfluous particle, a written plural, an anomaly). I know of none.
The second: 'corporate solidarity', which would explain how Yeshu can be at once the singular and the many. That category, taken from H. Wheeler Robinson, was thoroughly revised by J. W. Rogerson (JTS 21, 1970), who showed that it does not describe Hebrew thought: it is not a foundation on which to rest anything. And even if it were admitted, it would rescue inclusion (3:29), not exclusion ('not seeds, but one'), which is what 3:16 needs. With σπέρμα as a collective that includes Yeshu and the believers, the only textual support of the identification disappears, and what remains is the assertion ὅς ἐστιν Χ[ριστό]ς, an identification imposed on the text from outside.
VIII. The pesharim
The comparison with Qumran fails by genre and by method. The pesharim are continuous commentaries: they cite a lemma and apply it to the community with the formula פשרו על. Gal 3:16 does not comment on a lemma; it constructs, upon a phrase without variant, an argument about number, and the identification arrives with an ὅς ἐστιν that is not a pesher formula. When the pesharim read otherwise, they play with the letters of the consonantal text or with a known variant: they have a textual basis on which to rest the reading. Paul's alternative, σπέρματα, has none: neither in the Masoretic Hebrew, nor in the Samaritan Pentateuch, nor in the LXX, nor in the scrolls. And what Lim calls literalism in the pesharim, attention to the form of the text, has nothing to do with what Paul does in Gal 3:16, which is the opposite: to separate the letter from its context. The only thing Paul shares with the pesharim is what all ancient exegesis shares: to read the text in favour of the community that reads it. That is neither literalism nor a technique, and it does not distinguish him from Philo, whom nobody calls a literalist.
There is, besides, a difference of authority. The author of a pesher does not found his reading on a datum of grammar, but on a revelation: God made known to the Teacher of Righteousness:
אשר נתן אל ב[לבו בינ]ה לפשור את כול
דברי עבדיו הנביאים
'to whom God gave [in his heart understand]ing to interpret all the words of His servants the prophets' (1QpHab II, 8–9).
His reading stands because it comes from one who received the secret, and it does not need the text to authorise it by its morphology. Paul, in 3:16, does the opposite: he appeals to the form ('it does not say… but') and wants the letter itself to authorise him, and the letter does not.
IX. Paul's best defences, and why they are not enough
The serious defence of Gal 3:16 does not maintain that זרע is grammatically an individual. It maintains, with Hays, Wright, C. J. Collins and T. D. Alexander, that Genesis presents a seed that is concentrated in a lineage, receives the inheritance as a recipient distinct from Avraham, and resolves itself in a royal figure, whom Paul identifies with Christ. It is fitting to give it its best form, because Lim barely touches the texts (he cites 13:15 and 17:8 as mere occurrences of the formula) and because each piece of that defence meets its limit in the same text.
The first piece is Gen 13:15: 'all the land that you see, to you I will give it, and to your seed' (לְךָ אֶתְּנֶנָּה וּלְזַרְעֲךָ עַד עוֹלָם, 'for ever'). In the LXX the phrase coincides almost word for word with the one Paul cites (σοὶ δώσω αὐτὴν καὶ τῷ σπέρματί σου), and the defenders see in it a second recipient in the singular. But 'to you and to your seed' is the formula of a grant to a person and to his own, and the next verse sets its measure: 'if anyone can count the dust of the earth, your seed also shall be counted' (גַּם זַרְעֲךָ יִמָּנֶה). The same noun, in the same context, is counted as the dust is counted.
The second is Gen 17:8, which adds 'after you' (אַחֲרֶיךָ; LXX μετὰ σέ), a formula of succession that the defenders read as a hereditary line that narrows: the covenant passes through Yitzchak and not through Yishmael (17:19–21), 'for in Yitzchak shall seed be called for you' (כִּי בְיִצְחָק יִקָּרֵא לְךָ זָרַע, 21:12; Rom 9:7). But to choose a branch is not to choose an individual: the promise to Yitzchak is, in its turn, that of a multitude (26:4: וְהִרְבֵּיתִי אֶת זַרְעֲךָ כְּכוֹכְבֵי הַשָּׁמָיִם, 'I will multiply your seed as the stars of heaven'), and 17:9–10 continues with 'your seed after you, throughout their generations' and with the circumcision of 'every male'. The formula, moreover, is not exclusive to Avraham: to Ya‘akov it is said, in the same words as 13:15, לְךָ אֶתְּנֶנָּה וּלְזַרְעֶךָ (Gen 28:13), and at once וְהָיָה זַרְעֲךָ כַּעֲפַר הָאָרֶץ (28:14, 'and your seed shall be as the dust of the earth'). It is a mould that is repeated with each patriarch and that in each repetition is filled with multitude. And the selection of Yitzchak (21:12) does not change the number of the word: in the next verse the Torah calls Yishmael 'your seed' (כִּי זַרְעֲךָ הוּא, 21:13).
The third is Gen 22:17, where the seed is the subject and the enemies carry a singular suffix: 'shall possess the gate of his enemies' (וְיִרַשׁ זַרְעֲךָ אֵת שַׁעַר אֹיְבָיו). The defenders see in the singular a person, and add the picture of victory, the 'only son' of 22:2 and the blessing of the nations in 22:18. But the verse opens with the stars and the sand, and the same formula reappears in Gen 24:60, in the blessing of Rivkah: אֲחֹתֵנוּ אַתְּ הֲיִי לְאַלְפֵי רְבָבָה וְיִירַשׁ זַרְעֵךְ אֵת שַׁעַר שֹׂנְאָיו ('our sister, may you become thousands of myriads, and may your seed possess the gate of those who hate it'). The same verb, the same 'gate', the same singular suffix and, in the same sentence, an explicit collective of 'thousands of myriads'. Nobody reads an individual there. A singular suffix with a collective noun is ordinary Hebrew syntax.
