¹ For details see the complete survey of divergent views in Herzfeld's, Der Tell Halaf und das Problem der hettitischen Kunst (Archäologische Mitteilungen aus Iran, VI, 3–4)Google Scholar.

² The Art and Architecture of the Ancient Orient (Pelican History of Art, Harmondsworth), pp. 172 ffGoogle Scholar.

³ Tell Halaf, III (Berlin, 1955), pp. 3 ffGoogle Scholar.

⁴ Bibliotheca Orientalis, XIII (1956), pp. 36 fGoogle Scholar.

⁵ Cf. From the Stone Age to Christianity (Baltimore, 1940)Google Scholar, in which I still dated the “earliest orthostates” from Gozan in the eleventh century (p. 120), and Archaeology and the Religion of Israel (Baltimore, 1942)Google Scholar, by which time I had settled on the tenth century (p. 213, n. 5). In a copy of a letter to R. A. Bowman, dated the 14th January, 1942, I wrote: “for over three years I have maintained a date in the tenth century for Kapara and for the bulk of the orthostates.”

⁶ Op. cit., p. 166, supported by numerous other similar assertions.

⁷ Cf. Weidner, E., Archiv für Orientforschung, III (1926), pp. 151–161Google Scholar, and Forrer, E. O., Reallexikon der Assyriologie, I, p. 290 (1930)Google Scholar.

⁸ For this date, which is now assured, see Gelb, I. J., Jour. Near East. Stud., XIII (1954), pp. 221, 229Google Scholar, and Gurney, O. R., Anatolian Studies, III (1953), p. 17Google Scholar. The date of the prism from Tell 'Ashârah thus becomes 886 instead of 884.

⁹ Les Annales Archéologiques de Syrie, II (1952), pp. 169–190Google Scholar. Tournay's interpretation of the difficult text is in the main correct, though some minor corrections will be necessary.

¹⁰ For details see my treatment in the Hetty Goldman Festschrift.

¹¹ The line between Assyrian and Babylonian influence generally ran across the Middle Euphrates in periods known to us.

¹² Cf. Barnett, R. D., Carchemish, III, p. 263Google Scholar, and Anatolian Studies, III, pp. 90 fGoogle Scholar.

¹³ See especially Dunand, M. in Bulletin du Musée de Beyrouth, III (1939), pp. 65–76Google Scholar, and my study in BASOR., No. 87 (1942), pp. 23–9Google Scholar.

¹⁴ See Layard, A. H., Discoveries among the Ruins of Nineveh and Babylon (London, 1853), pp. 235 ff.Google Scholar, and on their place in the history of Assyrian art see von Bissing, F. W. in the Abhandlungen of the Bavarian Academy, XXVI (1914), pp. 12 ffGoogle Scholar.

¹⁵ See Unger, E. in Ebert, , Reallexikon der Vorgeschichte, XI, p. 181Google Scholar, and BASOR., No. 130 (1953), pp. 15 ffGoogle Scholar.

¹⁶ von Luschan, F., Sendschirli, IV (Berlin, 1911), p. 375Google Scholar, and for a detailed bibliography see Rosenthal, F. in Pritchard's, J. B.Ancient Near Eastern Texts Relating to the Old Testament, 2nd ed. (Princeton, 1955), p. 500Google Scholar.

¹⁷ See Carchemish, III (1952), especially pp. 261 ffGoogle Scholar.

¹⁸ Studi Classici e Orientali, I (Pisa, 1951), pp. 35–67Google Scholar.

¹⁹ See Barnett, op. cit., p. 262 f.

²⁰ See his excellent study, Späthethitische Bildkunst (Ankara, 1949)Google Scholar, and for minor modifications see my paper in the Hetty Goldman Festschrift.

