Tatami: The Enigmatic Toponym of Western Judah, and Use of Suffixes in Dating Toponyms

ABSTRACT

Sometimes, all that remains from the history of a place is its name. Thus, toponyms may provide valuable information. Moving backward in time, we trace the ancient history of the Arab/Crusader village Bēt ˁAṭab in western Judea, with remains from Middle-Bronze II (MB II), Iron III and Hellenistic periods. In Roman times it was Eusebius’ Ēnadab, ‘Spring of ˁaṭab’, identified near the spring of this name mentioned in the Survey of Western Palestine. While missing from Masoretic Biblical town lists, Tatami of the Septuagint derives by dropping the b- from reconstructed batˁaṭami, traced to Northwest Semitic languages exhibiting the ay>ā diphthong contraction. The MB II final -i is analysed as an enclitic -mi, found also in the Execration Text (ET) Shechem, and as enclitic -ma in personal names from Alalaḫ VII. The ‘prince’ of nearby ET ‘Ashnah (=Beth Shemesh) had an identical name as later Niqmi-epuh, King of Yamḫad, suggesting settlers from Yamḫad/Alalaḫ during the early 2nd millennium bce. In Bronze Age Bēt-X toponyms X was a deity. Thus, batˁaṭami = ‘place of the vulture-goddess shrine’. This, and additional evidence, suggest that Egyptian king Senwosret III could have set a shrine to goddess Nekhbet on Bēt ˁAṭab’s summit during his Shechem campaign.

KEYWORDS:

• Alalaḫ
• Bēt ˁAṭab
• case endings
• Nekhbet
• Onomasticon
• Septuagint
• southern Levant
• toponyms

Introduction

Languages evolve rapidly, so that today’s English speakers find it difficult to understand the Shakespearean language. Rapid language changes may reflect periods of prominent social, technological and material changes, such as the transition from hunting to agriculture (Agmon 2010; Agmon and Bloch 2013). However, certain linguistic niches are more change-resistant, for example frequently used words (Pagel et al. 2007). One of the slower evolving semantic fields is placenames (toponyms), perhaps because they are frequently used by the inhabitants. ‘Place names are a window into the history and characteristics of a country. Their names reflect the migrations of peoples, their religious and cultural traditions, local languages, conquests and fortifications long since disappeared’ (Everett-Heath 2000).

In the Levant, the population and/or their religion and language have been replaced at least thrice in 4000 years: From pagan Canaanites to monotheistic, Hebrew speaking Israelites, then to Aramaic/Greek speaking people, and then to a Muslim, Arabic speaking populace. Yet Levantine toponyms survived these upheavals. ‘What surprised western scholars and explorers the most was the amazing degree to which biblical names were still preserved in the Arabic toponymy of Palestine’ (Rainey 1978, 7). In a similar vein, Indo-European speaking Philistine settlers of the Late Bronze (LB) southern Levant retained the West-Semitic names of the Canaanite towns they inherited (Shai 2009).

Many Iron-Age biblical settlements originated in the Bronze Age. As an example, take bēt šemeš (as a linguist would write), or Beth Shemesh (as an archaeologist might spell, š = sh).¹ This border-town between Judah, Philistia and Canaan (Bunimovitz and Lederman 2009), apparently centred at some period around the worship of the Canaanite sun-deity Šapaš/Šamaš. We know this primarily because its name means ‘House [of] Šamaš’, where ‘house’ here stands for a temple dedicated to the named god (Aharoni 1979; Elitzur 2013; Isserlin 1957; Rainey 1978).² While the town came under Israelite control during parts of the Iron Age, it has never occurred to them to change its pagan name. In fact, it was even assigned a religious role as one of the designated priestly towns (Josh 21:16). Today’s ultra-orthodox populace is equally undisturbed by (and probably unaware of) the connotation of their town-name. Although the biblical town was demolished during the Assyrian/Babylonian conquests c. 2500 years ago, its name was preserved in the near-by spring, ˁēn šams, ‘Spring [of] Šamaš’ (Robinson and Smith 1841, II, 339) [where ˁ is the Semitic consonant ˁayn].

This work focuses on a deserted Arab village, Bēt ˁAṭāb, in the mountains above bēt šemeš, whose early history is presently unknown. Toponym use thus becomes indispensable. Travelling back in time, from the Crusader period to the Roman era (Eusebius’ Onomasticon) and the Iron Age (the Septuagint Book of Joshua), allows one to reconstruct its ancient name, bēt ˁāṭām, which is the enigmatic village Tatam(i) appearing in the Septuagint. Proceeding further into the Middle Bronze Age (MBA) is particularly challenging. Using archaic features of Semitic languages surviving in some toponyms (Layton 1990), such as the suffixes -i, -im, -am, -mi, and the -ay diphthong contraction, helps in dating this toponym and determining the origins of its first inhabitants. This analysis shows that the village was established by (Amorite) immigrants from the north-western Levant during Middle Bronze II (MB II).

The study is further expanded to the Egyptian Execration Texts (ET) and the neighbouring town of Beth Shemesh. Arguably, in the MBA it was known as Ashnah, which is ‘asinnu of the ET. Its settlers came, specifically, from Yamḫad or Alalaḫ (north-western Syria), and likely controlled nearby bēt ˁāṭām. The bēt- prefix in BA settlements is shown to designate a shrine, with ˁāṭām implicating a vulture deity. This surprising conclusion raises the possibility that a Middle Kingdom (MK) pharaoh marched here with his army, perhaps Senwosret III on his way to Shechem. If so, bēt ˁāṭām might mark the location where the Egyptian and Amorite cultures have first collided.

Bēt ˁAṭāb

On a hilltop, 14.5 km northwest of Bethlehem (aerial distance) and 7.5 km southeast of Tel Beth Shemesh (Figure 1) stand the ruins of Bēt ˁAṭāb [alternately Bēt ˁIṭāb, where ṭ = emphatic t]: A Crusader castle (fortified farmhouse, ‘maison forte’), around which an Arab village has been built (deserted and demolished during the 1948 Arab Israeli war). Today the site constitutes a national park (Figure 2). During the Ottoman period it was the chief town of the ˁArqūb district (Khalidi 1992, 274). A description of the Crusader structure is given by (Pringle 1997, 26–27), who lists its alternate Latin names: Bethaatap, Beitatap and Bethahatap. However, the Archaeological Survey of Israel (hereafter ASI) found sherds also from the MB II, Iron III, Hellenistic and Roman periods (Weiss et al. 2013). Hence the place was inhabited much earlier than the Crusaders’ time.

Figure 1. Portion of sheet XVII from the SWP map (Conder and Kitchener 1880) for the vicinity of Bēt ˁAṭāb. Villages existing at the end of the 19th century appear in red, black lines are wadis, dashed black lines are trails and double dashed lines are roads. Superimposed text: R (yellow) = Roman Road, A (yellow) = Ancient Road, ○ (cyan) = three of four springs near Bēt ˁAṭāb, where the northern one is ˁĒn Ḥod, the south-eastern one is ˁĒn Bēt ˁAṭāb and the southern is ˁĒin el-Birkeh. The southern of the two Roman roads leads from Beit Gubrin to Jerusalem.

Figure 2. Bēt ˁAṭāb knoll and the Roman road as seen from Sifleh on the west. Israel Transverse Mercator (ITM) 205092-626908.

During the 19th century, several exploration teams reported on this site (Rainey 1978, 1–2). Notably, British officers Conder and Kitchener conducted a Survey of Western Palestine (SWP 1871–1878) under the auspices of the Palestine Exploration Fund.³ Their impressions from Bēt ˁAṭāb are as follows (Conder and Kitchener 1883, 22–23) [square brackets added by author]: ⁴

A small village, standing on a remarkable knoll of rock which rises some 60 to 100 feet above the surrounding hilly ridge. The knoll is extremely bare and rugged. There are cisterns among the houses, but the main water-supply is from ‘Ain Haud [upper cyan circle in Figure 1], near which, north-east of the village, the Survey camp was fixed. There are here a few olives on a terrace above a deep valley [Wadi ed-Dilbeh] which runs north of the village. A little further west is another small spring (‘Ain el Khanzireh), by which is a rock-cut tomb. A third small spring (‘Ain Beit ‘Atâb) exists south-east of the village, coming out of a rock. A remarkable cavern (Mŭghâret Bir el Hasûtah) runs beneath the houses. … Though not really at a great elevation as compared with the surrounding hills, Beit ‘Atâb is very conspicuous on all sides.