The strongest case is another: 2 Sam 7:12, 'I will raise up your seed after you, who shall come forth from your body' (וַהֲקִימֹתִי אֶת זַרְעֲךָ אַחֲרֶיךָ אֲשֶׁר יֵצֵא מִמֵּעֶיךָ), and its messianic reading in the Qumran Florilegium (4Q174 1 I, 10–13), which identifies that seed with 'the Branch of David' (צמח דויד):
[וה]ג֯יד לכ֯ה יהוה כיא בית יבנה לכה והקימותי את זרעכה אחריכה והכינותי את כסא ממלכתו [לעו]ל֯ם֯ אני א֯ה֯י֯ה֯ לוא לאב והוא יהיה לי לבן הואה צמח דויד העומד עם דורש התורה
'[And] YHWH [de]clares to you that He will build you a house, and I will raise up your seed (זרעכה) after you, and I will establish the throne of his kingdom [for ev]er. I will be a father to him, and he shall be a son to Me. He is the Branch of David, who shall arise with the Interpreter of the Torah (Doresh ha-Torah).'
Here there is indeed a זרעכה that an ancient text reads as an individual. But the text of Samuel individualises it by itself: the relative is singular (אֲשֶׁר יֵצֵא) and the following verses continue in the singular ('he shall build a house for My name… I will be a father to him and he shall be a son to Me', 7:13–14). It is the principle that operated in Seth and in Samuel: the context marks the individual. With Avraham the context marks the opposite. It might be objected that Paul's ὅς ἐστιν recalls the formula of identification 'he is…' of the Florilegium; but the Florilegium identifies a singular that the text already brings, and Paul identifies a collective that the text brings as a collective. As far as I can see, no source earlier than Paul applies an individual messianic reading to the seed of Avraham.
There remains the last resort: to speak not of 'corporate solidarity', but of representation. The king represents his people, and Yeshu, a Davidic king, would represent the seed. But to represent is not to exclude. Even granting the category, the phrase 'not to the seeds, but to one' still needs a plural that nobody read, and in 3:29 the believers are not 'represented by' the seed, but are seed (σπέρμα ἐστέ). The defence explains how Paul could reconcile the two verses; it does not explain why the first was formulated as a grammatical argument.
Observe what this defence does: it abandons Paul's argument. Galatians 3:16 does not say 'Genesis concentrates the seed in a lineage'. It says οὐ λέγει· καὶ τοῖς σπέρμασιν, ὡς ἐπὶ πολλῶν, ἀλλ᾽ ὡς ἐφ᾽ ἑνός ('it does not say "and to the seeds", as of many, but as of one'): an assertion about what the word says, about its number, which is the one that Lim calls literalism and compares with al tiqrei. Hence a dilemma that neither position resolves. If Gal 3:16 is an assertion about the letter, it is false, because the letter does not say that. If it is an assertion about the narrative, it is not literalism, and the kinship with Qumran and with the Sages that Lim needs disappears. To sustain Lim's thesis one must keep the first horn; to save Paul, the second. One cannot have both.
X. Three promises in the Greek of Constantinople
Up to here we have measured Paul against the Hebrew and against the LXX. A witness is missing that has not entered the discussion: the Greek that Greek-speaking Jews themselves wrote when they translated the Torah from the Hebrew. It is preserved in the Constantinople Pentateuch, printed in Hebrew characters in 1547 and edited in Greek letters by D. C. Hesseling (Les cinq livres de la Loi, Leiden: Brill, 1897). It is the version of the Romaniote community, which used that language, yevanit, for centuries. Its relationship to the ancient Jewish versions, that of Aquila among them, is disputed; the Greek texts of Jewish use from the Cairo Genizah (de Lange, 1996) show that the practice of translating from Hebrew into Greek was never interrupted. We need not resolve that filiation. What we can observe is how a Jewish Greek behaves when it translates with the Hebrew before it and not with the LXX: where it coincides with it, where it departs from it, and what it does with the three promises on which the argument of Paul is defended (two of them, 13:15 and 17:8, carry the formula he cites; the third, 22:17, is the favourite of the defenders of the individual reading). We transcribe from each the text in Hebrew letters, the text in Greek letters of Hesseling, and our translation.
Genesis 22:17
Hebrew letters (consonantal), page 72:
כי אוטי [אבלויימו] נא [שי] [אבלוישו] קי פליתימו נא פליתינו טי שפורא שו שאן טא אשטרא טו אוראנו קי שאן טון אמו אוש איפי טו כילוש טיש תאלשוש קי נא קלירונומישי אי שפורא שו טו קשטרו טון אוכטרון טו׃
ὅτι εὐλογημὸ νὰ σ' εὐλογήσω καὶ πληθυμὸ νὰ πληθύνω τὴ σπορά σου σὰν τὰ ἄστρα τοῦ οὐρανοῦ καὶ σὰν τὸν ἄμμο ὃς ἰπὶ τὸ χεῖλος τῆς θαλασσοῦς, καὶ νὰ κληρονομήσῃ ἡ σπορά σου τὸ κάστρο τῶν ὀχτρῶν του.
Translation: 'Because with blessing I will bless you and with multiplication I will multiply your seed (spora) like the stars of heaven and like the sand that is upon the lip of the sea; and your seed shall inherit the fortress of his enemies.'
