^{page 425 note 1} Periplus, 29, p. 9.Google Scholar

^{page 426 note 1} Two slightly diflFerent proposals, that in Institution, 311, the beginning of the 4th century, and that of Dr.Beeston, A. F. L. in BSOAS., 16, 41Google Scholar, about A.D. 280–300, both depend on the mention of Nağran in the lmru l-Qais inscription from Namara as ‘ town of Šammar’. It is not clear that Šammar Yuhar'iš is meant. M. J. Ryckmans has rightly preferred to consider Sammar in the inscription Ry 508/4 a known geographical name, see Muséon 66, 334.Google Scholar The inscription may indicate the relation of Nağran to an important region north of it in the years before 328.

^{page 426 note 2} A typical example of the use of this expression by Cosmas, see p. 454.

^{page 426 note 3} Cosmas, , 2, 27 C-D: pp. 69–70.Google Scholar

^{page 426 note 4} ibid., XI, 447 D–448 A: p. 322.

^{page 426 note 5} ibid., II, 27 A: p. 89.

^{page 426 note 6} Procopius, Anekdota, 25, 13: see LRE, 2, 330–2.Google Scholar

^{page 427 note 1} θ for b, as in Avάσαpθov for Anazarbus, Malalas, p. 444, corrected by Theopbanes, , 1, p. 263.Google Scholar Professor Guillaume has proved the alternation of f and t in some early roots, BSOAS., 16, 1–12.CrossRefGoogle Scholar

^{page 427 note 2} Haury did not admit the variant Esimiphaios; it seems a secondary error. The form, which occurs only in Procopius, shows how such names were treated.

^{page 427 note 3} ἀξιŵv might be translated ‘ demanding ’, but Justinian was not in a position to demand, and aίτησιv subsequently shows the correct nuance.

^{page 427 note 4} ‘Wars’, 1, 20, 9–13.Google Scholar

^{page 428 note 1} ‘Wars’, 1, 12, 2–16.Google Scholar

^{page 429 note 1} Periplus, 20. The Kanraitai are otherwise unknown.

^{page 429 note 2} Nöldeke, 400–34.

^{page 430 note 1} Nöldeke, 170, note 1, cites the references.

^{page 430 note 2} Theophanes, , 1, p. 371.Google Scholar

^{page 430 note 3} Figures from Jacob of Edessa; Agathias generally wrong if varying.

^{page 430 note 4} Synchronisms: Nöldeke, Ṭabari, 78–9: 85: 132–3: 169–72: 345–8.

^{page 430 note 5} The figures presuppose coregencies.

^{page 430 note 6} Broken years counted with each reign.

^{page 430 note 7} 7 years᾽ interval not accounted for; possibly Suhrab is in wrong order.

^{page 431 note 1} RES 3383, unpublished.

^{page 431 note 2} This was proposed almost as soon as the inscription was known, and the date ia given in all the standard reference books, e.g. by J. H. Mordtmann, art. Ḥimyar, in the Encyclopaedia of Islam.

^{page 431 note 3} B80AS, 16, 37–40.Google Scholar

^{page 431 note 4} Kaθєipξαv: Procopius wrote Attic Greek modelled on Thucydidea and Polybius, so the natural meaning is not doubtful. There is no reference to a siege.

^{page 432 note 1} τὴv άpxὴv єxpaτúvαro: the translation in the Loeb edition, ‘ he strengthened his rule’, does not adequately render the nuance, see Thucydides iii, 82, 12.

^{page 432 note 2} ‘Wars’, 1, 20, 2–8.Google Scholar

^{page 432 note 3} The sources are John of Asia, ‘Ecclesiastical History’, iii, iv, 6–7Google Scholar, Dittenberger I, no. 201, and Procopius, ‘Wars’, 1, 19, 37Google Scholar, the account of the destruction of the temples at Philae by command of Justinian.

^{page 432 note 4} Published partially in RES 3904, fully in Muséon, 59, 165–72Google Scholar, by Professor G. Ryckmans. The restoration of the royal titulature after the names of the sons by M. J. Ryckmans in Inaitution, 242, quoted in Muséon, 64, 338Google Scholar, and B8OAS., 16, 38Google Scholar, is not indicated in the preserved text or in any source. Restoration is not evidence.

^{page 432 note 5} See p. 458.

^{page 433 note 1} (k)m śmlkn bḫh mik ḥbšt: mlk is mulk, šmlkn, which Professor Ryckmans emends to the plural mlkn, is clear in a well-cut inscription. As the mason omitted the ṣad in the name, he more probably omitted the infix t in śtmlkn than mistook alif for sin.

^{page 433 note 2} The assumption that they are identical has obscured the relation of the inscription to the Arab sources; the deputies for the nagašǎt are known to those sources, not the grant of kingship to them by the nagaśi. The only co-regents with Śumu-yafa were Abyssinians.

^{page 433 note 3} tḫlhmw It bḥrn: Arabic ‘ to flow ’, root .

^{page 434 note 1} Masūdi, iii, 172, 167, 162.

^{page 434 note 3} Nöldeke, 172.

