A Figurine with a Possible Early Aramaic Inscription

Abstract

This article deals with an inscribed amulet in the form of an animal that was acquired by the British Museum in 1883 with the claim that its provenience was Tartus, opposite the ancient Phoenician city of Arwad. The five graphemes of the inscription, trztn, are interpreted as written in early Aramaic script, dated to the 9th or 8th centuries BCE. Most plausibly the graphemes constitute a flora-derived personal name, Tirzatan, likely that of the onetime owner of the amulet. Strengths and weaknesses of other interpretative options are also considered.

Keywords:

• Aramaic
• Northwest Semitic
• Epigraphy
• Paleography
• Onomastics
• Amulet
• Tartus

Acknowledgments

I thank James Fraser and G. Nathan Harrison for arranging and overseeing my visit to the British Museum, and Andrew Burlingame, Jessie DeGrado, and Dennis Pardee for reading drafts of this paper.

Notes

1 The museum’s only other acquisition on that day was an Egyptian Old Kingdom copper knife—museum number EA12277 and registration number 83-06-27,1—identified as having also been purchased from Reverend Greville John Chester. Chester is known to have sold other items that he identified as coming from Ṭarṭūs to the British Museum, e.g., a pyramidal glass pendant (EA90486), a chalcedony stamp-seal (BM 102869) and a satyr amulet (BM 102873).

2 Modern Ṭarṭūs (سوطرط) was known in the Hellenistic period as Ανταραδος (< *ἀντί Αραδος, ‘opposite Arwad’) and is presumed to have functioned as the mainland partner of Arwad. Elayi 2015 is a history of Arwad; comparatively little is known about Antarados.

3 I have recently made similar comments regarding a bronze basin allegedly from Cumae and apparently incised with a Phoenician inscription (Richey 2017: 220).

4 Bovid figurines in clay come from Byblos (no. 7892; Dunand 1954: 153, Pl. 55) and Tell Halaf (von Oppenheim 1931: Pl. 56.8; Hrouda 1962: Pl. 20 no. 181). The latter has longer, stylized ears and has been identified as both a Stier(?) (von Oppenheim 1931: 188, Pl. 56) and an Esel (Hrouda 1962: 17). A further figurine from Byblos (no. 3580; Dunand 1939: 245, Pl. 56) went unidentified by its excavator and is also perhaps a donkey.

5 For the representation of dogs in ancient Near Eastern iconography, see especially Haussperger 1994, with a focus on cylinder seals.

6 Because the figurine includes no non-paleographic data that would allow restriction of date and location of manufacture, several script traditions and periods are theoretically possible. Alternative hypotheses falter on the absence of comparanda for one or another sign, e.g., late Neo-Punic has parallels for the cross-shaped sign (as ˀ) and r, but Z-shaped z’s are very rare in this corpus (e.g., Peckham 1968: 203–205), and n is nearly always a simple vertical (ibid.: 213).

7 There are at least four alternatives not described in detail below. First, trztn might be a nominal or verbal form of the Aramaic root √trz, ‘to be fulfilled, [D-stem] to fulfill.’ This verb is attested primarily in Syriac, e.g., ܐܬܪܒܠ ܗܬܙܪܬ ‘it filled up the egg’ (S. Isaaci Antiocheni II, ed. Bickell 1877: 236.281). But the connotations of this verb are usually negative and tend towards ‘bursting’(Brockelmann 1928: 834, ed. Sokoloff 2009: 1665–66). Second, trztn might be a personal name including a divine name from Hattic (Taru) or Hittite/Luwian (Tarḫunza). There are, however, phonological problems with the latter and no way to understand at least the last two graphemes -tn in either case. Third, the inscription might consist of two lexemes, tr ztn (as in the second main option considered below), with the first lexeme being Aramaic tôr ‘bull’ and the second lexeme ztn being Aramaic zêtīn ‘olives’. The Syrian storm-god Adad is often depicted as a bull and is sometimes connected with plants (Schwemer 2001: 686–687), but not olives in particular. Fourth and finally, one could divide the inscription into three lexemes, but this yields only very unlikely analyses. Autopsy confirms that there is no further inscription elsewhere on the figurine.

8 The best summaries may be found in Goldingay and Payne 2006: 354–55 and Koehler and Baumgartner 1994–2000: 1792; see also Brown, Driver and Briggs 1906: 1076.

9 Aramaic names formed thus are documented and/or discussed in, e.g., Kornfeld 1978: 20; Abbadi 1983: 181; Negev 1991: 169–179. Examples include qlqln ‘cassia’ (Kornfeld 1978: 70) or, in

biblical Hebrew, רמָ תּ ‘palm-tree’ (Gen 38, etc.).

10 This nominal pattern does not occur in Old (Degen 1969: 50) or Egyptian Achaemenid Aramaic (Muraoka and Porten 1998: 82). The fact that closely related languages—e.g., Hebrew with ןתָ יָ וְ ִל ‘Leviathan’ and ןתָּשׁ חֻ נְ ‘Nehushtan’ (Bauer and Leander 1922: 500)—have only a few such forms may support the conclusion that such concatenations became common only later in the histories of individual Northwest Semitic dialects.