The yevanit opens with ὅτι, which reproduces כִּי; the LXX makes the oath explicit with ἦ μήν, and Heb 6:14 cites the passage with εἰ μήν (ἦ μήν in the textus receptus), whereas the yevanit does not mark the oath and follows the Hebrew כי. The infinitive absolutes בָרֵךְ אֲבָרֶכְךָ and הַרְבָּה אַרְבֶּה are calqued with a cognate noun followed by νά with a subjunctive of future value (εὐλογημὸ νὰ σ' εὐλογήσω, πληθυμὸ νὰ πληθύνω), where the LXX employs the participle (εὐλογῶν εὐλογήσω, πληθύνων πληθυνῶ). For זֶרַע the translator chose σπορά (שפורא, 'sowing, seed') and not the σπέρμα of the LXX. ὃς ἰπὶ calques אֲשֶׁר עַל, where the LXX recasts with τὴν παρὰ τὸ χεῖλος. And שַׁעַר, 'gate', is rendered κάστρο (from castrum, 'fortress'), a metonymy for a fortified place that preserves the singular of the Masoretic Text, whereas the LXX pluralises (τὰς πόλεις, 'the cities') and neither version preserves πύλη. The enemies are ὀχτρῶν του, with the singular possessive of the Hebrew (אֹיְבָיו), as against τῶν ὑπεναντίων of the LXX, which omits it. What is decisive for our subject: in a single verse the singular noun σπορά (שפורא) is 'like the stars' and 'like the sand', and at the same time governs a singular verb and a singular possessive. The translator felt no need to undo that tension, because in his Greek, as in Hebrew, the collective singular is not a problem.
Genesis 17:8
Hebrew letters (consonantal), pages 49–50:
ונתתי קי נא דושו אשין קי טיש שפוראשו קטפוד[ו] שו טין איי טיש קטיקאש שו אולי טין אי[י] טו כנען יא דיאקרטיקשי נאונא קי נא [אימי] אקינון יא תיאוש׃
καὶ νὰ δώσω ἐσὲν καὶ τῆς σπορᾶς σου καταπόδου σου τὴν ἡγῆ τῆς κατοικᾶς σου, | ὅλη τὴν ἡγῆ τοῦ Κενααν γιὰ διακράτηξῃ ναιῶνα καὶ f. 26. νὰ εἶμαι ἐκείνων γιὰ θεός.
Translation: 'And I will give you and your seed (sporas) after you the land of your dwelling, all the land of Kena‘an, for a perpetual possession; and I will be God (Theosh) for them.'
The indirect object of לְךָ וּלְזַרְעֲךָ is expressed with the accusative and the genitive (ἐσὲν… τῆς σπορᾶς σου), because the vernacular no longer uses the dative that the LXX preserves (σοι… τῷ σπέρματί σου); and σπορά (שפוראש) occupies again the place of σπέρμα, the term on which Paul relies in Gal 3:16, which cites precisely the formula 'καὶ τῷ σπέρματί σου'. Both are collective singulars that translate זֶרַע. In τῆς κατοικᾶς σου (מְגֻרֶיךָ, 'your sojournings', from גור, 'to reside as a stranger') the yevanit employs the generic noun κατοικιά, which dilutes the condition of stranger that the LXX preserves with the verb παροικέω (τὴν γῆν ἣν παροικεῖς). For לַאֲחֻזַּת it has διακράτηξῃ ('holding') as against the κατάσχεσιν of the LXX; 'perpetual' is ναιῶνα, probably a contraction of τὸν αἰῶνα, as an adverbial complement, where the LXX uses the adjective αἰώνιον. And the close, ἐκείνων γιὰ θεός (אקינון יא תיאוש), calques לָהֶם לֵאלֹהִים: it reproduces the preposition ל of the predicate with γιά and the dative with the genitive ἐκείνων. The text of the CCAT has only αὐτοῖς θεός, without preposition; other editions, such as that of Brenton, read αὐτοῖς εἰς Θεόν, with an εἰς that does reflect the ל, in line with the yevanit.
Genesis 13:15
Hebrew letters (consonantal), page 40:
כי אוטי אולי איי[ש] אוש אשי תוריש אשין נא טין דושו קי טיש שפוראשו אוש נאונא׃
ὅτι ὅλη ἡγῆς ὃς ἐσὺ θωρεῖς, ἐσὲν νὰ τὴν δώσω καὶ τῆς σπορᾶς σου ὡς ναιῶνα,
Translation: 'Because all the land that you see, to you I will give it, and to your seed (שפוראשו), for ever.'
As in 17:8, the indirect object goes in the accusative (ἐσὲν) and the genitive (τῆς σπορᾶς σου) and not in the datives σοὶ… τῷ σπέρματί σου of the LXX, and σπορά replaces σπέρμα. The formula of the LXX in this verse coincides with the one Paul cites in Gal 3:16, and it is precisely that link which the yevanit does not preserve. The yevanit uses ὅλος (אולי) instead of πᾶς for כָּל, θωρῶ (תוריש, from θεωρέω) instead of ὁράω for רֹאֶה, the invariable relative ὃς as a calque of אֲשֶׁר (the LXX agrees: ἣν) and ὡς ναιῶνα for עַד־עוֹלָם, with ὡς + accusative as against the ἕως τοῦ αἰῶνος with genitive of the LXX.
Editorial note. The brackets mark doubtful readings or restored letters. The Hebrew text is consonantal, although the edition is vocalised, and the filler letters at line-ends are not included. Two readings are by analogy: in 22:17 the final י of the scan is read as the ׃ mark of verse-end, and in 17:8 the stroke shaped like a ו after תיאוש is read in the same way, as confirmed in 13:15. The marks '|' and 'f. 26.' in 17:8 are editorial additions of Hesseling and do not enter the Hebrew text.