^{page 434 note 3} Table A shows that over 7 Seleucid years in the Persian reckoning were not accounted for in Hišam1FBD;s synchronisms. The record may have been deficient at this point.

^{page 434 note 4} He first proposed this theory in Journal Asiatique, 11 Série, Tome 18, 29–32Google Scholar; it was repeated in Storia d᾽Etiopia, 189, 197.

^{page 435 note 1} Nöldeke, 203–4, with note.

^{page 435 note 2} Müller, , Fragmenta Historicorum Graecorum, 4, 179.Google Scholar

^{page 435 note 3} Muséon, 66, 275–84Google Scholar, Ry no. 506.

^{page 435 note 4} Location uncertain; I take it to be the region north of early Saba.

^{page 435 note 5} The plural suffixed pronoun is regarded by some as a plural of majesty, though mlkn takes a singular verb. The singular suffix in ‘ Its month of’ so-and-so is deictic. Arabs who were in personal subjection to the king could hardly be put on a level with the kingdoms. The title is a claim to rule over Arabs dwelling on the plateau and in Tihama as opposed to other Arabs.

^{page 435 note 6} Indefinite for passive. Again sometimes taken as royal plural. The formula is common, and occurs in other than royal inscriptions.

^{page 435 note 7} k: clauses containing dependent statements not infrequently begin thus, instead of with ḏ, in the late Sabaean inscriptions.

^{page 435 note 8} qśdw: the translation ‘ rebelled ’ is not justified by CIH 541/9–10. For the social classes qśd-śd see Rhodokanakis, Altsabäische Texte, 1, 49–50.Google Scholar On ś;d as ‘ soldier ’, Mordtmann-Mittwoch, 232–4, to my mind convincing proof that both classes were men of military age liable to the levy. The distinction is between the ruling people and the subject provinces. The verb is both active and static, ‘ to subject’ and ‘ to be subject’.

^{page 435 note 9} This preliminary to the main battle was a custom of the period; the wording distinguishes the tribes from the main army.

^{page 435 note 10} whrgw w srw wmnmw ḏ śm wmḫḏ; mlkn: Professor Ryckmans takes the last w as introducing the principal clause after a subordinate beginning with wmnmw. If śm has its normal Arabic sense, mnmw is the subject of hrgw w srw, the w before it emphatic or an error.

^{page 435 note 11} ws hmw mrm bn mḏrn wrhnhmw bnhw wśtḫlfhw ly m dm: Amr is the subject throughout, the 3rd person suffix refers to Abraha. The -ta- infix forms in the South Arabian inscriptions commonly mean acceptance of the action expressed without the infix; compare ntṣr, to accept protection, śtrญw, to accept as pleasing, tmm, to accept guidance.

^{page 435 note 12} The month name is omitted by error.

^{page 436 note 1} Of the aญ ญawāhir: Masūdi, iv, 122; for present location, Philby, H. StJ. B., Arabian Highlands, 25.Google Scholar

^{page 436 note 2} Masūdi, iii, 119, iv, 121. The arrangement of Quṣaiy by which such tribes at Mecca were al biṭāḫ may correspond to the territories originally held; if so, there was a long-standing division of the whole confederation.

^{page 436 note 3} Muséon, 66, 342.Google Scholar

^{page 436 note 4} The dates are fixed by Procopius, ‘Wars’, 2: from 5, 540Google Scholar; from xiv, 8, 541; from xx, 542; from xxiv, 543; from xxvi, 544; xxviii, 11, 545.

^{page 436 note 5} Justinian did not know that the war of the Lakhmid against the Ghassanid would continue. In 545–6 Munḏr killed a son of Ḥariṭ, and later suffered a severe reverse. ‘Wars’, 2, 28, 12–14.Google Scholar The ‘ sacrifice to Aphrodite’ is an ‘ embroidery’ like those found in Arab stories, and is on a level with another by Procopius, the cannibal Saracens. It is surprising that it has been accepted as historical evidence that Mund.ir was a savage pagan by some who reject historical statements in the Arab sources because of the very human stories.

^{page 437 note 1} Glaser, transliterated into Hebrew: CIH 541, re-transliterated: Chrestomathia no. 64: Fakhry I, pp. 79–83, no. 4 bis. M. Fakhry᾽s copy has been extensively used for this translation.

^{page 437 note 2} See p. 435, note 4. This instance might be thought decisive.

^{page 437 note 3} Unexplained. Abramos, slave of a Roman, may have claimed Roman citizenship when he revolted from Aksum. The hard breathing might represent the diaeresis in Greek ‘Pωµαios. is the Greek s in krsts.

^{page 437 note 4} Za Yabman at Murayghan; the official inscription at Marib is likely to be correct. For Geez names with Za compare the king list printed in Budge, Ethiopia, 1, 206–7.Google Scholar

^{page 437 note 5} Generally translated ‘ Kiddat and Da. Compare Kiddat Wail. The Yazid family of Zaan held the Wadi Rukhailah, north of Azzan, By 340 in Muséon, 52, 312–6.Google Scholar

^{page 437 note 6} znn, plural, posits this form; the region yzn, in Arabic Yazān, has a not uncommon prefix, see p. 443, n. 3.

