Gesenius' Hebrew Grammar/88. Of the Dual
a 1. The dual is a further indication of number, which originated in early times. In Hebrew, however, it is almost exclusively used to denote those objects which naturally occur in pairs (see e). The dual termination is never found in adjectives, verbs, or pronouns. In the noun it is indicated in both genders by the termination ־ַ֫ יִם appended to the ground-form, e.g. יָדַ֫יִם both hands, יוֹמַ֫יִם two days. In the feminine the dual termination is always added to the old ending ath (instead of ־ָה), but necessarily with ā (since it is in an open syllable before the tone), thus ־ָתַ֫יִם, e.g. שָׂפָה lip, שְׂפָתַ֫יִם both lips. From a feminine with the ending ־֫ ־ֶת, e.g. נְח֫שֶׁת (from neḥušt) the dual is formed like נְחֻשְׁתַּ֫יִם double fetters.
b With nouns which in the singular have not a feminine ending, the dual termination is likewise really added to the ground-form; but the latter generally undergoes certain changes in consequence of the shifting of the tone, e.g. כָּנָף wing (ground-form kănăph), dual כְּנָפַ֫יִם, the first ă becoming Šewâ, since it no longer stands before the tone, and the second ă being lengthened before the new tone-syllable. In 1 K 1624, 2 K 523b the form כִּכְּרַ֫יִם (which should be כִּכָּרַ֫יִם) evidently merely points to the constr. st. כִּכְּרֵי, which would be expected before כֶּ֫סֶף; cf. כִּכָּרָ֑יִם in 2 K 523 a, and on the syntax see §131d. In the segholate forms (§84aa) the dual ending is mostly added to the ground-form, e.g. רֶ֫גֶל foot (ground-form răgl), dual רַגְלַ֫יִם; cf., however, קְרָנַ֫יִם (only in the book of Daniel), as well as קַרְנַ֫יִם from קֶ֫רֶן horn, and לְחָיַ֫יִם from לְחִי cheek (as if from the plurals קְרָנוֹת, לִחָיִם).—A feminine dual of an adjective used substantivally occurs in עֲצַלְתַּ֫יִם a sluggish pair (of hands) Ec 1018 from the sing. עָצֵל.
c Rem. 1. Certain place-names were formerly reckoned as dual-forms (so in earlier editions of this Grammar, and still in König’s Lehrgebäude, ii. 437), viz.— (a) those in ־ַ֫ יִן and ־ָן, e.g. דֹּתַ֫יִן Gn 3717a (locative דֹּתָ֑יְּנָה, but in 17b דֹּתָ֑ן), and דֹּתָן 2 K 613; קַרְתָּן Jos 2132, identical with קִרְיָתַ֫יִם in 1 Ch 661 (cf. also the Moabite names of towns in the Mêša‛ inscription, line 10 קריתן = Hebrew קִרְיָתַ֫יִם; line 30 בת דבלתן = בֵּית דִּבְלָתַ֫יִם Jer 4822; lines 31, 32 חורנן=חֹרוֹנַ֫יִם Is 155, &c.); (b) in ־ָם, Jos 1534 הָֽעֵינָם ( = עֵינַ֫יִם Gn 3821). The view that ־ָן and ־ָם arise from a contraction of the dual terminations ־ַ֫ יִן (as in Western Aramaic, cf. also nom. âni, accus. aini, of the dual in Arabic) and ־ַ֫ יִם seemed to be supported by the Mêša‛; inscription, where we find (line 20) מאתן two hundred = מָאתַ֫יִן, Hebrew מָאתַ֫יִם. But in many of these supposed duals either a dual sense cannot be detected at all, or it does not agree at any rate with the nature of the Semitic dual, as found elsewhere. Hence it can hardly be doubted that ־ַ֫ יִן and ־ַ֫ יִם in these place-names only arise from a subsequent expansion of the terminations ־ָן and ־ָם: so Wellhausen, Jahrbücher für Deutsche Theologie, xxi. 433; Philippi, ZDMG. xxxii. 65 f.; Barth, Nominalbildung, p. 319, note 5; Strack, Kommentar zur Genesis, p. 135. The strongest argument in favour of this opinion is that we have a clear case of such an expansion in the Qerê perpetuum (§17c) יְרֽוּשָׁלַ֫יִם for יְרֽוּשָׂלֵם (so, according to Strack, even in old MSS. of the Mišna; cf. Urusalim in the Tel-el-Amarna tablets, and the Aramaic form יְרֽוּשְׁלֵם): similarly in the Aramaic שָֽׁמְרַ֫יִן = שָֽׁמְרָן for the Hebrew שֹֽׁמְרוֹן Samaria.—We may add to this list אֶפְרַ֫יִם, נַֽהְַרַ֫יִם the river country (in the Tel-el-Amarna letters nârima, na’rima), מִצְרַ֫יִם Egypt, Phoenician מצרם; also the words denoting time, צָֽהֳרַ֫יִם midday (Mêša‛ inscription, line 15 צהרם), and perhaps עַרְבַּ֫יִם in the evening, if the regular expression בֵּין־הָֽעַרְבַּ֫יִם Ex 126, 1612, &c., is only due to mistaking עַרְבַּ֫יִם for a dual: LXX πρὸς ἑσπέραν, τὸ δειλινόν, ὀψέ and only in Lv 235 ἀνὰ μέσον τῶν ἑσπερινῶν. The Arabs also say el ‛išâ’ân, the two evenings, cf. Kuhn’s Literaturblatt, iii. 48.