²¹ For the architecture of Tell Ḥalâf see the splendid work of Langenegger, F., Müller, K. and Naumann, R., Tell Halaf, II (Berlin, 1950)Google Scholar, especially the “Tempel-Palast” (called “Hilani” in Vol. III) and the remaining constructions in the north-west section of the citadel, described by Langenegger, op. cit., pp. 17–113. On pp. 112 f. he has a most impressive tabular contrast between the north-western group of buildings and the remaining constructions on the mound, ending with the following words: “Danach überragte der Tempel-Palast alle übrigen Grossbauten auf der Burg in ihrer letzten Gestalt und am Ende der Herrschaft der Kapara-Dynastie.”

²² Cf. Place's plan of the palace of Sargon (e.g., in the frontispiece of Khorsabad, I, Chicago, 1936Google Scholar) and especially Residences J-M in Pl. 70 of Khorsabad, II (Chicago, 1938Google Scholar). Of course, the provincial residence at Gozan may go back to the early eighth century and looks in any case more archaic than the constructions of Sargon.

²³ For a good photograph of the spring of the arch in the Outer Portal of Gate 7 see Khorsabad, I, Fig. 3 on p. 3Google Scholar; the reconstructed section appears at the upper right of Fig. 4 in the next page. Unfortunately, all the vaults and arches represented in Khorsabad, II, are restored as semi-circular in section, though none of them was preserved as high as the spring.

²⁴ R. J. Braidwood's survey of the chronological situation with respect to small objects, Am. Jour. Sent. Lang., LVIII (1941), pp. 364 ff.Google Scholar, does not throw much light on the date of construction of the buildings in question, since practically all small finds naturally date from the Assyrian occupation.

²⁵ Tell Ḥalaf, II, p. 225Google Scholar, Fig. 116. I do not mean to imply that the “Altbau-Epoche” went back to the Kapara period, since the early phase seems to have been quite similar in plan to the later phase.

²⁶ See the fuller discussion of this point in my study in the Hetty Goldman Festschrift

²⁷ The comparative material is treated most carefully by Ekrem Akurgal in his Späthetitische Bildkunst (see above, n. 20); see especially pp. 23 (where the hair-do is shown to have changed in the eighth century), 30 ff., 66 ff., 73 ff., 83 ff. Akurgal's date for the sculptures of Tell Ḥalâf in the ninth century should be pushed back a little, but this does not affect his relative chronology or his treatment in detail.

²⁸ Comparison between the lily and palmette motifs among the Megiddo ivories and at Tell Ḥalâf is most instructive. There are two principal varieties, both of which appear in both places. One shows the lily stamens in three branching units between the two volutes, which spiral toward them, emerging from a proto-Ionic volute below (The Megiddo Ivories, Pl. 34, nos. 166 a and b, etc.; Tell Halaf, III, Pl. 70 b, 125 a–b, etc.). The other shows a crown originating in the six elements of the three stamens, projecting far above the inturned spiral volutes (Megiddo Ivories, Pl. 6, no. 14, etc.); by the time of the Kapara sculpture this has become a curved or flat-topped crown containing from six to a dozen stamen elements rising far above the top of a conical stand (?) from which project two everted volutes at some vertical distance apart (Tell Ḥalâf, II, Pls. 71–8). In the second variety there can be no doubt that a considerable period of development has intervened between the Megiddo ivories of the early twelfth century and the Kapara period at Tell Ḥalâf. On the other hand, both Gozan varieties are older than the closest parallels in the ninth–eighth century ivories of Phoenicia and Syria.

²⁹ See Barnett, R. D., Iraq, II (1935), pp. 186 ff.Google Scholar, and Jour. Hell. Stud., LXVIII (1948), p. 14Google Scholar; see also H. Frankfort, op. cit., p. 260, n. 135, who defends Barnett's dating, in spite of the fact that he repeatedly elsewhere in the same volume denies the existence of Syrian art before the middle of the ninth century.

³⁰ Published by Meissner, B. in the Oppenheim Festschrift (Berlin, 1933), pp. 71–9Google Scholar, and by Weidner, E. F. and Ungnad, A. in Die Inschriften votn Tell Halaf (Berlin, 1940)Google Scholar.

³¹ Meissner, op. cit., pp. 72 ff. For a rather indistinct photograph of the original see Tell Halaf, III, Pl. 133.