^{page 437 note 7} This may be a tribal name in the plural; if so, Sulaim.

^{page 437 note 8} śyw: RES 4193/8 requires this meaning: RES 4084/7–8, – because the god replied in an oracle that he would send them away in peace’.

^{page 437 note 9} Ry 520/1, ḏbnnr might be connected, Muséon, 67, pl. 1.Google Scholar

^{page 438 note 1} yfqn: Arabic afaqa: the glossary entry, Chrestomathia, incorrect.

^{page 438 note 2} śḥtw, as in Arabic, Akkadian šaḥatu, see Revue d'Assyriologie, 19, 68–9Google Scholar, common in the Maer archive, often wrongly transliterated with ṭ.

^{page 438 note 3} hgn: root hwg. A northerner: an ancestor of Ghassān was so named.

^{page 438 note 4} št : the preliminary to a campaign, RES 4193/8–9, causative in CIH 308/19, of calling up the levy for labour, CIH 541/94.

^{page 438 note 5} Armies formerly were levied by fifties, ḥmś: not verbiage.

^{page 438 note 6} mqly: plural construct of maqāwil, (Kamūs): not a preposition.

^{page 438 note 7} lw wlmd: hendiadys. In CIH 540/78 read klmdw.

^{page 438 note 8} A section of Zaan, as noted by Ryckmans, M. J. in Muséon, 66, 337.Google Scholar On the location see von Wissmann—Höner, 92–3.

^{page 438 note 9} This seems decisive; Yazīd had not rebelled at this time.

^{page 438 note 10} ‘rmn w wdn wḫbšm wmḍrft ḍ fn: construction of arim. CIH 540/6–22, of wd, ibid. 22–8. Combined they correspond to Akkadian arammu, see Thureau-Dangin, F., Textes mathématiques babyloniens, 21, no. 45.Google Scholarḫbšm corresponds to mḏbn in CIH 540/10–11, the reservoir. All had a facing of quarried stone, grb, and brick, Ibt, here called mḍrft, root ḍfr, used of lining wells, RES 2817/1, and of facing a mswd, RES 4198/2. In spite of the lack of mimation in mḍrft, ḏ'fn is the periphrasis for the genitive: f, in the form nf, ‘ façade,’ in the Mineaean RES 3535/1 and 3029/1.

^{page 438 note 11} brt: the meaning is settled by RES 3689/12 and Glaser 1399 in N. Rhodokanakis, Die Inschriften an der Mauer von Koḥlan-Timna , brṯm wm , brṯm wm brm, gyr brṯm, covenanted or agreed without covenant. Both are texts from Qataban. The word is Hebrew, not cognate with the brṯ which represents Arabic burūz(un).

^{page 438 note 12} gb'w m yzd: causative in Fakhry I, p. 117, fig. 66, II, no. 121, ‘to command by law’, and in the broken context CIH 398/8; either the causative or the simple form in the Qatabanian text RES 3858/1; the simple form in RES 3908/5,‘ because (the god) protected him in what he ordered him’. For the passive compare CIH 621/8. In Ey 3/4, Muséon, 40, 165Google Scholar, the causative has the specific sense ‘ to have goods ordered in advance ’. Not related in sense to the Arabic ğaba᾿a, ‘ to turn back in a cowardly manner ’. Compare Akkadian g/qabū, ‘ to say ’, frequently ‘ to order ’. Aramaic gb , ‘ to impose a tax ’ is cognate.

^{page 438 note 13} The other party.

^{page 439 note 1} ẓtm: possibly related to the adverbial ,‘ never ’.

^{page 439 note 2} The source of the material mbr , CIH 540/, 11, 24–5, 63, fictile clay.

^{page 439 note 3} qdśw: specifically Christian, contrast ญll or ṣll, line 92.

^{page 439 note 4} Glaser, ḏbmśtlh: Fakhry b mśtlh: meaning guessed.

^{page 439 note 5} rn, defective for wrn, iwārun, ‘ defect’.

^{page 439 note 6} The necessity of clearing a length of the dyke to repair damage there caused by the breach in the ramp would increase risks in Marib.

^{page 439 note 7} Glaser, ḍllm w wśm, a common phrase which does not mean pernicies et pestilentia, as CIH renders: CIH 540/68–9, ḍllm wmwtm, ‘ anxiety, even death,’ Fakhry, clearly bllm: if this is correct, compare Arabic balla, ‘ to moisten ’, then ‘ to exercise benevolence ’, the metaphor explained in Freytag I, 147 a.

^{page 439 note 8} , ‘ admiratione affectus fuit’, Freytag, i, 23 a. CIH: dimiserunt eos, but the sense ‘ to permit’ would require an addition with b.

^{page 439 note 9} Glaser, wkwṣḥw. mlkn: Fakhry, wkwṣḥw hmlkn: CIH, Imlkn, conjecture. The preposition h for l occurs only in inscriptions from Ḥaḍramawt, with the infinitive. At present, emendation seems out of place.

^{page 439 note 10} Note the use of the military force on labour for the state.