What the three texts show together
First, the seed remains a collective singular. In the three promises, with the Hebrew before him, the translator writes σπορά (שפורא) with σου and gives no sign of perceiving a problem of number. In 22:17 that σπορά is at once 'like the stars' and the subject of a singular verb and a singular possessive: it is exactly the tension that Gal 3:16 claims the text cannot bear, and which the Greek of those who read Genesis from the Hebrew bears without effort.
And in 17:8 the collective singular becomes plural before the verse ends. The yevanit closes with ἐκείνων γιὰ θεός, 'God for them'; the LXX, the text from which Paul takes the formula, with καὶ ἔσομαι αὐτοῖς εἰς θεόν, and already in 17:7 it had said εἰς τὰς γενεὰς αὐτῶν, 'throughout their generations'. The same verse that, according to Paul, names one, hands over in mid-line the plural that Paul denies.
Second, it is a control experiment. If Paul's argument lived in the Hebrew, or in the understanding that Jews who read in Greek had of it, it would appear in a Greek made from the Hebrew. It does not appear. What comes out is σπορά (שפורא), and the formula 'καὶ τῷ σπέρματί σου' of 13:15 and 17:8, from which Paul hangs his 'it does not say… but', does not exist. The hook is a property of the LXX: of its lexicon (σπέρμα) and of its morphology (the dative of the singular and of the plural), not of Genesis. Observe moreover that the Pauline contrast is a contrast of datives, τοῖς σπέρμασιν as against τῷ σπέρματι. The Hebrew has an invariable lamed (וּלְזַרְעֲךָ), which does not distinguish number, and the yevanit, which has lost the dative, renders it with genitive and accusative. That proves nothing about the first century, but it shows what it is that Paul manipulates: a distribution of Greek forms, not a property of the text he claims to translate.
Third, the yevanit makes visible two orientations of Jews who read in Greek. The translator of Constantinople places himself at the service of the Hebrew, to the point of violating his own Greek: he calques the infinitive absolute with a cognate noun, אֲשֶׁר עַל with ὃς ἰπὶ, לָהֶם לֵאלֹהִים with ἐκείνων γιὰ θεός, and he reads שַׁעַר as κάστρο without losing the singular. Paul places himself at the service of the Greek word, to the point of violating the Hebrew: he makes a collective say what a plural that does not exist would say. He who translates from the Hebrew translates words; he who reasons from the Greek reasons with them. That is where Paul's reading proves, strictly, defective: not for being ancient or allegorical, but because it demands of the Hebrew what the Hebrew does not have.
There is a grammatical trace that confirms the method. In 22:17 the seed, σπορά, is feminine, but the pronoun that follows is του (ὀχτρῶν του), masculine. A Greek that reasoned in Greek would say της; the translator follows the masculine suffix of אֹיְבָיו (zera‘ is masculine in Hebrew) and lets the Hebrew gender govern the Greek. The Hebrew spelling of the edition (קלירונומישי for κληρονομήσῃ, שפורא for σπορά) confirms moreover that it is heard Greek, living speech, and not the copy of a codex. In Gal 3:16 the inverse movement takes place: the antecedent is neuter (τῷ σπέρματι) and the relative masculine (ὅς), so that Greek agreement bends to the referent that Paul wishes to find. One must not ask more of the yevanit than it says: in 22:17 the verb is also singular (νὰ κληρονομήσῃ ἡ σπορά σου) and it places no mark of an individual; the explicit pluralisation is in 17:8.
Fourth, what we do not claim. The Constantinople Pentateuch is a late witness, and it is not invoked as though it registered what Pharisees or Essenes of the first century thought. It is invoked as a control of method and for an absence. A critic will object that σπορά (שפורא) also has a plural; that is true, and it confirms that Paul's contrast is a play with a Greek word that is not in the text being translated.
XI. The Paul who fits in no classroom
Two readers receive this article with a thesis in their pocket. The Hebrew Roots missionary maintains that Paul was an observant Pharisee, a disciple of Rabban Gamliel HaZaken; the apologist who follows Lim places him among the Qumran pesher authors. Both profiles leave traces, and those traces can be sought.
The disciple of Rabban Gamliel HaZaken. We know of Gamliel from Acts 22:3, not from the letters. Paul never names a teacher, although the maxim attributed to Gamliel himself is עֲשֵׂה לְךָ רַב (Avot 1:16, 'make for yourself a teacher'); his self-portrait is 'Pharisee according to the Torah' (Phil 3:5, κατὰ νόμον Φαρισαῖος), and in Gal 1:22 he says that the communities of Judaea did not know him 'by face' (ἀγνοούμενος τῷ προσώπῳ), which is hard to reconcile with the persecutor trained in Jerusalem painted by Acts. A Sage is defined by what he received and from whom (משה קִבֵּל תורה מסיני, Avot 1:1, 'Moshe received Torah from Sinai'). Paul, in Gal 1:12, defines himself by the opposite: οὐδὲ γὰρ ἐγὼ παρὰ ἀνθρώπου παρέλαβον αὐτὸ οὔτε ἐδιδάχθην ('for neither did I receive it from man, nor was I taught it'). The verb παραλαμβάνω is that of received transmission, the one that corresponds to קִבֵּל, and Paul denies it for his gospel. On how a disciple of that milieu reasons we have direct testimonies: the Sage of Sanhedrin 4:5 starts from a written form (דְּמֵי), the Mishnah coins the plural that it lacks (זַרְעִיּוֹת), and in the gezerah shavah the link is a shared Hebrew word. All the work is done on the consonantal text. In Gal 3:16 there is no Hebrew letter on which to work, no gloss, no etymology and no variant: there is a Greek dative form. (On the rest of the formation attributed to him, see The Disciple of Gamliel Who Never Named a Single Sage.)