³² So von Soden, W. in Tell Halaf, III, p. 20Google Scholar, for TI.ME (read AN-ŠI = ilim by Meissner).

³³ An Assyrian dialectal form, like ēpašūni below. Attention has already been called to the fact that this statement reflects Syro-Hittite usage (e.g. in the somewhat later Kilamuwa inscription).

³⁴ This word has been a crux interpretum, since the original is damaged at its beginning and since the parallel text is supposed to offer li-ši-ru-pu. However, I have never believed in the value li_x supposed to be attested here and in the next inscription, line 2 (see below, n. 39), for normal LID, since it has not yet been found anywhere except in the Kapara texts at Gozan, and suggest that we should read simply lit-táš(DIŠ)-ru-pu, disregarding the thin horizontal line (which replaces the normal wedge at the end of a character in the Kapara texts) and the damaged initial corner wedge as reconstructed after the ši in the parallel text. The form littašrupu is IV.2 from ŠRP; the u for a is dialectal Assyrian, and neo-Assyrian sometimes replaces the normal precative by this form, called “perfect” by von Soden, (Grundriss der akkadischen Grammatik (1952), §81Google Scholare, note), who also quotes (loc. cit., and Orientalia, XIX, p. 396Google Scholar) late Assyrian iterative forms such as littatlak and littataprar.

³⁵ Meissner's lu-ra-me makes no sense. I see nothing whatever in the photograph after the vertical wedge of “ME” and see no difficulty in reading lu-ra-a, where the last vertical wedge of RA has been dropped by haplography before A. In any case, the last upright wedge looks double. The text must be referring to the dedication of young girls as harlots to their patron goddess.

³⁶ Meissner's reading of the name of the scribe as ^mAb-di-ili is in itself very reasonable, but makes the rest of his translation improbable. His alternative proposal, ^mAb-di-ili-ia ₅(MU) is also possible and has parallels. However, elsewhere in our text abiya is written a-bi-ia, so I have adopted the reading Abdi-ilimu after two names found in the Ugaritic tablets of the fourteenth—thirteenth centuries published by Nougayrol (Schaeffer, C. F. A., Le palais royal d'Ugarit, III, Paris, 1955, index, p. 240Google Scholar). In several Ugaritic occurrences the name is spelled ^mARAD-AN-mu, corresponding to alphabetic 'bd-'elm. The final u may be an Accadian adaptation.

³⁷ This is Meissner's alternative reading, which is vastly preferable to his choice.

³⁸ Meissner, ibid., pp. 74 ff.; we have no published photograph.

³⁹ Though the word seems to be written PA-LID-E, there can be no doubt that the second character is not an otherwise unattested li _x but is simply the three-wedge form of the sign DIN, which has the value tì in Old Assyrian and Arnarna texts. The different stance of the wedges is not surprising in the Kapar a inscriptions. The form is presumably a gentilic like the Assyrian Ḫa-at-te-e of the texts of Tiglath-pileser III and Sargon. Since writing my study in the Hetty Goldman Festschrift I have become convinced that the word is identical with “Hittite”, especially since it appears in the inscriptions of Asshur-naṣir-apli and Shalmaneser III as Ḫat-ta-aya, which can also be duplicated in the Assyrian inscriptions in the sense of “Hittite”.

⁴⁰ This form, though not normal, is certainly possible in a barbarous orthography. However, the Kapara inscriptions are not nearly so poorly spelled as is sometimes thought, so I suggest a possible alternative li/liš-ru-pu as a simple scribal error for lišrupu. We must again remember that the horizontal wedge of ŠI is written only as a thin line with no head, which might easily be an illusion when one deals with squeezes of this kind.

⁴¹ Among other archaic northern forms is ṬU written with only two vertical wedges at the left, as sometimes in Mitannian.

⁴² See von Soden, W., Tell Halaf, III, p. 20Google Scholar, and for a detailed account of the Assyrian dialect grammar of the royal inscriptions between 932 and 859 see Seidmann, J., Mitt. Altorient. Ges., IX, 3 (1935), pp. 46 ffGoogle Scholar.