^{page 439 note 11} Glaser, .lmtm: CIH ślmtm: Fakhry, glṯm: g and l are indistinguishable, ṯ and y sometimes confused in the copy. I guess glym.

^{page 439 note 12} mt: i.e. matta, to ask for access to a person: ‘ when ’ makes no sense.

^{page 439 note 13} dl, an appellative, would normally have -m in a personal name.

^{page 439 note 14} Fayiš was not a normal tribe, but a body of courtiers. Adug was possibly an Abyssinian.

^{page 439 note 15} So Fakhry. Glaser, Dhu ŠWLMN.

^{page 439 note 16} On southern Ruain see von Wissmann-Höfner, index. Originally perhaps the stock from which the dynasty of Ḥimyar sprang, Mascūdi, iii, 203.

^{page 440 note 1} Either a region, not included in Ḥaḍramawt, or a description, ‘foreigners (literally, opponents),’ settled in the land.

^{page 440 note 2} mhškt: in inscriptions of this period ḥškt is used of a wife. The occurrences have been discussed by Beeston, Dr. and ProfessorRyckmans, in Muséon, lxv, 279, lxvi, 109–11, 67, 103.Google Scholar The combination bḥšk wrd wtḥrg implies that ḥše does not mean‘ help ’ or‘ instigation ’ but something similar, perhaps ‘ instruction ’. In the damaged text RES 4194/2, nśm ḥškm ks bbythmw must mean an employee of some sort, not one in command; ks is construct before a prepositional phrase, and accusative, ‘ the man looking after the household quarters ’. In Ry 520/5 it is permissible to doubt whether a worshipper of rḥmnn would proclaim his polygamy. In the 6th century women at Nagran engaged in business, ḥšktm may describe the wife who looks after a man's affairs, and need not be applicable only to a wife, wld in the same texts is not necessarily confined to the literal meaning. On the whole there is no reason to doubt that mḥškt, not mentioned in these discussions, is cognate and means something like ‘ those conducting negotiations for ’ a king. Some significance must be attached to the distinction from tnblt.

^{page 440 note 3} Glaser, ḍlln bhmd: Fakhry, i, with no space for i, which is not required. ḍll here for ṣll, which occurs in the simple form in CIH 540/79. Form: like for .

^{page 440 note 4} In the pact; the previous work was voluntary.

^{page 440 note 5} Either an intercalary month or two months had the same name.

^{page 440 note 6} Šaraḥ-b-il Yafur; his decree, CIH 540/64–74, included Ḥimyar and Ḥaḍramawt, and was issued as a result of a breach immediately after his own reconstruction.

^{page 440 note 7} Unexplained. Possibly there was a rota of duty for princes to take charge of annual plastering. If so, ṭnkn is a suffixed form from ṭīn.

^{page 440 note 8} Not‘ to alter ’. The Arabic nakira means ‘ to be ignorant of’ something, but the adjective nakir(un) seems to retain the sense required here,‘ intelligent’. The form is presumably intensive.

^{page 440 note 9} nmry mfggm: nmry dissimilation, dual construct of mamarr(un), Freytag, iv, 165 a. mfggm root fağğa, to diverge.

^{page 440 note 10} Glaser and Fakhry, rz w, which would mean‘ assigned’: CIH 540/82, rz w ḫrṣm … wnskm, ‘ they distributed gold and electrum ’. This does not fit the present context. I read rt w.

^{page 440 note 11} ṭbḫm, though the beasts are called ḏbyḥm. Note the intentional restriction of the blessing. The rest was not a sacrifice, but a feast.

^{page 441 note 1} Reading blm śfym wgrbbm wqṣym. The first and the last adjectives are common, but I do not know the significance of the dark colour.

^{page 441 note 2} śqym ḏtmrm: if this applies to the doves, I can make nothing of it.

^{page 441 note 3} Adopting the restoration in CIH, k(ml)w.

^{page 441 note 4} There are two nouns mqḥ, see Mordtmann-Mittwoch, 178–9. ‘ Booty ’ is surely from Iqh, as Conti Rossini gave it. The meaning ‘ building work ’ I would associate with the Arabic participle Freytag, iv, 941a ‘ expertus ’.

^{page 441 note 5} Restoring bṯmnyt wḫmśy ymim wq w Igzwhmw.

^{page 441 note 6} von Wissmann—Höfner, 123–4. The eastern end of the incense route, Periplus, 24: ‘ The exports from there (Mouza, Mūkha) are either local goods, myrrh, both the choice and the virgin, from Abeira and Minaea, (or) limestone, and all the cargoes already mentioned from Adulis ’.

^{page 441 note 7} Not to be confused with the kdr mentioned in the inscription of Karib-il Watar, which von Wissmann—Höfner, 38, locate near Timna, unless that was a tribe and had moved. The narrative of Abraha's dam inscription proves that Abran was a considerable distance east, perhaps a little north, of Ṣirwaḫ.

