Navigation – Plan du site

The Gates of Himyarite Ẓafār

Les portes de Ẓafār l’himyarite
Paul Yule
p. 23‑35


À propos des paysages urbains de l’époque himyarite, les sources écrites et archéologiques sont rares. La citation d’Abū Naṣr al‑Yahāri par al‑Hamdānī au xe siècle concernant la configuration des portes de Ẓafār a été discutée par des chercheurs comme Faris, al‑Shamārī et Muhammad al‑Akwa‘. Il existe des contradictions dans la transmission des noms dans les trois versions du manuscrit, du fait des divergences dans la ponctuation. En se basant sur des cartes à grande échelle et des prospections archéologiques à Ẓafār, il est possible d’examiner à nouveau la question de savoir où se trouvaient les portes de la cité antique. Et bien qu’aujourd’hui, aucune de ces portes ne soit intacte, des restes de murs de fortification permettent de discerner leur emplacement originel. Le complexe de l’entrée principale de la ville se trouve au sud et est restée en service, bien que sous une forme modifiée et réduite, jusqu’à récemment. Malgré certaines contradictions entre les indices textuels et archéologiques, ce rapprochement entre textes et archéologie pourrait conduire à de nouvelles questions et de nouvelles réponses.

Haut de page

Texte intégral

  • 1 The author thanks Yūsuf ‘Abdullāh (DFG Mercator Professor at the University of Heidelberg), who dur (...)

1Historiographically speaking, a main problem of Himyarite history and architectural history (110 BCE –570/630 CE) is that no descriptions of the appearance of centres has survived1. The capital Ẓafār is such a case — with 110 hectares for the core area, one of the largest archaeological sites in Arabia. The few recorded building structures and finds from this rupestrian site require time, study and patience to place them in a larger historical context. It would be presumptuous to declare the known cultural remains there as wholly representative of specific definable time horizons, although in recent years the dramatically increase in excavated material gives a surer impression of its overall character. The lack of ancient written sources is partly attributable to preservation, but presumably also to the Ḥimyar’s lack of interest in describing art and architecture, compared to contemporaries, the Greeks and Romans as well as others coming to mind, with their voluminous writings. To speculate about the relative amount of the Greco‑Roman as opposed to Himyarite period descriptive literature in general, is to compare two unequal historic entities, conditioned in terms of time span and area. In any case, the former outclass the latter (OSA =Old South Arabian) in quantity and in the amount of information which illuminate ancient life and social thought. OSA texts which would correspond in detailed scope, length and intensity to the rich and nuanced descriptions and anecdotes of Pliny, Pausanius or Vitruvius are simply lacking. In terms of the social context of the buildings and patrons, terse OSA texts render little more than a mere shadow of the real architectural situation relative to contemporary Greco‑Roman texts. Even OSA church records are largely silent regarding their own architecture, preferring to dwell on church history and hagiography (Cameron, 1982: 279). Although the Vita of Gregentius mentions the building of three churches in Ẓafār following the end of the war with Aksum, this source seems to be a later compilation (oral information, A. Berger, Munich).

2The OSA renowned Ghumdān palace (‘Abdullāh, 1990, 103–125; Lewcock, 2005, 74-75) and the cathedral both in Sanaa are exceptions, the former being known exclusively from diverse, imaginative Arabic sources, for example al‑Hamdānī (Müller, 1986 : 141–143). The latter is mainly known from a few OSA epigraphic texts (Ir 11, Ir 18, CIH 429, Ja 577) and from early medieval Arabic descriptions. Owing to the sketchy content of Himyarite sources, in terms of architectural history, the main task of the archaeologist amounts to little more than to venture guesses on the size and date of a given monument, or turn to buildings of earlier or later periods as points of comparison regarding the possible appearance. Recent research again raises the question of the development of the ancient city. Did it grow in size and elaboration up to the late period (c. 523–570/630 CE) or, following the Aksumite conquest, did the victors simply occupy what was erected in the centuries before ?

1. Map of places cited (Tactical Pilotage Chart, sheet K‑6A, edition 3‑GSGS).

1. Map of places cited (Tactical Pilotage Chart, sheet K‑6A, edition 3‑GSGS).

3The work of a research team based at Heidelberg University illuminates Ẓafār’s history and more particularly its architectural history. Inscriptions of the 3rd–4th centuries recount the refurbishing of the city defences, and mention « bastions and the city wall, which on the entry to the city Ẓafār were located… Ẓafār with the recovery of houses which…and the wbt and the gate and the city trench, and the bastions of the entry of the city Ẓafār… » (Excerpted from zm2263+zm2262+zm2264, Sima 2002). The urban area measures some 1200 x 800 m, the inner third of which lies inside the inner city wall. Magnetometer investigations of the remains of the southern gate area show images of building structures, but these are difficult to interpret (Yule, in press 2).

4Regarding the topography of Ẓafār, one notes first its high altitude, rising to more than 2800 m above the surrounding valleys and plains. Except for trails, this centre is largely blocked off from all directions by mountains except to the agriculturally rich al‑Ḥaql plain a few kilometres to the west (Fig. 1). In place of a dirt road used in antiquity, today, a fine tarmac road, which the Spanish government financed in 2002, links Ẓafār to the towns al‑Ribāṭ and Kitāb in that plain. Paradoxically, Ẓafār does not lie along the assumed antique north‑south trade routes which lie further east on the zone between desert and highlands.