The SWP map (Figure 1) shows that Bēt ˁAṭāb is situated on an elongated east-west ridge between Wadi ed-Dilbeh in the north and Wadi Saīd in the south. There is a ‘Roman Road’ running just south of the summit, then through Wadi ed-Dilbeh to ancient Beth Shemesh. Another trail down Wadi Saīd (also called Wadi of the Cave) goes west via Te’omim (Twins) Cave. This karst cave, 3.2 km aerial distance from Bēt ˁAṭāb, contains evidence of human activity as early as the Neolithic era. Recent Uranium-Thorium (U-Th) dating suggests that until about 1500 bce it served as a quarry for calcite flowstone (Egyptian alabaster), previously believed to originate exclusively from quarries within Egypt (Frumkin et al. 2014).

Conder’s etymology

The classical Arabic root ˁṭb means ‘to perish, die’ (Lane 1863, 2077), not a likely name for a village. ‘One sure test of a genuine ancient name is when it definitely does not mean anything in Arabic’ (Rainey 1978, 9). Hence Bēt ˁAṭāb should be an ancient name. Conder suggested (Conder and Kitchener 1883, 23) that it derives from Hebrew ˁēṭam (‘place of eagle/vulture’), identifying it as the ‘Rock of ˁĒṭam’ where Samson hid from the Philistines: ‘and he went down and dwelt in the top of the rock Etam’ (Judg 15:8, KJV translation). In his view, Samson would have used the underground cavern as a hiding place. It was secluded because ‘in the time of Samson no town existed here, as it would in such a case most probably have been mentioned with the fourteen Shephelah towns in its neighbourhood’. This is indeed supported by the absence of LB and Iron I/II sherds.

Phonetically, Conder’s suggestion that ˁēṭam > ˁāṭab is plausible, involving the interchange of two labial consonants, m > b (Lipiński 2001, Sec. 11.6). However, it does not explain the bēt- (‘house’) prefix or the ēṭ > āṭ shift. Additionally, a recent study identified Samson’s hideout elsewhere (Gass and Zissu 2009). Finally, this village was actually mentioned in the Judahite town list, albeit in a corrupted form that defied identification.

Roman Era
Ονομαστικον

The first comprehensive attempt to locate [biblical sites] was that of Eusebius, a bishop of Caesarea and fourth-century church historian (ca. 260-339 C.E.). In his Onomasticon, Eusebius cataloged most of the cities, sites and regions mentioned in the Old and New Testaments … including their distances in Roman miles from other well-known metropolitan centers in fourth century Palestine (Notley and Safrai 2005).

A critical edition of the Onomasticon has been published by Timm (2017). In its English translation by C. Umhau Wolf, one finds (Wolf 1971) ‘Ēnada.⁴⁷⁸ (In tribe of) lot of Issachar. There is now another village Ēnadab about ten miles from Eleutheropolis [ = Talmudic Beth Gubrin] on the road to Jerusalem’. His note 478 reads, ‘The distance puts Ēnadab at Beit Nettif if miles are measured as the road is here described. If coming the other way, the 10th milestone is from Jerusalem and points to location at Beit ‘Itab. Neither have anything to do with the biblical site’.⁵

Ēnada (Ηναδα) is the town ˁēn ḥaddah in the North (Josh 19:21). Eusebius comments that there is another town with a similar name, Ēnadab (Ηναδαβ), in Judah. The two possibilities mentioned by Wolf for its identification appeared in earlier literature: Thomsen suggested Beit Nettif (Thomsen 1907, 65), whereas Abel favoured the town that is ‘aujourd’hui [today] Beit ˁaṭab’ (Abel 1938, 316). The latter assignment has been adopted by Avi-Yonah (1976, 55) and his students (Tsafrir et al. 1994, 120). Likely, this relies on the above distances cited by Eusebius, as no further justification was given.

One can reconstruct the Hebrew origin of Ēnadab, and deduce its tentative location, as follows. Eusebius renders Hebrew ˁēn = ‘Spring (of)’ as ην, see the rules of Greek transliterations (Krašovec 2009, 22). The consonant ˁayn (ˁ) is not voiced in Greek, whereas the long vowel ē (contracted from the archaic diphthong ay) is rendered by Greek eta (η). For example, ˁēn ḥaddah > Ηναδα and ˁēn gannim > Ηνγαννι. The common transliteration of Hebrew ṭ is τ, but sometimes also δ, giving ˁēnˁaṭab > Ηναδαβ, ‘Spring of ˁaṭab’. Unlike ancient Semitic scripts, the Greek transliteration retains the vowels. Thus, αδαβ supports the pronunciation ˁaṭab of Arabic Bēt ˁAṭāb. The crusader’s Bethaatap also involves elision of the ˁ, plus b > p interchange, thus αδαβ > atap. It likely developed from the Medieval Arabic form, which was therefore Bēt ˁAṭāb and not Bēt ˁIṭāb.

The prefix ˁēn- (‘spring’) suggests that Roman-era Ēnadab was not situated on the bald hilltop, as the later Crusader/Arab Bēt ˁAṭāb, but rather near a spring called ˁĒn Bēt ˁAṭāb below the summit. One might expect it to be a synonym for ˁĒn Ḥaud, the major spring near the village, but the SWP report clarifies that these are two separate springs: ‘‘Ain Haud … north-east of the village … A little further west is another small spring … [while] a third small spring (‘Ain Beit ‘Atâb) exists south-east of the village, coming out of a rock’ (Conder and Kitchener 1883, 22). There is indeed a small spring coming out of a rock in Wadi Saīd south of the village (Figure 3), but it is called ˁĒn el-Birkeh, ‘Spring of the Pool’ (Palmer 1881, 279), which is a fourth water source, distinct from, ‘‘Ain Beit ‘Atab. The spring of Beit ‘Atab. q.v.; south-east of the village by main road’ (Palmer 1881, 278).

Figure 3. ˁĒn el-Birkeh flows out of a hole in the rock (now tapped into a pipe) under a fig tree near Wadi Said (Figure 1), south of Bēt ˁAṭāb, 31^o43’53” North, 35^o3’14” East (ITM 205182-626536). While this spring is indeed ‘coming out of a rock’, it is not ˁĒn Bēt ˁAṭāb, which according to Palmer (1881) is ‘by main road’. Note that coordinates in degrees were measured by the author and checked for agreement with Google Earth.

Today, there is no other spring ‘coming out of a rock’ that might be identified as ˁĒn Bēt ˁAṭāb. Fortunately, Palmer (above) gives another helpful hint: ‘by main road’. The PEF map (Figure 1) indeed shows a Roman road passing south of the Bēt ˁAṭāb summit, which is the only dirt-road leading there today.⁶ About 200 metres southeast from the summit, just a few metres north of this road, there is an inconspicuous structure with two openings in its roof, probably a cistern or spring (Figure 4).

Figure 4. This inconspicuous structure, some 200 metres southeast of the Bēt ˁAṭāb summit and directly on the Roman road (upper right), conceals a cistern that is likely the ˁĒn Bēt ˁAṭāb mentioned in the PEF report (Conder and Kitchener 1883, 22). Coordinates: 31^o44’2” North, 35^ο3’16” East (ITM 205235-626813), middle cyan circle in Figure 1.

A glimpse inside shows that it is currently dry and filled with dirt that needs to be removed before its role as a cistern/spring can be verified (Figure 5). If there used to be a conduit directing water to its entrance (Figure 4) it is long gone. Just under the olive tree seen in Figure 4, there is a set of two similar structures (Figure 6). Another water hole is located across the main road to the south, close to a burial cave and, a little further on, a vat with an assortment of cup marks hewn into the rock. The abundance of (seasonal) water sources around this spot supports its identification as ˁĒn Bēt ˁAṭāb, reported upon by the SWP team, and hence as the tentative location of Hellenistic Ēnadab.

Figure 5. The inside of the structure from Figure 4. Photograph by Ariel Agmon (with permission).

Figure 6. Openings of water holes located under a large olive tree on main road, across from the cistern in Figure 4.

Iron Age

Proceeding further into the past, the question is whether Bēt ˁAṭāb existed during the Iron Age, in spite of it not been mentioned in the Masoretic Text (MT) of the Hebrew Bible (HALOT = Koehler and Baumgartner 2001). It will now be shown that it was, and its name was probably bēt ˁāṭām. The key for this lies in one verse from the Greek translation of Joshua.