^{page 442 note 1} Two bronzes in a classical style, of the 1st century A.D., from Ukhdūd, with a South Arabian inscription, show clearly Nabataean influence: British Mueeum Quarterly, 9, 153–6Google Scholar; Ryckmans, G. in Muséon, 52, 62–4Google Scholar; Rostovtzeff, M., Social and Economic History of the Hellenistic World, 2, 855Google Scholar and III, 1537 (not quite accurate); Philby, Arabian Highlands, 85. The inscription: ‘ X has dedicated at the canal these recipients to the god Dhu Śamawi’: is significant in the future Christian centre; consideration of the character of this ‘ lord of heavenly ones ’ must take local developments into account.

^{page 442 note 2} Nöldeke, 56.

^{page 442 note 3} ibid., 238.

^{page 442 note 4} Loftus, W. K., Travels and Researches, 233Google Scholar: Capt. Shakespeare, in JRGS, 1932, 59Google Scholar, plate at p. 325; Col. Dickson, A. P. in Iraq, 10, 1–8CrossRefGoogle Scholar: Ryckmans, G. in Muséon, 1, 239–40Google Scholar (correct Thay to Thağ): Winnett, F. V. in BASOR no. 102 (03, 1946), 4–5.Google Scholar

^{page 442 note 5} Procopius, , ‘Wars’, II, 1, 798.Google Scholar

^{page 442 note 6} Dussaud, René, Topographic Historique de la Syrie, 255, 286.Google Scholar

^{page 443 note 1} Perhaps following Leo's introduction of it, p. 444.

^{page 443 note 2} If there has been discussion of this point since that of Glaser, it is unknown to me.

^{page 443 note 3} Not infrequently spelt Jotabe in English eference works. Yo- is connected with the prefix ya- in regional names already noted (see p. 437, n. 6). For examples in ancient and modern southern Arabia see von Wissmann—Höfner, 113, Anm. 3, and 140. It is ancient, for Amorite districts of the 18th century B.C. have such names, given in the Maer archive. One instance, Yamutbal, also called Emutbal, proves that ya- is a separable element, since it is also called Mutiabal.

^{page 443 note 4} i, 217–8.

^{page 444 note 1} CMH 472. The source ia Müller, Fragmenta Historicorum Graecorum, iv, fr. 1 of Malchus, De legibus gentium.

^{page 444 note 2} GMH, ibid. (E. W. Brooks).

^{page 444 note 3} Many places were so called; one in Egypt, see Meredith, D. in BSOAS, 16, 1954, p. 237.Google Scholar

^{page 445 note 1} See pp. 428, 435.

^{page 445 note 2} References in F. Krenkow, Kinda, in Encyclopaedia of Islam.

^{page 445 note 3} Nöldeke, 148. Extracts relating to two distinct men are put side by side. The confusion caused by like names is constant.

^{page 445 note 4} The texts are: Fakhry, i, p. 105, fig. 48, ii, no. 60; Ry 509 in Muséon, 46, 304Google Scholar, where sb is used of a peaceful expedition as in Ry 3 in Muséon, 40, 165Google Scholar, line 5; CIH 540/52–4. This Abi-karib Asad cannot be identified unhesitatingly with the co-regent of Maliki-karib Yuhamin who was an older brother of Dara-Amar Ayman in era year 493; the interval is too long. Otherwise Institution, 318.

^{page 445 note 5} So, if RES 4105 is to be restored as suggested in Institution, 222, and Abi-karib Asad and Šaraḥ-b-il were co-regent without Ḥassan being mentioned. Šarah-b-il, named first in Fakhry no. 60, must be the older brother, and subsequently ruled alone. Ry 509 by its nature does not prove that Šaraḥ-b-il was co-regent, disappeared and restored.

^{page 445 note 6} The bitter herbs meant must be the Akkadian irru, ‘ the bitter plant’, ša(m)mu marru, see Campbell Thompson, Assyrian Botany, 223–9, used to alleviate internal pains, identified by him with the opium poppy, by B. Landsberger with cucumbers. Ibn Isḥāq stated that bitter herbs were introduced among the Arabs during the smallpox plague of 569–70; the allusion must be to a drug, not to vegetables; Noldeke 219, Anm. 2, saw this. Perhaps the drug still used in Yaman is meant.

^{page 446 note 1} Malalas, pp. 434–5; the end of this work was mauled by an epitomist.

^{page 446 note 2} Obviously Arabs, Numan and Gafna; presumably sons of Harit, whose wife counted a Gafnah in her pedigree. Nu'man is strange, since it belonged to the Tunukh family, displaced by that of Saliḥ.

^{page 446 note 3} He avoided entering acknowledged Persian territory because Persia was at peace with the Roman Empire.

^{page 446 note 4} The kings of Hirah had Persian troops: Nüldeke, 83.

^{page 446 note 5} Nüldeke, 171, Anm.

^{page 446 note 6} See p. 461.

^{page 446 note 7} Nüldeke, 148–50.

^{page 447 note 1} ibid., 168.

^{page 447 note 2} CMH, 481.

^{page 447 note 3} Muaéon 59, 167Google Scholar, line 13.

^{page 447 note 4} Neither Ḥuğr nor Ḥariṯ is named in Ry 510, as a recognized feing would normally be.