The pesher author. A pesher is recognised by a lemma, by the formula פשרו על, by the attribution of the secret to the Teacher of Righteousness (1QpHab II) and by a textual anchor: a letter, a variant, an alternative reading of the consonantal text. Gal 3:16 has, at most, the identification (ὅς ἐστιν Χριστός); it lacks everything else. And its concept of the righteous does not coincide with that of the pesher author, as will be seen in section XIV.
It will be replied that Paul could have known Hebrew, and the contrary is not affirmed. A bilingual reasons in the language of the text that he has in view, and the one he has in view here is the Greek. Whoever had the Hebrew before him could not have written 'it does not say seeds' on זרע; whoever writes that is not reading זרע. If the Hebrew was in view and was ignored, Gal 3:16 is not 'literalism' (which is what Lim affirms), because it ignores the letter; if it was not in view, Paul does not work with the Hebrew text in which the talmid hakham and the pesher author were formed. In both cases, the profile that Lim needs disappears. Whoever adds the two absences obtains a profile: a Jew who reasons on Scripture from the LXX, with the vocabulary of Hellenistic law (κυρόω, ἀθετέω, ἐπιδιατάσσομαι), and without any of the marks that work on the Hebrew leaves in a text, the same ones that Josephus did leave and that we measured in The Greek That Betrays Paul. That profile may be called a 'Jew' of the diaspora, and it can be debated how much of his world was presumptively Jewish; what it cannot be called is talmid hakham or pesher author. A disciple of Rabban Gamliel HaZaken who read Gal 3:16 aloud in the classroom would have had to answer the first question of any fellow student: where is the plural written?
XII. What Lim does not explain
There remains what Lim's reading leaves unexplained. First, why the argument of Gal 3:15–18 rests on three Greek words whose interplay breaks when one returns to the Hebrew: σπέρμα as against σπέρματα, διαθήκη as testament and as covenant, ἀθετέω and ἐπιδιατάσσομαι as notarial verbs. Second, why an author who would have studied Genesis with the Hebrew before him does not hear the collective of זרע, which the text repeats in every promise. Third, why that argument leaves none of the traces that one who works in Hebrew does leave (glosses, etymologies, a Hebrew text of his own): the ones Josephus left, as we measured in The Greek That Betrays Paul, and which are not here. Fourth, why the Greek that Jews wrote from the Hebrew, that of Constantinople, returns in the same promises a collective without friction and none of Paul's words. And even the Hebraists who wished to write Paul in Hebrew, Delitzsch in 1901 and the ecumenical team of the Bible Society in Israel in the 1970s, had to fabricate the plural that Genesis does not have. The most economical explanation is the one Lim rejects: Paul reasons on the Greek word σπέρμα, with the grammar of Greek and with the testamentary law of Greek; and that is why his reasoning can be formulated in Greek and not in Hebrew.
XIII. The biblicist and the method of 'Judaising' Paul
The biblicist does not present Lim as an author who argues: he presents him as a link in a programme. Together with another text of his, the passage of Lim serves to celebrate that Paula Fredriksen and others are spreading the tendency to understand Paul, and the New Testament, 'within Judaism' in his time, and to dismiss those who oppose it as 'pseudo-academic Neanderthalism', fanatics enclosed in a 'bubble of dogmatic biases' (the biblicist's post, n.d.). Those are his words:
Let us contrast them with what we have just seen.
First, authority without reading. The passage that he spreads rests, as we saw, on three 'perhaps' and 'it is possible'; the post offers it as a demolition. Whoever calls dogmatic bias a reluctance before a conjecture should explain why a conjecture is a result.
Second, the confusion of theses. Pages 63–65 of Lim deal with whether there is literalism in the pesharim and in Paul; they say nothing about where Paul studied or whether he did so in a Jewish school. From Lim to 'Paul within Judaism' there is a leap that the post does not justify. And the function of the passage, according to Lim himself, is to exclude from the promise 'the many descendants who observe the law' (Lim, 1997, pp. 63–65): it does not serve to 'Judaise' anyone; it serves for the opposite.
Third, the accumulation of authorities from different disciplines. Lim compares exegetical techniques of Paul and of Qumran. Fredriksen, as far as I can see, is a historian of ancient Christianity whose thesis is that Paul was a diaspora Jew, apostle to the Gentiles, whose Bible was the Greek one (Paul: The Pagans' Apostle, Yale, 2017). Neither of the two maintains what the post needs: that Paul reasoned as a Pharisee formed in the Hebrew text. And if Fredriksen is right in what she maintains, she gives us reason in what is discussed here: the Paul of Gal 3:16 works in Greek.
Fourth, insult in place of philology. In the biblicist's text there is not a word about Gen 13:16, about 15:13, about 17:12 or about 24:60, nor about the fact that זרע lacks a plural in biblical Hebrew. There is, on the other hand, a text of Lim that concedes that, in Genesis, the formula refers collectively to the descendants. The author on whom he relies admits the datum that ruins the singular reading, and the post that cites him calls dogmatic those who read it with the same Hebrew.
Fifth, the use that will be made of it. The frame of a 'Jewish' Paul of Hebrew formation is the one that the Hebrew Roots missionaries, the Netzarites and the so-called Messianic 'Jews' need, whose edifice requires an observant Paul. That is why it matters that the passage of Lim, read in full, does the opposite of what is asked of it.