Septuagint

The Septuagint (LXX) is the first translation of the Bible from Hebrew to Greek, which started in the mid–3rd century bce. It has different Greek versions, most well-known are Codex Alexandrinus (LXX^A) and Codex Vaticanus (LXX^B). Alfred Rahlfs prepared a famous edition of the Septuagint (Rahlfs 1935)⁷ which is split into two columns whenever these two versions deviate from each other. The traditional LXX English translation by Brenton has recently been updated with a new English translation, NETS (Pietersma and Wright 2007).

Differences between the MT and LXX are sometimes substantial. Such is the case for some chapters of the Book of Joshua, suggesting that the translators have been using a different (and probably older) Hebrew source (Vorlage) for their work (Rösel 2002; Tov 2012). Margolis published a comparative study of all available LXX versions of the Book of Joshua (Margolis 1931). Of particular interest here is the list of towns of Judah, in Josh 15:20–63, ordered according to the 12 administrative districts of the Judean Kingdom (Aharoni 1979, map 26; Aḥituv 1993; DeVos 2003; Galil 1984; McKinny 2016; Rainey 1983, Fig. 1). Town names differ between MT and LXX, and even between LXX^A and LXX^B. Although LXX^A is c. 100 years younger than LXX^B, its toponyms agree substantially better with the MT (McKinny Tbl. 1–1).

Notably, the towns of the 9th district, to which Bēt ˁAṭāb belongs, were elided from MT Josh 15:59, but appear in the LXX. ‘It seems certain that the list preserved in the Septuagint has been part of the original text and was lost by homoioteleuton [common ending of consecutive verses]’ (Rösel 2002, 12). The missing verse is given in Table 1. It includes 11 towns, subdivided into three groups: 4 near Bethlehem, 3 near Koulon, and 4 near Jerusalem. The first group includes a town named ˁēṭam, one of Rehoboam’s fortified cities (2 Chr 11:6), identified at Ḫirbet el-Ḫoḫ near Solomon’s pools (McKinny 271), too far from Beth Shemesh to be Samson’s Rock of Eṭam. In the second triplet all identifications are problematic (Margolis 1931, 317). Κουλον has been identified with Chesalon (Kallai 1986, 392; Kiepert 1856, Chesalloth; McKinny 2016, 272),⁸ Arab village Keslā (ITM 204964-632016). LXX^A Σωρης, preferred over Εωβης of LXX^B (LXX^A is generally more consistent with the MT), is identified with the Arab village of Sarīs (ITM 207084-633852). It remains to consider the unidentifiable village called Tatam(i).

Table 1. The missing verse from Josh 15:59 according to the Septuagint (Codex A: Alexandrinus, Codex B: Vaticanus), with the ‘enigmatic’ toponym in boldface.

Greek LXX^B (Vaticanus)	Θϵκω καὶ Εφραθα αὕτη ἐστὶν Βαιθλϵϵμ καὶ Φαγωρ καὶ Αιταν καὶ Κουλον καὶ Ταταμ καὶ Εωβης καὶ Καρϵμ καὶ Γαλϵμ καὶ Θϵθηρ καὶ Μανοχω …
English LXX^B (NETS)	Theko and Ephratha (this is Baithleem) and Phagor and Aitan and Koulon and Tatam and Eobes and Karem and Galem and Thether and Manocho …
Greek LXX^A (Alexandrinus)	Θϵκω καὶ Εφραθα αὕτη ἐστὶν Βηθλϵϵμ καὶ Φαγωρ καὶ Αιταμ καὶ Κουλον καὶ Ταταμι καὶ Σωρης καὶ Καρϵμ καὶ Γαλλιμ καὶ Βαιθηρ καὶ Μανοχω …
English LXX^A (NETS)	Theko and Ephratha (this is Bethleem) and Phagor and Aitam and Koulon and Tatami and Sores and Karem and Gallim and Baither and Manocho …
Reconstructed Hebrew Vorlage	תקוע ואפרתה, היא ביתלחם, ופעור ועיטם וכ(ס)לון ו(ב)תעטמי ושרש וכרם וגלים וביתר ומנוחה …

The geography

Several geographical identifications for Tatam(i) were suggested: (a) unknown town in the vicinity of Bēt ˁAṭāb (Galil 1984, 217); (b) Kh. Darban (Ḫirbet esh-Šeḫetah), 1.5 km southeast of Bēt ˁAṭāb (McKinny 2016, 272); (c) Netophah near Bethlehem (Ofer 1993). To proceed, note that the 11 villages of Table 1 (excepting Koulon, which could thus be disregarded) are arranged in south-north pairs (Table 2), consult e.g., the map in Fig. 5-1 of McKinney (2016). Mostly, the orientation is from south-west to north-east. This applies also to Karem-Gallim, when adopting the traditional identification of Gallim in the territory of Benjamin (HALOT 193b). Consequently, Tatam(i) is located south or south-west of Sores. Indeed, as Koulon and Sores are neighbouring villages near the northern border of Judah (red villages in Figure 7), the only option is to place Tatami to their south. It is thus either Bēt ˁAṭāb or Kh. Darban (ASI site # 103).

Figure 7. Satellite-based topographic map of western District 9, prepared by QGIS 3.10. Villages in red are ‘Koulon and Tatami and Sores’ (LXX^A, Josh 15:59). Sites in ochre, plus Bēt ˁAṭāb, are all the MB II sites of map 104 (Nes-Harim) of the ASI (Weiss, Zissu, and Solimany 2013), including their site number (in parentheses). Distances from Bēt ˁAṭāb are indicated in metres (ochre). Elevation data are from NASA’s SRTM file N31E035.hgt, from which the orange elevation contours (in metres) were calculated. Black line is the Jerusalem-Eleutheropolis Roman road (McCormick et al. 2013).

Table 2. The 11 missing villages (NETS-LXX^A spelling) are arranged in south-north pairs, except for Koulon (that has no partner). Arrows (left to right) indicate the order in the list.

Corruption mechanisms

Ταταμ appears in both LXX versions,⁹ suggesting that it emerged during or before the translation process. It will now be argued that the original Iron Age toponym was ‘bēt ˁaṭām’, corrupted to ‘tatam’ by loss of the initial b-.

1. In the Hebrew Vorlage of Joshua, bēt-X toponyms had no space: bētX. Indeed, both LXX versions do not exhibit such spaces see, e.g., Βαιθλϵϵμ in Table 1. This facilitated the concatenation of the two-word toponym into one (e.g., Βαιθηρ was probably originally Βαιθ θηρ). The space in the MT is thus likely an innovation, corroborating the relative antiquity of the Vorlage.¹⁰

2. Although pronounced bētX, it was generally written plene: bytX. This is evident from LXX^A Βαιθσουρ (Josh 15:58) and LXX^B Βαιθλϵϵμ (Table 1).

3. Imagine next that, in contrast to this general rule, bēt ˁaṭam was written in the Hebrew Vorlage as btˁṭm (rather than bytˁṭm) i.e., without the mater lectionis. The translator might have concluded that the initial b- is the mono-consonantal preposition ‘in’, there by mistake, deleting it to obtain tˁṭm (pronounced tˁaṭam).¹¹

4. The elision could not have occurred after the translation process (i.e., in further copies of the Greek text). For comparison, deletion of the initial Βη- in LXX^A Josh 15:61 Βηθαραβα (Betharabah) results in LXX^B Θαραβααμ (on the -am ending later), with an initial theta. Thus, if Βη- was elided from Βηθαταμ after the Greek translation, one would expect Θαταμ, rather than Tαταμ.¹²

5. Finally, elision from tˁaṭam of the Semitic consonant ˁain (Krašovec 2009, 22), which does not exist in Indo-European languages, produces tatam.

In particular, this confirms Conder’s hypothesis that the final consonant was originally -m. It was thus corrupted to -b after the Iron Age but before the Roman era – perhaps during the Hellenistic period.

Vowels and diphthongs

The MT and LXX Vorlage did not include vowels. These could have been preserved orally, then transliterated into Greek. The LXX has Tatam and Eusebius (Ēn)adab, both with two -a- vowels. This agrees also with the Arabic form (Bēt) ˁAṭāb and the Crusaders’ Latin forms, e.g. (Beth)aatap. Hence, the consensus reconstruction (including vowels) is (bēt) ˁāṭām.¹³

In Biblical Hebrew (BH), ˁāṭām is meaningless, but there are two toponyms ˁēṭām (HALOT 817a). Thus, if the Iron-Age Israelites pronounced bēt ˁāṭām as bēt ˁēṭām, it could explain the alternate Arabic form, Bēt ˁIṭāb.