^{page 447 note 5} There is no phonetic objection to the identification of ṯ lbt with the Thalaban of Theophanes. In Arabic the tribal name is spelt as often with fatha as with dhumma in the first syllable.

^{page 447 note 6} This view would incidentally involve the denial of any connexion between the campaign of era year 662 and the attack of Abramos on Persian territory described by Procopius. ‘ Wars’, I–II, xxviii, 11, contains no reference to any event later than 545; see LRE, 2, 420Google Scholar, note 2

^{page 448 note 1} Procopius, , ‘ Wars’, 1, 18, 16.Google Scholar

^{page 448 note 2} ibid., I, xxi, 1.

^{page 448 note 3} ύπό generally means ‘ just after ’, but also ‘ just before ’, ύπό vύkra.

^{page 448 note 4} The statement in BSOAS., 16, 38Google Scholar, may be based on ‘ Wars ’, 1, 20, 13Google Scholar, which deals with Abramos.

^{page 449 note 1} Malalas, pp. 457–9.

^{page 449 note 2} Then by ship to Adulis along the Arab shore, the regular route.

^{page 449 note 3} Either Ameritai or Auxumitai; the later reference to Elesboas is suspect.

^{page 449 note 4} øαkὶλιv. presumably metal with openwork ribs like rods.

^{page 449 note 5} Part of the Sassanian insignia, with an earlier prototype in Assyrian reliefs. Such ribbons adorn a baitulion in the relief published by Professor W. B. Henning in Asia Major, New Series, ii, plate x. The inscription, p. 174, no. 3, states that the relief was cut by ‘ townsmen of (the settlement named after) Am-wafa, son of Bod-Aqi, of the Bany Kuzai’. In the Aramaic of southern Irāq the name is equivalent to southern Arabian Am-wafay or Am-wafa.

^{page 449 note 6} That is, the Arab phylarchs of the Empire.

^{page 449 note 7} The first person indicates direct copying of the original text.

^{page 450 note 1} Mordtmann, J. H. in ZDMG., 35, 694.Google Scholar

^{page 450 note 2} The statement of Procopius that the promise was not carried out will also be correct. When negotiations for peace began, the Himyarite campaign was abandoned.

^{page 451 note 1} Institution, 323.

^{page 451 note 2} p. 432.

^{page 451 note 3} Nüldeke, 215.

^{page 451 note 4} BH, Syriac text, 49, a, b.

^{page 451 note 5} Nüldeke, 188, Anm.: a remarkable example of critical acumen.

^{page 451 note 6} The attribution of this form (in the Greek text in the accusative) to Procopius, in BSOAS., 16, 38Google Scholar, must be an accident.

^{page 452 note 1} See the Nau, Abbe in JA., 9 Serie, viii, 346 ff.Google Scholar, ix, 529. Nüldeke's review in WZKM., 1896, 160–1.Google Scholar

^{page 452 note 2} Mordtmann in ZDMG., 35, 700Google Scholar, criticizing Nüldeke, invoked the name of Bichard Bentley. Bentley proved, by cumulative evidence, that Phalaris was a late school exercise, and not a particularly good one; he was contradicting those who stated that it was one of the most ancient and admirable works of Greek literature. He did not claim to have proved that every letter preserved in an ancient book is a ‘ forgery’. He would have distinguished between the reproduction of Simeon's dated letter, and the ‘ letter ’ in it.

^{page 453 note 1} Chabot, J.-B., Chronique de Michel le Syrien, tome 2, 184.Google Scholar

^{page 453 note 2} i, xe.

^{page 453 note 3} ibid., Ii.

^{page 454 note 1} Budge, , Ethiopia, 1, 261.Google Scholar

^{page 454 note 2} Cosmas, Logos B, 73: p. 52.

^{page 454 note 3} The first five books were published separately, books VI–XII later in reply to adverse criticism. The eclipses dated by modern calculation to 547 are described at the beginning of book VI in an argument meant to refute attacks; book II must be earlier.

^{page 454 note 4} Cosmas, II, 101 C: p. 72.

^{page 454 note 5} Dittenberger, , II, no. 198, pp. 292–4.Google Scholar The date of this inscription is still uncertain. Conti Rossini connected it with Aphilas. The Kinaidokolpitai, ‘ men of Kogues’ Gulf’, correspond to the Kanraītai.

^{page 455 note 1} CIH 621: Chrestomathia no. 65: Mlaker, K. in WZKM., 34, 54–75Google Scholar: RES no. 2633.

^{page 455 note 2} von Wissmann—Hüfner, 92–3.

^{page 455 note 3} The first son named after an uncle; the second after the predecessor of Yusnf, and associated with Yazid, the deputy of Abraha.

^{page 455 note 4} bany is not used literally here; this Laḥayat Yarḥum should be the traditional founder of the family, not the brother of the qayl Šaraḥ-il Yaqbul, and not the son of Sarah-b-il Yakmul.