⁴³ Die Inschriften vom Tell Halaf (Berlin, 1940), pp. 68–70Google Scholar.

⁴⁴ Am. Jour. Sem. Lang., LVIII (1941), pp. 359–367Google Scholar.

⁴⁵ ZDNT is the clear reading of G. R. Meyer's copy, on which Friedrich based his attempt at decipherment. This would be the Hittite royal and noble name Zidantas, etc., from the fifteenths—fourteenth centuries. However, the second and third letters of the name are not quite certain. I have thought of a good later Aramaic name Zabînat, but it never occurs in such early times. Hittite royal names recur again and again among the Syro-Hittite rulers of northern Syria in the ninth–eighth centuries, and it is hard to escape the fact that the reigning dynasty of Gozan considered itself as Hittite. The second word is particularly clear in Bowman's copy, all that is needed to make it certain being the line closing B at the top. The title ba'al, “lord,” is well attested: Niqmadda of Ugarit is called in the alphabetic inscriptions 'adn Yrgb b'l Ṯrmn, “lord ('adon) of YRGB and lord (ba'al) of ṮRMN,” and bēl āli, “lord of the city,” is a common Assyrian expression for “local ruler”. If the first letter of the second word is B, the first letter of the last must also be B, again with the top line worn away. Bowman's copy shows clearly that the word need not end with Y, so in the light of the excellent attestation of Bīt-Baḫiāni as name of the district we are almost compelled to read as we do.

⁴⁶ The usual rendering of u-kal-lu-ú-ni as “inne hat, held” from kālu is very improbable; read the pi'el of kalū, which occurs in the suggested sense.

⁴⁷ For the text see Seidmann, J., Die Inschriften Adadniraris II (Mitt. Altor. Ges., IX, 3, 1935, pp. 28 ff.Google Scholar).

⁴⁸ First proposed by Opitz, D., Zeits. f. Assyr., 1927, p. 300Google Scholar.

⁴⁹ See Schrader, , Keilinschriftliche Bibliothek, I (1889), p. 74Google Scholar; the text appears in I R, Pl. 20, ii; 21 ff.

⁵⁰ Exactly the same spelling, Ḫat-ta-aya, appears twice in the inscriptions of Shalmaneser II I (Monolith, ii: 85, Obelisk, 40) for “Hittite”, referring to Pitru (Pethor) in northern Syria.

⁵¹ See Schrader, op. cit., p. 105, and Luckenbill, , Ancient Records of Assyria, I, p. 164Google Scholar. It may be observed that the Assyrians took with them native levies from all the conquered states as far as Carchemish, including Gozan.

⁵² In the Hetty Goldman Festschrift I have made a slight mistake, following a lapsus calami of Weidner, reading māt Guzana in the entry for 808; actually we have āl Guzana in both entries—for 808 as well as for 758.

⁵³ This is obviously the same name as Ugaritic Ḫdyn (Virolleaud, , Syria, XV, p. 245 f.Google Scholar, line 8). This name has been overlooked by De Langhe, and the damaged Ḥdyn, which he quotes after Virolleaud, Rev. d'Assyr., XXXVII (1940), p. 144Google Scholar, RS 141, 7, may just as well have been Qdyn. In BASOR., No. 87 (1942), p. 26Google Scholar, n. 7, I have made the identifications, pointing out that the verbal stem was ḫḏy or ḫḏw, in view of the Heb. Z. The biblical Ḥezyôn would presumably have been a contemporary of the Ḫadiānu of Gozan.

⁵⁴ Pronounced Abišalâm (Heb. Abšalôm or Abīšālôm), since the Assyrians reversed the Babylonian sibilants and no longer pronounced final short vowels. This is naturally the same name as was borne by the slightly earlier biblical Absalom, whose mother was also an Aramaean princess.

⁵⁵ This represents a change in my attitude since writing my paper in the Hetty Goldman Festschrift, where I left the identity of Ḫattê and Ḫattāya with “Hittite” open for discussion; see n. 39, 45.