As to the semantics of the name, there is no triconsonantal Hebrew root ˁṭm, whereas the biconsonantal root¹⁴ ˁṭ means ‘to fall upon with shouting’ (HALOT 816b). The noun ˁayṭ, obtained by inserting the diphthong ay between the two consonants, means (appropriately) ‘vulture’, ‘eagle’ or ‘bird of prey’ (HALOT 816b). The toponym ˁēṭām likely derives by adding the accusative ending -am to ˁayṭ (Bauer 1930, 77), ‘place of birds of prey’. As the first syllable is no longer stressed, an ay > ē diphthong contraction has occurred: ˁayṭ-am > ˁēṭ-am. Yet, this development, which occurred in Old Canaanite languages (Lipiński 2001, Sec. 22.6), does not explain the reconstructed form.

In contrast, in Eblaite and Amorite of Northwestern Syria an ay > ā contraction was the rule (Lipiński 2001, Sec. 22.3; Rendsburg 1990, 95–97). An example¹⁵ given is bayt > bāt (‘house’), hence similarly ˁayṭ > ˁāṭ. Indeed, the noun ˁṭ existed in (Northwest Semitic) Ugaritic as either a fish or a bird of prey (del Olmo Lete and Sanmartín 2004, 192). The full toponym reconstruction for the BA is thus bāt ˁāṭām. In consonantal script this would be written btˁṭm, as postulated in item (c) above. Hence the contrast with the conventional bytX spelling may have instigated the loss of the initial b.

Notably, the ay > ā contraction never occurred in (Iron Age) Judahite territory (Rendsburg 1990, 109). Thus, the people who coined this toponym must have arrived from the northern Levant before the Iron Age. Such a southward (Amorite) infiltration is known from the early MB II (Burke 2021), therefore the next sections focus on evidence from that era.

Middle Bronze

While the ASI found a few MB II sherds at Bēt ˁAṭāb’s summit, no excavations were performed there. Is this conclusive evidence for an MB II village there? To answer this, consider the sherd distribution from all MB II sites in ASI map 104.

Geographical site distribution

In the 10 × 10 km² of ASI map 104 (Nes-Harim) only six MB II sites were identified (Table 3), to which the Te’omim Cave quarry was recently added (Frumkin et al. 2014). These sites are depicted as ochre circles in Figure 7. Excepting site no. 81, all are within 3 km of Bēt ˁAṭāb. This tight distribution contrasts with the Iron Age sites that are scattered all over this map. Moreover, excepting no. 103, they are too small to represent a village. Table 3 lists the reported remains as agricultural installations and burial caves, presumably belonging to the inhabitants of a central village. Moreover, all these sites (excepting no. 81) were deserted in the LB, corroborating their affiliation with a single population centre. Only Bēt ˁAṭāb (no. 67) and Kh. Darban (no. 103) are large enough for a village.

Table 3. Sites with MB II remains in the surveyed area of ASI map no. 104, see Figure 7. Periods from which pottery sherds were found: MB – Middle Bronze, LB – Late Bronze, P- Persian, H- Hellenistic, ER- Early Roman, R- Roman, B- Byzantine, MA- Middle Ages, O- Ottoman.

site no.	site name	periods	remains found
23	Har Yaˁala	MB II, Iron III, H, R, B	Agricultural terrace
45	Kh. es-Sīra	MB II, H, ER	Crushing basin, structural remains, rock-hewn caves, steps, cistern
67	Beit ˁAṭāb	MB II, Iron III, H, ER, MA, O	Castle, ruined Arab village
70	Bar Giyyora	MB II, H	Burial cave, hewn cistern, quarry, rectangular structure
81	Har Refa’im	MB II, LB, B	Burial cave, agricultural terrace, structural remains
103	Kh. Darban	MB II, Iron II/III, P, B, MA	30 dunam site
–	Te’omim Cave	MB II	Ancient calcite-alabaster quarry

Kh. Darban is believed to be one of two fortified cities built (together with towers and fortresses) by the latter Kings of Judah (Ganor 2016, 83–84). Hence its area grew only during Iron II/III, whereas in MB II it was likely inferior to Bēt ˁAṭāb.

A closer inspection of Figure 7 reveals an elongated site distribution, rather than circular. This ‘distribution pattern may attest to the existence of an ancient road that ascended to the hill country along the spurs in this area’ (Weiss et al. 2013). Indeed, on the Roman dirt road near the deserted village of Sifleh, c. 1 km west of Bēt ˁAṭāb (see Figure 1), one identifies ‘cart ruts’ engraved on the bedrock (Figure 8), which could corroborate the conclusion that an ancient road passed here.

Figure 8. Cart ruts on the rocks of the Roman road near Sifleh, 31^o44’6” North, 35^o2’37” East (ITM 204209-626938). Average gauge c. 55 cm, wheel width 8 cm. Possibly engraved by carts carrying heavy stones from a quarry.

Archaic suffixes

The reconstructed LXX^A form, bāt ˁāṭāmi, exhibits a final -i not yet discussed. Possibly, this is not a scribal error, but an archaic suffix. Classical Semitic languages had a ‘case-system’—suffixes indicative of the grammatical role of a word in the sentence. The suffixes -u, -a and -i are nominative, accusative and genitive endings (for singular nouns), respectively (Lipiński 2001, Sec. 32, 13–15). Case inflection was used by early 2nd millennium Amorites e.g., ˁabdu baˁlati, ‘servant of the mistress’ (Waltisberg 2011, Sec. 3), as evidenced by ‘Amorite’ personal names (PNs) in Akkadian cuneiform texts. In Canaanite languages the case system became obsolete during the LB (Layton 1990, Chap. 2), but it survived in classical Arabic. Additionally, an -m was suffixed to the case vowel (‘mimation’), obtaining -um, -am and -im endings (Layton 1990, Chap. 4; Lipiński 2001, Sec. 33.16). Mimation was dropped before the case system collapse, perhaps in MB II (Albright 1944, 26).

A related suffix with uncertain grammatical role is the ‘enclitic -m’, which is always followed by a vowel in cuneiform texts, resulting mostly in -mi or -ma endings (del Olmo Lete 2008; Layton 1990, Chap. 4). The existence of similar suffixes for toponyms is demonstrated below and shown to be an important vehicle in dating them.

Execration Texts (ET)

‘The most precious source bearing on Egypt’s relations with Asia in the MK, the so-called execration texts … involve … writing the names [of enemies] on pottery vessels. The curse formula was then undoubtedly pronounced and the object broken’ (Redford 1992, 87). Such pottery bowls, in the Berlin Museum (Sethe 1926, e- and f-series), date to Amenemhat II (12th dynasty, 1929–1895 bce according to Burke (2021, Tbl. 4.1)). Pottery figurines, now at the Royal Museums of Art and History, Brussels, were bought on the market and analysed by Posener (1940, E-series). These conceivably date to Senwosret III (1878–1843 bce) or later. On the body of each statue, a list of enemy rulers and their cities was written in red ink and hieratic script. There are slight differences between versions on different figurines. These are not necessarily ‘scribal errors’: they may be variants from different periods of habitation. Several attempts were made to identify the listed toponyms (Aḥituv 1984; Hoch 1994, 192–95; Maizler (Mazar) 1954; Rainey and Notley 2006, 58; Yeivin 1956), with partial success.

Do the Brussels ET depict the geopolitical reality during the reign of Senwosret III? Older Levantine chronologies suggested that he reigned during Levantine MB I, when places like Shechem were tiny villages (Ben-Tor 2006). Recently, extensive radiocarbon dating of sites in Egypt (Ramsey et al. 2010) and the Levant (Höflmayer 2017) revealed that the chronology of the Levant should be shifted to earlier dates. The transition from MB I to MB II (when fortified towns re-emerged) occurred during the reign of Senwosret III and/or his son Amenemhat III. Hence, the ET do convey important historical information (Streit 2017).