d 2. Only apparently dual-forms (but really plural) are the words מַ֫יִם water and שָׁמַ֫יִם heaven. According to P. Haupt in SBOT. (critical notes on Isaiah, p. 157, line 18 ff.), they are to be derived from the old plural forms (found in Assyrian) mâmi, šamâmi, whence the Hebr. מים, שמים arose by inversion of the i mâmi, mâimi, maim. It is simpler, however, to suppose that the primitive singulars may and šamay, when they took the plural of extension (§124b), kept the tone on the ay, thus causing the îm (which otherwise always has the tone, §87a) to be shortened to im. Cf. the analogous formations, Arab. tarḍaina, 2nd fem. sing. imperf. of a verb ל״י, for tarḍay + îna, corresponding to taqtulîna in the strong verb; also bibl.-Aram. בָּנַיִ֫ן the abs. st. plur. of the ptcp. Qal of (ל״י) בְּנָה, which otherwise always ends in în with the tone, e.g. in the ptcp. Qal of the strong verb, דָּֽבְחִין sacrificing.
e 2. The use of the dual in Hebrew is confined, except in the numerals 2, 12, 200, &c. (see §97), practically to those objects which are by nature or art always found in pairs, especially to the double members of the body (but not necessarily so, cf. זְרֹעִים and זְרֹעוֹת arms, never in the dual), e.g. יָדַ֫יִם both hands, אָזְנַ֫יִם both ears, שִׁנַּ֫יִם teeth (of both rows), also נַֽעֲלַ֫יִם a pair of sandals, מֹֽאזְנַ֫יִם a pair of scales, Lat. bilanx, &c.; or things which are at least thought of as forming a pair, e.g. יוֹמַ֫יִם two (successive) days, Lat. biduum; שְׁבֻעַיִ֫ם two weeks; שְׁנָתַ֫יִם two years (in succession), Lat. biennium; אַמָּתַ֫יִם two cubits.
f In the former case the dual may be used for a plural, either indefinite or defined by a numeral, where it is thought of in a double arrangement, e.g. אַרְבַּע רַגְלָ֑יִם four feet, Lv 1123; שֵׁשׁ כְּנָפַ֫יִם six wings (i.e. three pairs), Is 62, Ez 16; even שִׁבְעָה עֵינַ֫יִם seven eyes, Zc 39, כָּל־בִּרְכַּ֫יִם all knees, Ez 717; כָּל־יָדַ֫יִם all hands, Ez 2112; מְצִלְתַּ֫יִם cymbals, Ezr 310; שְׁפַתַּ֫יִם double-hooks, Ez 4043.—To express a certain emphasis the numeral two is used with the dual, as in Ju 1628, Am 312.—See some other remarks on the use of the dual in §87o and s.
g It is not impossible that Hebrew at an earlier period made a more extensive and freer use of the dual, and that the restrictions and limitations of its use, mentioned above, belong to a relatively later phase of development. The Arabic literary language forms the dual in the noun, pronoun, and verb, almost as extensively as the Sanskrit or Greek; but in modern Arabic it has almost entirely disappeared in the verb, pronoun, and adjective. The Syriac has preserved it only in a few stereotyped forms, with which such duals as the Latin duo, ambo, octo may be compared. In the same way, the dual of the Sanskrit is lost in the modern Indian languages, and its full use in Old Slavonic has been restricted later, e.g. in Bohemian, just as in Hebrew, to pairs, such as hands, feet, eyes, ears. On the Germanic dual, see Grimm’s Gramm., 2nd ed., i. p. 814.