Sixth, translation as programme. Whoever wishes to see how Paul is 'Judaised' does not need to look among recent scholars: it suffices to open the Hebrew version of Delitzsch, which presents the New Testament with the title page of a Jewish book and in the Hebrew of the Sages, and which, in Gal 3:16, can make Paul 'argue on the letter' only by writing a plural, לִזְרָעֶיךָ, that Genesis does not have (section IV). What that Hebrew passes off as a Jewish background is, in reality, the work of the translator.
XIV. Findings
Eight observations, each verifiable in the text, which the previous study had not put on the table.
The relative that breaks the grammar. In Gal 3:16 the relative is masculine, ὅς ἐστιν Χριστός, although its antecedent, σπέρμα, is neuter: grammatical agreement would call for ὅ. The masculine is an agreement ad sensum: it is governed by the person whom Paul identifies, not by the word whose form he has invoked. Some Western witnesses correct it to ὅ, a scribal reaction to the discord. The argument that begins by appealing to the grammar of the noun ends with a pronoun that obeys another law. The identification does not come out of the form; it is imposed upon it.
The same dative, in Romans, with the opposite effect. In Rom 4:16 the promise is secured 'to all the seed' (παντὶ τῷ σπέρματι), and it is specified: 'not only to that which is of the Torah, but also to that which is of the faith of Avraham' (οὐ τῷ ἐκ τοῦ νόμου μόνον ἀλλὰ καὶ τῷ ἐκ πίστεως Ἀβραάμ). The same singular dative, τῷ σπέρματι, which in Galatians leaves outside the many who observe the Torah, includes here those of the Torah; and in 4:18 Paul cites Gen 15:5, whose context is the stars, with its collective sense: οὕτως ἔσται τὸ σπέρμα σου ('so shall your seed be'). Lim refers to Rom 4:13–18 as a natural complement to Galatians. It is two uses of the same dative in two letters, with contrary effects on the observers of the Torah: there is not here a literal technique, but a rhetorical necessity that changes according to the interlocutor.
Habakkuk 2:4: the same verse, the same genre, the opposite conclusion. Gal 3:11 cites Hab 2:4, ὁ δίκαιος ἐκ πίστεως ζήσεται, in a form without pronoun that coincides neither with the Hebrew (וְצַדִּיק בֶּאֱמוּנָתוֹ יִחְיֶה, 'the righteous shall live by his faithfulness') nor with the LXX (ἐκ πίστεώς μου, 'by my faith'), and uses it to oppose faith to the Torah (ἐν νόμῳ οὐδεὶς δικαιοῦται παρὰ τῷ θεῷ, 'in the law no one is justified before God'). The Pesher Habakkuk, one of the pesharim most often cited alongside Paul, reads the same verse thus (1QpHab VIII, 1–3):
(4Q82 frg. 102,3) וצדיק באמונתו יחיה ]
פשרו על כול עושי התורה בבית יהודה אשר
יצילם אל מבית המשפט בעבור עמלם ואמנתם
במורה הצדק
'[“But the righteous shall live by his faithfulness” (Hab 2:4b; 4Q82 frg. 102,3)] Its interpretation concerns all the doers of the Torah in the House of Judah, whom God will deliver from the House of Judgement on account of their toil and their faithfulness to the Moreh ha-Tzedek (Teacher of Righteousness).'
The same lemma, the same gesture (to apply it to the community) and the opposite relation to the Torah: for the pesher author, the righteous man who lives by his faithfulness is the one who does the Torah; for Paul, the righteous (δίκαιος) who lives by faith is the one who is not justified by it. If Paul shared a method with the pesharim, he would share their reading of this verse. It is the most direct proof of where his loyalty does not lie.
The Greek 'all' and the link that Paul constructs. In Gal 3:10 and 3:13 Paul cites two texts of Deuteronomy. Both carry an 'all' that the Masoretic Text does not have and that the LXX does: in 27:26, ἐπικατάρατος πᾶς ἄνθρωπος… ἐν πᾶσιν τοῖς λόγοις ('cursed is every man… in all the words'), as against the Hebrew 'Cursed be he who does not confirm the words of this Torah' אָרוּר אֲשֶׁר לֹא יָקִים אֶת דִּבְרֵי הַתּוֹרָה הַזֹּאת; in 21:23, κεκατηραμένος ὑπὸ θεοῦ πᾶς κρεμάμενος ἐπὶ ξύλου ('cursed by God is everyone who hangs on a tree'), as against כִּי קִלְלַת אֱלֹהִים תָּלוּי ('for a hanged man is a curse of God'). There is more. To join the two texts, Paul needs them to share a word, as in a gezerah shavah. In Hebrew they do not share it: the first says אָרוּר, and the second, קִלְלַת אֱלֹהִים. In the LXX they do not either: ἐπικατάρατος in 27:26, κεκατηραμένος in 21:23. Paul constructs the link: in 3:13 he writes ἐπικατάρατος πᾶς ὁ κρεμάμενος ἐπὶ ξύλου, substitutes the participle with the adjective of 27:26 and suppresses ὑπὸ θεοῦ. The verbal bond between the two verses is not in the Hebrew or in the LXX: Paul himself produces it by editing the Greek. Whoever practises gezerah shavah in a halakhic school cannot fabricate the link, because the link has to be a word that was already in both verses. Paul does it on the Greek page.