Consider the first 15 toponyms from the Brussels figurines, which include all the listed towns from the Southern Levant (except E45, Jerusalem) up to the border with Lebanon. Table 4 gives their Middle Egyptian spelling, with tentative vocalization following Hoch (1994, 488–96).¹⁶ Important for the present analysis is that Egyptian ἰ or ἰm endings are equivalent to Semitic -i or -im, respectively.

Table 4. Levantine town names from the Brussels figurines (E-series, up to E15). The f-entries correspond to the same toponyms from the Berlin bowls, when available. The Egyptian spelling, from Posener (a₂= Gardiner’s A2), is based on single-letter (alphabetic) hieroglyphs (excepting ḥa in E1). Their likely vocalization follows the Hoch rules (note 16). Identification – with biblical toponyms.

number	Egyptian	vocalization	identification
E1	ḥa ᴈ ἰ m	ḥāl-im ?
E2 f15	ἰ s q ᴈ ἰ ἰ s q ᴈ n w	'asqal-i 'asqal-ānu	Ashkelon
E3, f14	ἰ s ἰ n n w	'asinnu	Ashnah
E4	h ᴈ m u	har-ram-u ?
E5	m k t r ἰ ἰ	magdal-ay
E6	s k m ἰ m s k m m ἰ s k m ἰ m ἰ	ṯakm-im ṯakm-mi ṯakm-im-mi	Shechem
E7	k n ἰ	gan-i	Engannim (Jenīn)
E8	p ἰ ḥ ᴈ u m	piḥl-um	Pella (Jordan)
E9	ἰ p q u m	’apq-um	Apheq (near Akko)
E10	ˁ n ἰ ἰ	ˁen-ay
E11	ἰ a2 k s p ἰ	’aksap-i	Achshaph
E12 f21	ἰ s ἰ a2 p ἰ ἰ a2 s ἰ p ἰ ἰ ἰ ἰ s ἰ p ἰ	’asip-i yasip-i
E13	m ˁ š ἰ ᴈ ἰ	maš’il-i	Misheal
E14 E14, f8	ἰ ᴈ ḥ b u m ἰ ᴈ ḥ b u	’arḥab-um ’arḥab-u	Rehob
E15	ḥ ḏ w ᴈ ἰ ḥ ḏ w ἰ ᴈ ἰ	ḥaḏur-i	Hazor

All 15 toponyms are suffixed, apparently a common practice in the early 2nd millennium. From the distribution of suffixes in Table 5, the most common is -i. The -im ending is barely discernible; its mimation was already lost, yielding -i. Even E2 Ashkelon has here an -i ending, whereas on the earlier Berlin bowls its suffix is -ānu.¹⁷ There are three towns with the nominative -um ending: E8 Pella and E9 Tel Apheq, which have remains as of the Chalcolithic era, and E14 Rehob. Of the three towns with the latter name, recent research points to Tel-Rosh of the Upper Galilee as Rehob of Josh 19:28. It was ‘settled almost continuously from the Early Bronze (EB) Age to the Ottoman period’ (Katz and Levin 2021, 36). Thus, the -um suffix correlates with sites that were settled already in the EB. However, 3/5 of the attestations of E14 Rehob, and f8, appear without mimation, which is characteristic of the MBA.

Table 5. Distribution of suffixes for the ET toponyms of Table 4.

Suffix	Toponym
-mi	E6
-ānu	f15
-ay	E5, E10
-um	E8, E9, E14
-u	E3 = f14, E4?, E14 = f8
-im	E1?, E6
-i	E2, E5, E7, E11, E12 = f21, E13, E15

Interestingly, Shechem (E6), from northwest Semitic ṯakm ‘shoulder’, exhibits three different suffixes: -im (mimation), -mi (enclitic), and both: ṯakm-im-mi (duplicated m is not always explicit in Egyptian scripts). In the last variant, the MB II -mi suffix is added on top of the -im suffix, so that the latter must be older than MB II (MB I or EB). Notably, MB II Shechem, Megiddo and Hazor are the only fortified sites in the Southern Levant (c. 40 in total) exhibiting three (or more) material elements (e.g., migdāl temples) observed in MB II northern Levantine Alalaḫ and Ugarit (Buck 2020, Map 4.25). ‘The preferred explanation for the spread of the material assemblage known from the current type site of Ugarit is as a result of population migrations at the start of the MBA’ (Buck 2020, 176).¹⁸

A remaining question is whether such endings could have survived >1,000 years, to be manifested in LXX Tatami? To answer this, the next section identifies such suffixes in BH toponyms.

Suffixes in biblical toponyms

The idea that archaic case endings in toponyms can be used to date them is important not only for Tatami. Moreover, when a given toponym carries different suffixes in different texts it could be because these are relics from different historical periods of the site’s occupation. To further corroborate these ideas, consider some toponymic suffixes from the Hebrew Bible and Septuagint (Table 6).

1. ˁAdullam is identified with certainty and surveyed archaeologically, with remains from all periods as of the Chalcolithic (McKinny 2016, 178). Its -am ending is common to both major LXX versions. Thus, like -um and -im in the ET, -am is also a marker of an EB settlement.

2. ˁEnam, like ˁadull-am, has the ‘accusative’ -am ending, ˁēn-am, and means ‘place of the spring’ (HALOT 820a). According to LXX^A the name is Ηναιμ¸ like the ˁēn-ayim where Judah met Tamar (Gen 38:14, 21).¹⁹ The LXX^B version is maˁian-i, ‘spring’, with the -i ending. Thus, three suffixes (-am, -ayim and -i) are manifested for this entry, which could indicate the existence of a settlement through three different periods starting with the EB. Unfortunately, this site remains unidentified.

3. ˁAy is the second city conquered after Jericho (Josh 8:1), generally identified with the EB II site of et-Tell. In the Benjaminite townlist, ˁavv-im is generally interpreted as the ‘people of ˁAy’. This has no support in the text, which lists only towns: ‘the cities … of Benjamin … were Jericho, … and Zemaraim, and Bethel, and ˁAvim, … twelve cities with their villages’ (Josh 18:21–23, KJV). This sums up to 12 cities only with ˁAvim included. Thus ˁavv-im (mimation) could be the EB name of ˁay (no mimation).

4. Zanoaḥ (of the Maon district) is written without a suffix in the MT, with a suffix -im in LXX^A, and -ayim in LXX^B.²⁰ In BH, -ayim is the dual suffix. However, ‘place names ending in -ayim are no duals, but toponyms formed with an archaic … gentilitial suffix -ay with the genitive ending -i, followed by the mimation … ’ (Lipiński 2001, Secs. 29.54, 67.16). The -ay and -im suffixes are attested in Table 4 (E5, E10 and E1, E6, respectively), but not -ayim, which must be a layer development.

5. Zemaraim and Shaˁaraim are two MT -ayim cities. Remarkably, none of their LXX variants exhibit the -ayim morphology. For Zemaraim, LXX^A manifests the -im ending, while LXX^B has an -a ending (and misses the μ: Σαρα instead of Σαμρα). Shaˁaraim derives from the Semitic root ṯġr, ‘gate’, so that the strange combination ργ in LXX^A Σαργαριμ is an effort to emulate the sound of Semitic ġ (more conventionally rendered as γ). If the Joshua Vorlage is indeed older than the MT, -im would be the original ending, suggesting EB origins for this city. If true, then Shaˁaraim cannot be identified at Kh. Qeiyāfa, which dates to the 11–10 centuries bce. This adds to the argument contra its etymology as ‘two gates’, because the -ayim ending is not dual (Na’aman 2008).

6. There has been a debate on the identification of the City of Ephraim. A recent publication (Keinan et al. 2015) supports Albright’s identification at Kh. Marjameh, where sherds have been found from the EB onwards (Mazar 1995). Here the LXX versions concurr, but the Onomasticon is of interest: ‘Ephrōn. (In) tribe of Juda. There is now a (very) large village (named) Ephraim (Efraea) twenty miles north of Jerusalem’. In parentheses is Jerome’s Latin translation, in which some additional sources must have been used. These suggested Efraea, which could attest to an archaic 'ephr-ay-im (Lipiński 2001, Secs. 29.54) after losing mimation to yield 'ephray-i.

   Table 6. Biblical toponyms from the southern Israelite tribes with archaic suffixes.