11 Syriac and Mandaic occurrences of *–atān(-) are catalogued by Nöldeke (1880: 72; 1875: 139).

12 The proto-Semitic noun *θawr- is reflected in the cognate set that includes Hebrew רוֹשׁ šor, Ugaritic ṯr (ṯôr-), and Arabic روْ ث^َ ṯawr-. As is well-known, a number of Indo-European languages have phonologically and semantically similar lexemes that are almost certainly related, e.g., Latin taurus, Greek ταῦρος, German stior > Stier and English stēor > steer.

13 Attestation is summarized in Hoftijzer and Jongeling 1995: 1118, to which add Katumuwa l. 3 šwr (θawr) ‘bull’ (Pardee 2009: 61) and Bukān l. 5 šwrh (θawrā) ‘cow’ (Lemaire 1998: 22–23).

14 Nevertheless, the noun continues to be spelled with w in the Achaemenid Aramaic texts from Egypt—in a single occurrence at Cowley 1923 no. 33:10 (Porten and Yardeni 1986–1999: A4.10): wqn twr ˁnz mqlw [l]ˀ ytˁbd tmh “and sheep, bull, and goat are [n]ot sacrificed as a burnt-offering there”—and Bactria (Naveh and Shaked 2012 A1:8 [353 BCE]; C1:5 ,7 [pl. twrn]; 330 BCE).

15 I have recently discussed Old Aramaic reflexes of *θ and *ð in Richey 2020: 22–25. For attestations and summaries of this apparent rule, see Degen 1969: 35, 43; Segert 1975: 92; Martínez Borobio 2003: 38–39; Fales and Grassi 2016: 41–43.

16 These are (1.) Sefire (ca. 725 BCE) IA:32 btn (batn) ‘snake’ (< *baθn–) (e.g., Fitzmyer 1995: 89, 120, 188); (2.) Sefire IC:24 yrt (yirat) “he will inherit” (G PC 3.m.s. *√yrθ; see e.g. Fitzmyer 1995: 89, 120, 188). This orthography is possibly the result of dissimilation across word boundaries, i.e., yrt šr/[š]h represents here yirat šurših < *yiraθ šurših (so Degen 1969: 43). (3.) Aššur Ostracon (ca. 650 BCE) 11 yhtb (yVhtīb) ‘he will (cause to) return’ (C PC 3.m.s. *√θwb; Fitzmyer 1995: 188). The further example tnn (Bukān 8; see Lemaire 1998: 23 n. 43) claimed by, e.g., Gzella (2015: 92) is unlikely, since this and Syriac ܐܢ_ܵ ܢ_ܵ ܬ ‘smoke’ are not likely cognates of Hebrew ןשׁ ע ‘smoke.’

17 Its terminus ante quem is the reign of Sargon II (721–705 BCE) The weight is published in Al-Rawi 2008: 127–30 and Fig. 15i.

18 Most exceptions involve the reflex of the lexeme *θiql- ‘shekel’ written šql; this is generally recognized as a lexically-specific archaic orthography (Folmer 1995: 70–74; Muraoka and Porten 1998: 7–8).

19 E.g. in the Zincirli Hadad inscription (Donner and Röllig 1966–69: 214): (1.) ymy (yōmay[?]) ‘my days’ (ll. 9, 10, 12; Tropper 1993: 67); and (2.) mšb (mōθab[?]) ‘throne’ (l. 8; Tropper 1993: 66, 186); and (3.) in Nerab Stele 2:10 (Donner and Röllig 1966–69: 225): mmth (mamōtatih[?] < *mamawtati–hu) ‘his death’, but the morphological analysis is hardly certain (compare Jewish Palestinian Aramaic אתוֹממְ ?; see Jastrow 1903: 794b).

20 For this assimilation *nC > CC in Old Aramaic, see Degen 1969: 39–41; Segert 1975: 112–13; Martínez Borobio 2003: 41–42; Fales and Grassi 2016: 43; DeGrado and Richey 2019: 38–39.

21 For this orthography (later –nˀ) of the first common singular genitive pronominal suffix, see Fales and Grassi 2016: 49; Martínez Borobio 2003: 52; Segert 1975: 173; Degen 1969: 55–56. An alternative involves interpreting this -n as an additional nominal morpheme, but the co-occurrence of the feminine singular morpheme -(a)t and the nominal morpheme –ān is—as noted in nn. 10–11, above—rare in older Semitic in general and earlier Aramaic in particular.

22 For lexical documentation see Brockelmann 1928: 192b, ed. Sokoloff 2009: 378b; Payne-Smith 1879: 1103a.

23 These are all letters from the House of ˀUrtēnu: RS 94.2284:29 (Dietrich, Loretz, and Sanmartín 2013: 2.87; Bordreuil and Pardee 2012: 180–86); RS 94.2383+94.2619:8–9 (Dietrich, Loretz, and Sanmartín 2013: 2.104; Bordreuil and Pardee 2012: 190–92); and RS 94.2479:19 (Dietrich, Loretz, and Sanmartín 2013: 2.89; Bordreuil and Pardee 2012: 163–66). Of course, orthographic identity does not require lexical identity, and alternative interpretations that do not incorporate a root √z-n ‘to provide’ are encountered in various scholarly treatments (e.g., Garbini 2014: 84 n. 7).

24 Here both verbal and nominal forms of those roots listed in von Soden (1981: 1509–10) and CAD (Z [1961]: 41–45) as zanānu A/I ‘to rain’ and zanānu B/II ‘to provide’ are considered to be relevant, since they either represent the same root—with the former as a semantic specification of the latter—or are likely to have influenced one another semantically as homonyms.