^{page 455 note 5} Including ksrn. Mlaker suggested that this tribe inhabited the Wadi Kasr, described in von Wissmann-Hüfher, 125 S. The writer of the Periplus heard that Eudaimon, Aden, was taken by Kaisar after the time of Chariba-el, i.e. Karibi-il Watar Yuhanim. Frisk, Periplus, 110–1, rightly joined others in rejecting emendation of Kaisar into Ilasar or anything else. Could a Dhu Kaisaran be meant ?

^{page 455 note 6} Abyssinians resident in Arabia, as opposed to Ḥabašat.

^{page 455 note 7} w i ḥbšn zrftn b rญ ḥmyrm: the Syriac zrīphūthā is an abstract, and could not be the object of a verb meaning ‘ to send ’. zrft is the Arabic zarāfatun, a small troop, here messengers.

^{page 455 note 8} khrgw mlk ḥmyrm w qwlhw hmrm w rḥbn: usually translated ‘ when they (the Abyssinians) slew the king of Himyar and his princes, men of Ḥimyar and Rahab’. If ḥmrm were in apposition to qwlhw, nunation would be required; the tautology after mlk ḥmyrm. is improbable. The distinctions ‘ (some) Himyarites ’, ‘ the men of Rahab ’, is due to the peculiar, but undefinable, importance of Rahab, cf. CIH 540/79–80, where the reading seems to be ṣlwhmw w r shmw w bty rḥbm. von Wissmann-Hüfher, 23, connect the name with Khuraibat Rahaba near Ṣirwaḫ.

^{page 456 note 1} K. Krumbacher, Geschichte der byzantinischen Litteratur, 42.

^{page 456 note 2} Nüldeke, 191, Ibn Isḥāq: 194–5, Hišam.

^{page 456 note 3} The word is naturally to be connected with Syriac serīqo, ‘ worthless’. The name of Abraha's second son, also Masrūq, is probably a pure Arabic laqab not the royal name.

^{page 456 note 4} This seems to be a play on the regional name Saṭiran, Fakhry II, no. 63, or the tribal Bany Saṭiran; in the pagan period a kabir of the Fayišan was named Laḥayat Saṭiran, RES 3913/1 and Fakhry II, no. 61. It cannot be connected with Sanatruces, the Parthian prince of Hatra who met Trajan: that name was discussed by Nüldeke, 500, and has been found at Hatrah by Safar, Sayid Fuad, Sumer, 8, 191–2Google Scholar, nos. 36 and 37, and IX, 19. Otherwise Arabische Frage, 490. Šanatir is not a place name, and none of the traditional explanations seems likely; could it be a laqab based on a combination of šana and šatir,‘ infamous scoundrel’ ?

^{page 456 note 5} Muséon, 59, 167Google Scholar, line 12. Marḥabum is presumably collective, Raḥab.

^{page 457 note 1} Institution, 341–6.

^{page 457 note 2} BH, Syriao Text, 43 b. In Akkadian as in Sabaean the root is karabu, in Arabic karama. Lidzbarski's assertion, Ephemeris, 1, 263Google Scholar, that m is ‘ original’, paid little attention to the phenomena. Alternation from b to m not only characterizes different languages of the Semitic family, but can occur within one and the same language.

^{page 457 note 3} One of the aḍ ḍawāhir: Masūdi, iv, 122.

^{page 457 note 4} Boot hyg.

^{page 457 note 5} Ry 63, RES 4069, discussed with different results by Ryckmans, M. J. in Institution, 245–50Google Scholar and Muséon 66, 337.Google Scholar

^{page 457 note 6} The word mahir was used in ancient Egyptian of an official scribe, and in texts from Ras aš Šamra of a household official of the goddess, Anat, ii, 15, 21.

^{page 458 note 1} Emendations proposed neglect the nature of the document.

^{page 458 note 2} See pp. 432–3.

^{page 458 note 3} Muséon, 66, 295–303Google Scholar: Ry no. 508.

^{page 458 note 4} This passive construction is parallel to the indefinite

^{page 458 note 5} udk khm rg m: the h must represent an adverbial ending in ā; khm is an unusual orthography for km, representing the long vowel; rg m is the Arabic raj i, a place through which water runs.

^{page 458 note 6} These figures may mitigate criticism of the credibility of the Syriac sources; they obviously derived their figures from accounts current in the Himyarite kingdom. There is no reliable criterion for any estimate of the population.

^{page 458 note 7} wbn ḏkyhw mlkn: this temporal adverb corresponds to the Phœnician locative adverb bn, Eshmunazar 5, ‘ therein ’. This use of deictic i is comparable to that in the compound preposition ln, ‘ from ’, where it anticipates a noun in apposition, and in the conjunction b dn, where it introduces a subordinate clause. Dr. S. Y. Bakr has proved to me that this corresponds to the -mā mukqffah of the Arab grammarians.

^{page 459 note 1} Moberg's discussion of the reading in BH, clxvii–clxix. The corruption has been recognized by M. J. Ryckmans.

^{page 459 note 2} BH, Syriac text, 43b.

^{page 459 note 3} See pp. 432–3.

^{page 460 note 1} BH, Syriac text, 7 b.

^{page 460 note 2} Ry 446, now 510, Muséon, 66, 307–10.Google Scholar

^{page 460 note 3} wtf: land charters were called wtf as being grants for ever.