   Joshua verse	Biblical name	Suggested arch. site	Era	LXX^A	LXX^B	McKinny (page #)	HALOT (page #)
   15:35	ˁadull-am	Khirbet Tel Sheikh Madhkûr	EB	Οδολλαμ	Οδολλαμ	178	792a
   15:34	ˁēn-am	?		Ηναιμ	Μαιανι	176	820a
   15:36	šaˁar-ayim	?		Σαργαριμ	Σακαριμ	181-3	1619a
   15:56	zanoaḥ	?		Ζανωακιμ	Ζακαναιμ	257-58	275b
   18:22	ṣemar-ayim	?		Σϵμριμ	Σαρα	321-23	1035b
   18:23	ˁavv-im	et-Tel	EB	Αυιμ	Αιιν	324-25	797a
   2 Sam 13:23	'ephray-im	Kh. Marjameh	EB	Eusebius: Ephraim	Jerome: Efraea	–	80b-81a

It thus appears that the LXX -im/-i suffixes are indeed very old. The Septuagint translators could not have invented them, 1500 years after the ET, unless they were still preserved in the language. Consequently, there is no reason not to consider LXX^A Tatami as preserving a relic of an ancient suffix that should now be analysed.

What’s in a suffix?

How could the suffix of LXX^A Tatami, reconstructed batˁaṭami, be interpreted? The first inclination is to analyse the -i as a genitive suffix, which is dominant in the ET (7 out of its first 15 entries, see Table 5). This is acceptable if the name is derived from a triconsonantal root ˁṭm. However, such a root is unknown in Northwest Semitic languages, and therefore ˁaṭam has been analysed as biconsonantal ˁ(y)ṭ + the accusative ending -am (Bauer 1930, 77). One cannot have two different case endings on the same word, hence the -i in batˁaṭami is not a genitive suffix. What is it then?

A resolution of this dilemma is possible if the ending was not -i but rather -mi namely, bat-ˁaṭ-am-mi (assuming that a consonant duplication is not always explicit in the LXX). While an enclitic -mi is rare for southern Levantine toponyms, it is attested in the E6 variants ṯakm-mi and ṯakm-im-mi. In the second example, -mi follows an -im case ending, analogously to ˁaṭ-am-mi, where it follows an -am ending. Given that Shechem was an Amorite stronghold in the early 2nd millennium bce (Buck 2020, Sec. 4.3), the enclitic -mi could have originated with some Amorite dialect, which is considered next for Amorite PNs.

Enclitic -m in personal names during MB II

Layton has searched for evidence of enclitics in PNs of early West Semitic (Layton 1990, 156–60). In the ET he did not find any, but did find a few examples in the Alalaḫ Tablets [AT = (Wiseman 1953, 125–64)], though only from level IV (1550–1473 bce). MB II better corresponds to Alalaḫ level VII (1720–1650 bce). To obtain the required information, the > 2100 PNs of the AT were filtered as follows: only level VII, with an enclitic -m suffix, and clear Semitic origin (Table 7), verified by comparison with Appendix A of Buck 2020.

Table 7. Amorite PNs from Alalaḫ level VII with enclitic -ma.

Transcription	Transliteration	Translation	AT Page
Ab-be-li-im-ma	’abi-li’m-ma	The tribe is my father	125
Am-mi-ta-kum-ma	ˁammi-taqum-ma	Rise, my paternal uncle	127
A-na-i-lim-ma	ḥanna-’ilim-ma	Grace of God (’Ilu)	127

Ammitakum was a king of Alalaḥ in the 17th century bce. His name appears frequently in the tablets, sometimes with the final -ma and sometimes without, indicating that this particle is not an integral part of the name. However, the three PN examples involve the suffix -ma, not -mi as in batˁaṭam-mi. Possibly, the opposition between -ma and -mi indicated human vs non-human beings, respectively (del Olmo Lete 2008, 29). Therefore -ma is attached to PNs and -mi to toponyms.

The etymology of ’ilim-ma in the last PN:²¹ ’Ilu = ’El, ‘god’, followed by the genitive case marker + mimation + enclitic -ma. The same analysis applies to ˁaṭ-am-mi: accusative case marker + mimation + enclitic -mi. Furthermore, the last two examples in Table 7 are the only cases of enclitic -m listed in a recent compilation of MBA Western Amorite PNs between roughly 1900 to 1600 bce (Buck 2020, Appendix A). They are all from Alalaḫ (Tel Atchana, in the ˁAmuq valley, near the bend of the Orontes). Hence, the simplest interpretation is that the settlers of Bēt ˁAṭāb emigrated from Alalaḫ, bringing with them the -mi ending. Unexpected support for this interpretation comes from the largest town in its vicinity, Beth Shemesh.

’Ashnah is Beth Shemesh

As of the MBA, Beth Shemesh was a major site, a few km west of Bēt ˁAṭāb. Nevertheless, it is not mentioned in any source outside the Hebrew Bible (Bunimovitz and Lederman 2009, 116), and is even absent from the Judahite town list in Josh 15. However, the lot of Dan (before moving north) included ‘Zorah, and Eshtaol, and Irshemesh’ (Josh 19:41, KJV). Most researchers equate ˁir šēmeš with bēt šēmeš. Zorah and Eshtaol are always mentioned together, close to Beth Shemesh. They are included also in the Judahite town-list: ‘And in the valley, Eshtaol, and Zoreah, and Ashnah’ (Josh 15:33). Here 'Ashnah (LXX^A Ασνα) replaces Irshemesh. A simple explanation resolving the above issues is that 'ašna was the BA name for Beth Shemesh, just as ’Ephratha was the BA name for Bethlehem.

'Ashnah was identified with E3, 'asinnu, see Table 4 (Yeivin 1956, 37). It appears also on the earlier Berlin bowls (f14, e13–15). This was criticized because although ‘'as’annu should be located in southern Palestine … proposals to identify it with Ashna are linguistically improbable [why so?] and must be rejected on archeological grounds as well, since the latter was an insignificant settlement’ (Aḥituv 1984, 68). However, if 'Ashnah was Beth Shemesh, then it actually was a significant town.

In both E- and f-series 'asinnu (E3, f14) appears next to Ashkelon (E2, f15). Therefore, as Aḥituv stated, it ‘should be located in southern Palestine’. Interestingly, in the f-list 'asinnu (f14) is also next to yarmut (f13), which is therefore the EB city Yarmuth of the southern Levant.

The inhabitants of Yarmuth … reverted to living … as pastoral nomads around 2300 bce. The country thus glided … into a Dark Age … from which it emerged only at the beginning of the second millennium bce, with the process of reurbanization that characterizes the onset of MB II (de Miroschedji 1999, 15).

Tel Yarmuth is just 4.6 km (aerial distance) south of Tel Beth Shemesh, so that an identification of 'asinnu with Beth Shemesh indeed places it near Yarmuth.

This identification, if correct, allows one to date Beth Shemesh. Because 'asinnu appears on the Berlin bowls it should predate MB II (before 1850 bce). Because the ending is -u, and not -um, it should postdate the EB (after c. 2000 bce). Excavations at Tel Beth Shemesh have not yet systematically probed the BA levels, but segments of a city wall from MB IIB (c. 1750 bce) have been observed (Bunimovitz and Lederman 2009, 120).

Niqmi-Epuh prince of ’Ashnah

Interestingly, the ET ‘prince’ of 'Ashnah is listed as nqmwpˁἰ (Posener 1940, E3), reconstructed niqmi-yapuˁ, ‘my vengeance shall appear’. This Amorite name is attested in Akkadian syllabic texts from Alalaḫ and Yamḫad of north-western Syria (Buck 2020, 310–11) e.g., Ni-iq-mi-e-pu-uḫ.²² In fact, a person with that name was a Great King of Yamḫad (capital Ḫalab), 1700–1675 bce (Middle chronology), about 150 years after the ET figurines were crafted. Because it is a rather non-common PN,²³ limited to the geographical area of Alalaḫ/Yamḫad, this area must be the homeland of the early settlers of Beth Shemesh. Because a PN can reside in a family for generations, E3 niqmi-yapuˁ was possibly one of King’s niqmi-yapuˁ familial ancestors, who has left Yamḫad to settle in 'Ashnah. Likely, the nearby village of Bēt ˁAṭāb fell under his jurisdiction. Given the difficulties in purchasing a ‘settlement’ in Yamḫad (Lauinger 2015), obtaining one (for free?) in the fertile Soreq valley must have seemed like a lucrative option—until it vexed an Egyptian Pharaoh. This possibility emerges from analysing the bēt- prefix.