The Aramaic and the plural that the Hebrew does not have. Targum Onkelos, which the synagogal tradition reads alongside the Torah, does not write zar‘akh in the passages that matter here, but bnakh בְּנָךְ, 'your children', and it does so with verbs and pronouns in the plural:
אֲרֵי בָרָכָא אֲבָרְכִנָּךְ וְאַסְגָּאָה אַסְגֵּי יָת בְּנָךְ כְּכוֹכְבֵי שְׁמַיָּא וּכְחָלָא דִּי עַל כֵּיף יַמָּא וְיִרְתוּן בְּנָךְ יָת קִרְוֵי סַנְאֵיהוֹן:
'For I will surely bless you, and I will greatly multiply your children (bnakh) like the stars of heaven and like the sand that is on the shore of the sea; and your children (bnakh) shall inherit the cities of their enemies.'
In 22:17: וירתון בנך ית קרוי סנאיהון ('and your children shall inherit the cities of their enemies'); in 22:18: ויתברכון בדיל בנך כל עממיא דארעא ('and all the nations of the earth shall be blessed on account of your children'); in 17:7: ובין בנך בתרך לדריהון ('and between your children after you, throughout their generations'); in 17:8: ואהוי להון לאלהא ('and I will be to them for God'); in 13:15: ולבנך עד עלם ('and to your children for ever'); in 13:16: ואשוי ית בנך סגיאין כעפרא דארעא… אף בנך לא יתמנון ('and I will make your children many as the dust of the earth… your children also shall not be counted'). Aramaic has, moreover, bar for 'son' in the singular; the translator would have used it had he heard an individual. What Paul affirms the text does not say, 'and to the children', is what the Targum says. Its final redaction is later than Paul and it is not adduced as a witness to what his adversaries thought, but as a methodological control, as proof that a Jewish translation made from the Hebrew wrote the plural without fear. There is one more detail: in 22:17 Onkelos renders שער by קרוי ('cities'), like the LXX (πόλεις) and unlike the yevanit (κάστρο). The coincidence may be explained by a shared reading of 'gate' as city, and the divergence confirms that the yevanit does not depend on the LXX.
The Torah has its own lexicon for saying 'one', and it is not the number of זרע. In Gen 15:4 God specifies to Avraham who will inherit him: לֹא יִירָשְׁךָ זֶה כִּי אִם אֲשֶׁר יֵצֵא מִמֵּעֶיךָ הוּא יִירָשֶׁךָ ('This one shall not be your heir, but one who shall come forth from your body, he shall be your heir'). And in the next verse, to say how many they will be, it no longer uses 'the one who shall come forth' but כֹּה יִהְיֶה זַרְעֶךָ, 'so shall your seed be', before the stars. A single verse separates the individual, marked by אֲשֶׁר יֵצֵא מִמֵּעֶיךָ and by הוּא, from the collective, זַרְעֶךָ. The mark returns in 17:19–21 (the covenant 'with him', אִתּוֹ; 'with Yitzchak', אֶת יִצְחָק), in 22:2 (אֶת בִּנְךָ אֶת יְחִידְךָ, 'your son, your only one': the individual is ben, not zera‘) and in 2 Sam 7:12, where the same formula, אֲשֶׁר יֵצֵא מִמֵּעֶיךָ, individualises the Davidic zera‘. That is the ultimate reason why 4Q174 can read an individual in 2 Sam 7:12 and Paul cannot do so in Gen 13:15: in 4Q174 the mark is written; in Gal 3:16, it is not.
A Greek witness earlier than Paul by almost two centuries. The grandson of Ben Sira, who translates in Alexandria around 132 BCE, summarises the oath of Gen 22:17–18 by saying that God swore that in his seed the nations would be blessed (ἐν σπέρματι αὐτοῦ ἐνευλογηθῆναι ἔθνη), and that He would exalt his seed like the stars and make them inherit from sea to sea (ἀνυψῶσαι τὸ σπέρμα αὐτοῦ καὶ κατακληρονομῆσαι αὐτούς) (Sir 44:21). The σπέρμα of Avraham, in the same Greek as Paul's, is taken up as a plural, αὐτούς, in the same sentence. It is a Greek-speaking Jewish reader who does not hear what Paul hears.
The vocabulary of Avraham at Qumran points to practice. The Damascus Document presents Avraham as a 'friend' of God for keeping His commandments, and Yitzchak and Ya‘akov as those who kept them and 'were written down as friends of God and members of the covenant for ever' (CD III, 2–4). And 4QMMT uses the verb that Gen 15:6 made famous, 'to be reckoned', but ties it to a doing: ונחשבה לך לצדקה בעשותך הישר והטוב לפנו ('and it shall be reckoned to you as righteousness, when you do what is right and good before Him', Miqsat Ma‘aseh ha-Torah, 4Q398 frags. 14–17 ii, 7 / composite text C 31). 4QMMT does not comment on Avraham, and it is not claimed that it does; what it shows is how 'reckoning as righteousness' sounds in a mouth that speaks from the covenant. Paul, in Gal 3:6, takes from Gen 15:6 the verb 'to believe' (ἐπίστευσεν… καὶ ἐλογίσθη αὐτῷ εἰς δικαιοσύνην) and separates it from doing. Together with the Pesher Habakkuk of finding 3, the pattern is the same: the common lemma, and the verb 'to do' on one side, 'to believe' on the other.