^{page 460 note 4} ly mhn śb tm: mhn is surely an indefinite pronoun, ma and hn.

^{page 460 note 5} rq: ‘cliff’. If the tribal name !§ubaic is correctly read, more probably the ranges of Philby, Arabian Highlands, 109–10, than the hillocks mentioned ibid., 95, though they are called Arq al Subai.

^{page 460 note 6} lhm ḏndynhmw rbn: Ihm surely consists of the preposition with a pronoun, combined demonstratives, followed by the relative. The -i form of the verb cannot here be an infinitive, the usual explanation.

^{page 461 note 1} wyḥn: I take this form to be for yauḥinu, root wḥn, singular for pi. An unknown tribe could not be joined to the nations preceding.

^{page 461 note 2} Theodoros Anagnostes (Lector), ii, 58, quoted LRE, 2, 322.Google Scholar

^{page 461 note 3} See p. 446.

^{page 461 note 4} Muséon, 66, 333.Google Scholar

^{page 461 note 5} Invariably either Abyssinia, or Abyssinians in their own land. The order is therefore significant. Part of Arabia was included with the African territory, as it is named between Raydan and Saba.

^{page 461 note 6} Miscellanea Academiae Berolinensis, 97–127, especially 107–8. The counter-argument, Institution, 309–11, takes no account of the character of the inscriptions.

^{page 461 note 7} Fakhry i, p. 109, fig. 55, ii, pp. 46–9, no. 74.

^{page 461 note 8} Haṣbaḥ was associated by M. Hartmann with the name of the father of Abraha al Ašram. On the Aṣbaḥ and modern Ṣubaihi see Dr.Serjeant, R. B. in Muséon, 66, 126–9.Google Scholar

^{page 462 note 1} Clearly put in H. StPhilby, J. B., The Background of Islam, 116–9.Google Scholar

^{page 462 note 2} Andrae, D. Tor, Der Ursprung des Islams und das Christentum, 7–24.Google Scholar Contrast Margoliouth, D. S., Relations between Arabs and Israelites (Schweich Lectures), 59–83.Google Scholar

^{page 462 note 3} Texts published before 1945 were catalogued and discussed by Ryckmans, G. in Miscellanea Historica Alberti De Meyer, 194–205.Google Scholar

^{page 462 note 4} See pp. 432, 437.

^{page 462 note 5} I owe this information to the kindness of Professor B. Lewis who referred me to Ḥayyim Zeeb Hirschberg, Yisra'el ba-Arab: the i in Arabia from the fall of the Second Temple until the Crusaders. (Modern Hebrew.) Unfortunately I have not had access to this book.

^{page 463 note 1} See Table B. The effect of reckoning the era from 115 B.C. or 110 B.C. can be judged by shifting years marked E.

^{page 466 note 1} The last re-statement of this view known to me is in BH, xlv–xlviii. Detailed arguments, e.g. that of Moberg about the reference in the Koran to the Nağran martyrdoms, frequently break down. Ukhdud explains itself.

^{page 466 note 2} e.g. Masudi, iii, 122: ‘ there are many accounts about Quraiš, Ǧurhum, Khuzaa and the others of Maadd ’: iii, 391–2, the wars of Ghassān and Maadd.

^{page 466 note 3} Masūdi, iv, 236–241: vi, 137–55. Contrast the many cases cited in von Wissmann-Hümer which prove that tribes still occupy territory they held in pre-Islamic centuries.

^{page 466 note 4} von Kremer, Ueber die südarabische Saga, vi. von Kremer himself frequently anticipated points subsequently proved. So far as I can judge the references in the traditionalists are, by comparison, worthless.

^{page 466 note 5} Ibn Hišam, 963, quoted in Muir-Weir, 456, as ‘ curious’. The pagan sticks are figured in A. Grohmann, GōttersymboU und Symboltiere.

^{page 466 note 6} Published by ProfessorRyckmans, G. in Muséon, 67, 181–5.Google Scholar The monogram in nos. 2 and 10, read ḥb, should I think be regarded as that of the tribe Ḥabum, and compared with the monogram yz for Yazan. The interpretation as a statement, ‘ he loved ’, cannot be considered natural in this case; the occurrence of ḥb in graffiti not in holy places is another matter. Ḥabum was a section of Yazan. There were adherents of Hūd among the Jews and Christians there.

^{page 466 note 7} This seems to be the view taken in Muir-Weir, 51–2, and in Margoliouth, D. S., Mohammed, 131.Google Scholar

^{page 467 note 1} Ibn Sad, 117–8, quoted in Muir-Weir, 436. The attack implies that Jiddah belonged to Quraiš. The Abyssinians were not subjects of the nagaši, but settlers there, who frequently served as professional soldiers and the like; see Father Lammens, H., L'Arabie occidentals avant l'Hégire, 244–57.Google Scholar

^{page 467 note 2} Ostrogorsky, G., Geschichte des byzantinischen Staates, 48Google Scholar: ‘ wirtschaftlich und finanziell völlig zerr00FC;ttet’.