Bronze-Age Bēt-X toponyms

Bēt-X toponyms ‘were in active use during both the Israelite and pre-Israelite periods’ (Isserlin 1957, 133). The Hebrew Bible mentions 52 such toponyms (HALOT, 125–29). Of these, a few were identified with known BA archaeological sites. Eight of these, from the Judean hill country (Rahkonen 2016, Sec. 5.2), are listed in Table 8. Excluded from the list are Bēt-Tappuaḥ, which is an Iron II site with no BA remains (McKinny 2016, 248), and Bethlehem, which has EB remains, but its older name, ’Ephratha (Table 1), is not of the Bēt-X type. The remaining sites have BA remains and their names are not known to have changed. If ṣur (‘rock’) is understood as a synonym to ’el (HALOT, 1017a), then all X’s are deities. Therefore, a BA/Bēt-X toponym designated a temple to the deity X.

Table 8. Bronze-Age Bēt-X toponyms in the Judean hill country. References for BA remains: G = (Greenberg and Keinan 2009), followed by site no.; M = (McKinny, 2016), followed by page no.; P = (Peilstöcker 2006).

Period	X	Deity type	Identification	Reference
EB	’El	Chief	Beitin	G249
	Zur	Chief ?	Kh. eṭ-Ṭubeiqeh	M264
	ˁAnoth	Oracle or War goddess	Kh. Bēt ˁAnun	G882 (M266)
MB	Shemesh	Sun	Tel er-Rumeilah	M66
	Dagon	Wheat	Tel Bēt-Dagan	P
LB	Ḥoron	Underworld	Lower Bēt Ġur	G276

Bēt ˁAṭāb fits in this list as an MB II site. Its reconstructed name, bātˁāṭām, is the north Levantine equivalent of bētˁēṭām, following the alternate ay > ā diphthong contraction. Therefore, ˁaṭam should represent some deity. However, the accusative/locative -am ending suggests reading ‘where the birds of prey are’ (HALOT 817a) namely, a place rather than a deity. This can be resolved by first attaching the prefix bat, obtaining batˁaṭ, ‘vulture shrine’,²⁴ followed by the locative ending -am, (batˁaṭ)-am. Thus, its etymology is ‘place of (-am) the vulture (-ˁaṭ-) shrine (bat-)’. The deity is a vulture, and the indicated place is where its shrine used to stand.

A vulture shrine at Bēt ˁAṭāb is a surprise, because Semites did not worship avian deities. Egyptians believed in them, goddess Nekhbet being their vulture deity (Hill 2008; Wilkinson 2003, 213). Indeed, Bēt ˁAṭāb and Beth Shemesh are on the border between the mountains and the Shephelah, a region that was frequently under Egyptian influence, especially during the New Kingdom (NK). Did Egyptian influence reach this far during the MK?

Nekhbet in the Levant?

During the MK, Nekhbet was not a widely worshipped goddess in Egypt. She was the protector of the royal family, often depicted hovering above the pharaoh in battle with her wings wide spread, a protective posture,²⁵ holding an ankh or shen ring. Scenes of the king smiting enemies within a wooden/reed Nekhbet shrine are depicted in the two pectorals of the Egyptian princess Mereret (Aldred 1971 no. 41, 42), daughter of king Senwosret III (1878–1843 BCE). Given this symbolism, one could imagine a portable Nekhbet shrine accompanying the king to battle. Thus, Nekhbet could have arrived in the Levant only in the context of a royal military campaign.

The local population, if originating from northern Syria, would have been aware of the Nekhbet–vulture symbolism. Indeed, MB II cylinder seals from Alalaḫ and vicinity contain ample Egyptian iconographies (Teissier 1996). Specifically, a Nekhbet vulture in its protective posture is attested on the seal of ’Abba-El (Collon 1975, 6), Great King of Yamḫad (c. 1750–1720 bce) and grandfather of king Niqmi-Epuh. Thus, it is not inconceivable that they would call the place where Egyptians set their Nekhbet shrine batˁaṭāmi, if indeed Egyptians have reached this far inland.

Egyptian kings of the MK showed little interest in military campaigns into the Levant, except for Senwosret III, considered the greatest military leader of the 12th dynasty (Gee 2004). In 1900, the archaeologist John Garstang found a limestone stele of an Egyptian official named Khusobek (Sebek-Khu) near his tomb in Abydos (presently in the Manchester Museum). It bears testimony to a military campaign of Senwosret III against Shechem (Peet 1914; Wilson 1969, 230).²⁶ Apparently, he reached the ‘district named Sekmem’ (perhaps not Shechem itself), turned back to go to the palace and was attacked by Asiatics. ET toponyms E1-E7 (and E45), Table 4, might depict his premeditated itinerary: ’Ashkelon to ’Asnah, up the mountains to Jerusalem, and along the ridge-road to Shechem. The stretch from ’Asnah, if identified with Beth Shemesh, to Jerusalem could have passed by Bēt ˁAṭāb. Indeed, 19th century travellers have taken such a road (Kiepert 1856, red lines). Had he chosen this route, Senwosret III would have had the opportunity to set a Nekhbet shrine there.

Walking up Wadi Saīd, his army engineers may have noted the flowstone deposits in Te'omim Cave. ‘The use of local calcite-alabaster sources may have demanded the transfer of technological knowledge from Egypt’ (Frumkin et al. 2014, 757a), which could occur following an Egyptian military expedition. Demand for alabaster in Egypt increased during the MK, while possessing only 9 travertine vs 99 limestone quarries, all periods considered (Harrell and Storemyr 2009, at end), and none of the flowstone type. Therefore MK Egypt would have been interested in this quarry.

Conclusion

Traditionally, toponym identification in the Levant is based on sound similarity with Arab placenames. This subjective criterion may be of limited validity, unless supportive evidence exists. Moreover, it does not help in dating the toponym, even in the fortunate case that the site itself can be dated archaeologically. Here, an interdisciplinary (language, geography and archaeology) multi-source approach succeeded in tracing a toponym’s history.

Bēt ˁAṭāb is a deserted Arab village, built around a Crusader castle in western Judah, not far from Beth Shemesh. Its name does not appear in the MT, and thus its origins remained obscure. This was rectified here by reconstructing Bēt ˁAṭāb‘s ancient names in different periods (Table 9). Subsequent etymological analysis even revealed the origins of its first settlers.

Table 9. Evolution of Bēt ˁAṭāb’s names throughout history.

Name	period	location	comments
bātˁāṭāmi	MB II	knoll	Regional centre, road, temple, quarry, agriculture
	LB, Iron I, II		Uninhabited
bētˁāṭām = Tatam	Iron III	knoll?	Judahite citadel?
Ēnadab	Hellenistic, Early Roman	spring on main road	Non-Hebraic populace
Bethaatap Bēt ˁAṭāb	Crusader/Moslem	knoll	Citadel, farmhouse, Arab village

The key is the enigmatic toponym Tatami of LXX^A Josh 15:59. Comparison with the Arab and Medieval Latin names, and Ēnadab of Eusebius, establishes the reconstruction bātˁaṭāmi, in alphabetic script: btˁṭmi. The missing mater lectionis in bt, ‘house’, which facilitated the loss of the initial b- during LXX translation, is typical to Semitic languages of the northern Levant. Deriving from Northwest Semitic ˁayṭ, ‘valture/eagle’, the ay diphthong contracted to ā rather than ē, also characteristic of the northern Levant, and hence the likely origin of the first Bēt ˁAṭāb settlers.

During the BA, names had suffixes that could be dated. In bātˁaṭāmi, the suffix is an enclitic -mi (rather than genitive -i). It is found in two variants of E6 Shechem, demonstrated archaeologically to be an Amorite stronghold in the southern Levant. A few Amorite PNs from the early 2nd millennium exhibit analogous morphologies with an enclitic -ma and, remarkably, they are all from Alalaḫ. Thus, both Shechem and Bēt ˁAṭāb had population from the same Amorite tribe, whose encroachment southward could have worried an Egyptian king.