XV. For whoever confronts a Hebrew Roots missionary
This material needs nothing more than the texts on the page. The discourse of Hebrew Roots, of the Netzarites and of the so-called Messianic 'Jews' demands a Paul who was an observant Pharisee, a disciple of Rabban Gamliel HaZaken, whose letters Gentile Christendom would have misunderstood. Five questions, each anchored in a text, suffice to put it to the test. Where, in the Hebrew of Gen 13:16, 15:13, 17:12 or 24:60, does the plural 'seeds' appear, or could it appear, which Paul says the text does not have? What disciple of Rabban Gamliel HaZaken reasons from a form that no text writes, when the Sages who do reason from the letter (Sanhedrin 4:5) write the plural they need? If Gal 3:16 serves, according to Lim himself, to leave out of the promise 'the many descendants who observe the law', how is that Paul a defender of observance? If Rom 4:16 includes in the same seed those of the Law, which of the two passages is that of the Pharisee? And if the argument of 3:15–18 is sustained on διαθήκη, ἀθετέω and σπέρμα, in what language did he think it? None of the five has an answer within the frame of a Paul of Hebrew formation. All five have it in a Paul who reasons in Greek, and that Paul, whatever his place in the history of ideas, belongs neither to the classroom of Rabban Gamliel HaZaken nor to the desert of Qumran.
Closing
Lim presented Galatians 3:16 as the best example of Paul's literalism and of his kinship with Qumran and with the rabbis. It is, but of the opposite. His own text concedes what ruins it: that in Genesis 'and to your seed' refers 'collectively' to the descendants, and that Paul goes 'against the more natural interpretation'. A reading that acknowledges going against the natural sense of the text cannot be called literal. Paul himself classifies it as allegory (Gal 4:24), and his support is not in Genesis, but in three Greek words, σπέρμα, διαθήκη and ἀθετέω, whose interplay does not survive the Hebrew.
Every link of the defence gives way when examined. Al tiqrei requires an alternative consonantal text, and Paul opposes a plural that no text brings. The pesharim individualise when the text has already done so ('the one who shall come forth from your body', 2 Sam 7:12), and Gal 3:16 individualises where the text counts stars and dust. Corporate solidarity, which Rogerson dismantled, does not rescue exclusion, which is the only thing Paul needs. And the most serious defence of Paul can be sustained only by changing the argument: either Gal 3:16 speaks of the letter and is false, or it speaks of the narrative and is no longer literalism. Besides, Paul contradicts himself: thirteen verses later the seed are the many (3:29), and in Romans 4:16 the same singular dative includes those of the Torah whom in Galatians it leaves out. It is not a technique, it is a rhetorical necessity that changes with the interlocutor.
The witness that Lim does not even name is missing. The Greek-speaking Jews who rendered the Torah from the Hebrew, in Constantinople, return the three promises with a collective that asks no permission: σπορά and not σπέρμα, without the pair that sustains the argument, and in 17:8 with 'God for them'. He who translates from the Hebrew translates words; he who reasons from the Greek reasons with them. That text is not invoked as though it registered what Pharisees or Essenes of the first century thought, but as a control of method: where the Hebrew governs, Paul's contrast does not form.
Hence the profile that emerges from the text. A disciple of Rabban Gamliel HaZaken would have had to answer the first question of any fellow student, where is the plural written?, and the Sages who do reason from the letter write the plural they need (Sanhedrin 4:5). A pesher author would have had his lemma, his formula and a mark in the consonantal text. Gal 3:16 has none of this. It has a Greek dative, Greek testamentary law and a purpose that Lim himself declares: that the promise be fulfilled in the single heir 'and not in the many descendants who observe the law'. An exegesis that serves to remove from the promise those who keep the Torah is not that of a teacher of the Torah, however meticulous his citation.
What remains is what the text says: a man who reasons about σπέρμα without hearing zera‘. The Hebrew did not hear it, for it writes it as a collective in every promise. The Jews who put it into Greek from the Hebrew did not hear it. Only one who reads the Bible that already came in Greek, and reasons with its grammar, hears it.
Minimal bibliography: T. H. Lim, Holy Scripture in the Qumran Commentaries and Pauline Letters (Oxford: Clarendon, 1997), 63–65; J. M. G. Barclay, 'Mirror-Reading a Polemical Letter: Galatians as a Test Case', JSNT 31 (1987): 73–93; J. W. Rogerson, 'The Hebrew Conception of Corporate Personality: A Re-Examination', JTS 21 (1970): 1–16; W. Bauer, F. W. Danker et al., A Greek-English Lexicon of the New Testament (BDAG), s.v. ἀθετέω, διαθήκη, ἐπιδιατάσσομαι; A. Rahlfs and R. Hanhart, Septuaginta; D. C. Hesseling, Les cinq livres de la Loi (Le Pentateuque) (Leiden: Brill, 1897); N. R. M. de Lange, Greek Jewish Texts from the Cairo Genizah (Tübingen: Mohr Siebeck, 1996); P. Fredriksen, Paul: The Pagans' Apostle (New Haven: Yale University Press, 2017); Mishnah, Sanhedrin 4:5; Avot 1:16 and 6:2; Bemidbar Rabbah 13:15; Babylonian Talmud, Yevamot 24a; 1QpHab (Pesher Habakkuk) VII–VIII; 4Q174 (Florilegium). J. Ziegler (ed.), Sapientia Iesu Filii Sirach (Göttingen, 1965); A. Sperber (ed.), The Bible in Aramaic, vol. 1 (Leiden: Brill, 1959); E. Qimron and J. Strugnell, Discoveries in the Judaean Desert X (Oxford, 1994). F. Delitzsch, ספרי הברית החדשה / Sifrei ha-Berit ha-Ḥadashah, translated from Greek into Hebrew, 11th edn (1892), England; Ha-Berit ha-Ḥadashah [title as in your copy] (Jerusalem: The Bible Society in Israel), version by an ecumenical team (J. Atzmon, Y. Elihai, B. Lindsay, G. Grossman, M. and Y. Ronen, Y. de Waard), Gal 3:16, p. 241 [year of the edition consulted to be completed].














