On the connective forms שְׁבַע, תְּשַׁע, cf. the analogous forms in § 93 h.
[c] The other Semitic languages also exhibit the same peculiarity in the external differentiation of the numerals from 3 to 10 as regards gender. The fem. form of the numeral abstracts is only rarely found in connexion with feminine nouns,[1] e.g. שְׁל֫שֶׁת נָשִׁים Gn 713, 1 S 103, Jb 14, Ez 72 Keth.; probably also Jos 1711, where we should read with Dillmann שׁ׳ הַגָּפּוֹת. In apposition, Zc 39, 42, cf. Jer 3623. From what was said above, under a, it follows that these cases are not a return to original usage, but only an intrusion of the form used before masculines into the sphere of the feminine. Conversely in Gn 3824 שְׁלשׁ חֳדָשִׁים (but in the Samaritan שְׁל֫שֶׁת).—For שִׁבְעָה seven, there occurs in Jb 4213 the strange form שִׁבְעָ֫נָה, according to Ewald [Ausführl, Lehrb.8, § 269 b] an old feminine substantive (German ein Siebend, a set of seven), but more probably a scribal error.
[d] 2. The numerals from 11 to 19 are formed by placing the units, without the copula, before the number ten (in the form עָשָׂר masc., עֶשְׂרֵה fem.), but without the two words being joined into one. As was said above, under a, and as is proved by the use of אַחַד, אַחַת in the numeral 11, the feminine numerals from 13 to 19 are to be regarded as construct forms in a genitive connexion. The connective forms of the masculine abstracts, like שְׁל֫שֶׁת, &c., are not admitted in combination with עָשָׂר, since they are merely in apposition, and not in a genitive relation (see the rare exceptions at the end of e). On the other hand שְׁנֵי and שְׁתֵּי in the numeral 12 are undoubtedly true constructs, like אַחַד and the fem. numerals 13–19. But instead of שְׁנֵי (Ex 2821, Jos 312 and four other places) and שְׁתֵּי (Jos 48 and three times in Ezek.), we generally find שְׁנֵים and שְׁתֵּים. Two explanations have been given of these forms: (1) that the Kethîbh really intends שְׁנַ֫יִם, שְׁתַּ֫יִם, in the absol. st., which was first introduced in the case of שְׁנַ֫יִם, on the analogy of עֲשָׂרָה, &c., and then extended to שְׁתַּ֫יִם; the Masora, however, required שְׁנֵי, שְׁתֵּי (but see below), and therefore pointed שְׁנֵים, שׁתֵּים as a Qerê perpetuum (see § 17).—(2) that the absolute forms שְׁנֵ֫יִם, שְׁתַּ֫יִם (introduced on the analogy of שְׁלשָׁה, &c.) were contracted to שְׁנֵים, שְׁתֵּים to facilitate the pronunciation of the duals when closely