The Alalaḫ/Yamḫad connection is demonstrated for nearby Beth Shemesh, identified with ’Ashnah (E3 ’Asinnu, following E2 ’Ashkelon). This identification is corroborated by the Berlin bowls where f13 Yarmuth is followed by f14 ’Asinnu, then f15 ’Ashkelon. The Amorite ruler of ’Asinnu was named Niqmi-Epuh, a somewhat unusual name known exclusively from Yamḫad and Alalaḫ. This must then be where the original settlers came from. One may imagine this Niqmi-Epuh, a cadet prince of Yamḫad, leaving home with his followers to establish a new city-state in the southern Levant. Their journey south may have been facilitated by access through the southern port city of Ashkelon (42 km aerial distance from Beth Shemesh), which was already an Amorite enclave (as testified by its -ānu suffix).

Bēt ˁAṭāb manifests also Egyptian connections. It is close to ’Ashnah (Tel Beth Shemesh, 7.5 km away), on the route of the army of Senwosret III if he followed the ET toponyms from ’Ashkelon to Shechem. Then, a reasonable road from ’Ashnah to Jerusalem would go by Bēt ˁAṭāb, which was then only a hill with a nice spring (ˁEn Ḥaud) at its foot (otherwise it would have been mentioned in the ET). The Egyptian army may have camped in its vicinity, setting a portable Nekhbet shrine on the knoll. This would explain the Bēt ˁAṭāb etymology: ‘place where a vulture shrine stood’. Evidently, the Yamḫad settlers did not believe in avian deities, hence the name for them had only topographical connotation. From their camp, the Egyptians would send scouts, possibly discovering the flowstone deposits in Twins Cave. Perhaps Senwosret III turned around without seizing Shechem because he reached an agreement with the Amorites on joint exploitation of natural resources, including the Twins Cave deposits. Regardless of whether such a scenario is real or illusionary, it is probable that this region has indeed witnessed the first collision of the Egyptian and Amorite civilizations.

Acknowledgments

I thank Yigal Bloch for suggesting the diphthong contraction, and for discussions and comments on various versions of this work. I thank Ariel Agmon for the photograph presented in Figure 5, and comments on an earlier version of the manuscript. The Fritz Haber Center is supported by the Minerva Gesellschaft für die Forschung, München, FRG.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Funding

No funding was reported by the author.

Notes on contributors

Noam Agmon

Noam Agmon (PhD in theoretical chemistry, 1980). After a two-year postdoctoral fellowship at Harvard and Caltech, he joined the chemistry department of the Hebrew University of Jerusalem, promoted to full professor in 2000. He has authored around 190 scientific papers, e.g., on proton mobility. During a 2006 Sabbatical he became interested in the interrelations between language and archaeology, establishing an interdisciplinary group to study this, and showing (with Yigal Bloch) that pre-Semitic languages transitioned from bi- to tri-consonantal morphology concomitant with the transition to agriculture. His interest in Bēt ˁAṭab arose after relocating to its vicinity, finalizing this work during the COVID years.

Notes

1 The first is transliteration from Hebrew, which does not capitalise proper nouns, while the second is from an English translation of the Hebrew Bible (e.g., The KJV).

2 Such a temple, if existed, has not been unearthed, perhaps because its remains are under the Iron Age layers. In this genitive relationship between bēt (‘house’) and šemeš (‘sun’), called the construct state, the original ay diphthong (in bayt) has been contracted to ē.

3 On the PEF history, see their internet site http://www.pef.org.uk/about/history/.

4 Transliterations in quoted texts are as in the original source. Transliterations of toponyms from Arabic in the present text follows the SWP style (Palmer 1881), except for the diphthongs. Consequently Arabic toponyms are capitalised, whereas Biblical Hebrew transliterations are not, compare (Elitzur 2013). For consistency with Hebrew, Semitic consonant symbols are used, and the -ay diphthong is contracted to ē in the construct state.

5 ‘When a village was located along a lane off a main road, the distance to that village was reported only up to the last milestone on the main road before the branching point’ (Keinan et al. 2015, 226). According to Thomsen (1917, 80), mile XI on the main road from Jerusalem to Eleutheropolis would be near today’s Zur Hadassah junction, which is close to where the Bēt ˁAṭāb road may have branched off.

6 It is not the main Roman road from Jerusalem to Eleutheropolis, but rather a secondary road connecting it to Beth Shemesh (Figure 1).

7 Available online: http://www.hs-augsburg.de/~harsch/graeca/Chronologia/S_ante03/VT/vte_pd00.html

8 This is more plausible than Ḫirbet Umm el-Qalˁa (ITM 211800-624350), where an Iron II fortress was uncovered (McKinny 2016, 456), because Arabic qālˁa (and not Qa'la) means (appropriately) ‘fortress’, and has nothing to do with the (proto-Semitic) root ksl, ‘loin, side’ (HALOT 489b).

9 Margolis compared with additional LXX versions (Margolis 1931, 317), finding for LXX^B ταταμ the variants ταταν, ταιταταμ and δαταμ that reconstruct back to ταταμ. For LXX^B ταταμι he finds the variant ταμι.

10 Concatenation of two different towns also occurred: LXX^A Josh 15:23 Ιθναζιφ clearly stands for Ithnan and Ziph (KJV translation).

11 This scenario is more likely than a parchment defect because the many translators must have used more than one scroll, and it is unlikely that they all had a defect in the same spot.

12 In contrast to a final Hebrew tav (ת) that is transliterated as θ, an initial tav can be transliterated as either θ or τ. For example, the transliterated toponym tappuaḥ, ‘apple’, begins sometimes with θ and others with τ (HALOT, 1774b).

13 LXX^A has a final -i vowel, Tatami. This, and the bēt prefix, are discussed below.

14 On the classification of bi- vs tri- Proto-Semitic roots and their dating to before and after the ‘agricultural revolution’ see Agmon (2010).

15 An interesting example from the Hebrew Bible involves Saray, Abram’s wife, ordered by God to change her name to Sarā (Gen 17:15). Conceivably, this ay > ā contraction also characterized the city of Ḫaran (on the border of Yamḫad in NW Syria), where Abram’s family was stationed (Gen 11:31).

16 The following ‘Hoch rules’ for Middle Egyptian are relevant to the table: Egyptian ᴈ (vulture hieroglyph, Gardiner’s G1) is always a consonant: l or r (never the vowel a); Egyptian ἰ (reed hieroglyph, M17) is a vowel, i, except when it appears before certain hieroglyphs, and then it is a vocalized aleph: ἰ a₂ = 'a, ἰ ᴈ = 'il/r, ἰ s = 'as, where a₂ = Gardiner’s A2; Egyptian ἰ ἰ = y, ya or ay; The consonant combination m ˁ (G17 and D36) is pronounced ma.

17 According to Gelb (1961, 42), the suffix -ānu is manifested between Mt Lebanon and Alalaḫ, and is hence non-Canaanite (i.e., Amorite), whereas its later Canaanite equivalent is -ūnu.

18 The verse in Gen 48:22 may be interpreted as an alternate tradition of Jacob the conqueror, who has conquered Shechem from the Amorites (Farber 2013, 18–21). Hence, if the ‘Patriarchal Era’ is placed in the early 2nd millennium bce, this tradition concurs with the archaeological conclusion of an Amorite Shechem in MB II.

19 The morpheme -ayim in BH generally indicates duality, but not in toponyms. Indeed, the two other variants indicate a ‘spring’, and not ‘two springs’.

20 Evidently, LXX^B Ζακαναιμ is a common metathesis error, and should be read Ζανακαιμ. Both versions transliterate the ḥ in Zanoaḥ with κ, which is unusual for Greek transliterations. However, compare 2 Sam 3:16, baḥurim, with LXX Βαρακιμ.

21 This is not the plural ending -ima. Compare with bney-'elim in Ps 29:1; 89:7 = children of 'el (rather than ‘of gods’).

22 A similar conclusion is reached by searching e.g., in ARCHIBAB: Archives Babiloniennes, http://www.archibab.fr/, under Niqmi-epuh. Out of 5 attestations, 2 are from Mari (1770 bce) and 3 from the settlement of Alaḫtum, generally believed to be Alalaḫ.

23 In comparison, assuming the correct spelling of the prince of Shechem (E6) was ’Abi-Haddu (Hoch 1994, 494), ‘Hadad is my father', such a common PN precludes any attempt to determine his place of origin.

24 From here onwards ‘vulture’ is used as a shorthand for ‘birds of prey’.

25 Compare: ‘He shall cover thee with his feathers, and under his wings shalt thou trust’ (KJV Ps 91:4).

26 Another MK military intervention, this time along the coast of Lebanon, is deduced from the inscription on the mastaba of Khnumhotep at Dahshur (Allen 2008).