5Other texts give further information about the fate of the city. In 523 Yūsuf ‘As’ar Yathar reports burning the church and killing the Abyssinians in Ẓafār, but clearly does not mention destroying the city (Ry 507/4, Ry 508/3, Ja 1028/3). Since in Ry 507 he explicitly discusses the destruction of other fortifications aside from the ones in Ẓafār, one can assume that the latter he left intact. A main question remains whether the Aksumite king, Abraha, chooses to move the capital as a result of the stamping out of Sumūyafa‘ Ashwa‘’s revolt in 535, as a result of the insurrection of 547, or as a general reorientation in the 560s. A possible later destruction of the capital makes more sense than one assumed prior to this during the Aksumite‑Himyarite war, after which, the traditional capital is at least partly still in use. In any case, in the 560s Sanaa undisputedly superseded Ẓafār as the capital.

2. Ẓafār in the surrounding landscape (N. Carstensen, 2005).

2. Ẓafār in the surrounding landscape (N. Carstensen, 2005).

6Subsequently, in light of 1500 years of war, weathering and finally stone robbing, the preservation of the once proud gates of Ẓafār, not surprisingly, is hardly worth mention. Until recently there was no archaeological evidence for building activity in the late period (523–630 CE), the emphasis period of a research project currently being carried out by the Heidelberg team. Despite Ḥimyar’s great size, military power and economic influence, after losing the war in 525 to Aksum and an insurrection in the 6th century against those rulers, its power is definitely spent and in 630 officially succumbs to a new power — Islam.

7Al‑Hamdānī’s 10th century brief description in al‑Iklīl, is one of the few to discuss Ẓafār’s cityscape: « Ẓafār lies on the top of a mountain in the north of Qitāb, close to the city of the Sukhtiyīn which is Mankath. According to Abū Naṣr al‑Yahārī (same source), Ẓafār had nine gates: as listed, the Walā’ (a place known as « friendship »), Aslāf, Kharaqah, the Manah (navel), the Hadwān, the Khubān, the Ḥawrāh, the Ṣayd also called Sumārah, and the Ḥaql gate. It is said that the waters of Ḥaql are drawn from a distant place. » (Faris, 1938, 22-23). Even if N.A. Faris did not recognize them in the 1930s, at a time when few maps and no gazetteers of place names for the Yemen were yet available for study, today these place‑names are well‑known in the area surrounding Ẓafār. Contradictions in the correct rendering of the names lies in that in the three Arabic manuscripts which contain the text, the pointing is erratic. It has been long suspected that the gates of Ẓafār face toward other outlying places, but an update on this matter seems appropriate in light of new information. Muḥammad al‑Akwa‘’s edition of al‑Iklīl (2004) proposes some names for the gates and their name‑sakes.

Historic‑geographical notes on the names of the gates

8a. Walā’: First of the gates of Ẓafār which Abū Nar al‑Yahārī mentions, is probably the main gate. For the position of this gate there is no external corresponding place‑name. The term means perhaps the « first gate ».

9b. al‑‘Aslāf (Salāf): In the direction of Iryān (14°16’N; 44°12’E) lies a place called Najd al‑‘Aslāf (14°18’N; 44°17’E).

10c. Kharaqah, mentioned with other areas near Ibb. In the census of Yemen cf. al‑Yesah. Akwa‘ reads this name as « Kharafah » (Akwa‘, 2004: 53). However he village al‑Khalaqah lies nearby (14°25’N; 44°28’E). The liquid r could conceivably be replaced by the liquid l.

11d. Man’ah: al‑Akwa‘ reads this name Mābah (14°17’N; 44°16’E), a village on a mountain in al‑Ḥārith in the direction of Ba‘dān district (sic). Another possibility is the town of Māwah (14°14’N; 44°20’E), 2 km north‑west of Ẓafār.

12e. Hadwān: al‑Akwa‘: town in the mountains of Iryāb (14°09’N; 44°19’E).

13f. Khubān area lie between the al‑Raẓmah landscape (centre: 14°12’N; 44°27’E) and al‑Saddah (14°07’N; 44°25’E). It is one of the easiest names to localise. Khubān occurs in Sabaic texts (Robin/Brunner 1997).

14g. Ḥawrāh means the « settlement », and lies south‑west from Ẓafār. Not unexpectedly, a number of settlements in the Yemen bear this name.

15h. Ṣayd (14°11’N; 44°16’E) pass (old name: Sumārah pass), toward Ibb (14°12’N; 44°16’E). Ṣa’idam occurs in Sabaic sources (Robin/Brunner, 1997). A Turkish fort is preserved nearby.

16i. Qa‘a al‑Ḥaql (‘the plain of al‑Ḥaql’). Today a major agricultural area centring on Kitāb and al‑Rubāṭ (60 000 tons of agricultural produce per year). It does not simply mean ‛field’ in Arabic, but from other sources it is known to refer to a particular area.

17Turning closer to the gate‑names, Table 1 localises some of them on the basis of nearby present‑day places (cf. also Fig. 2 and 3).

Table 1. Summary of the names of the gates of Ẓafār (14°12’N; 44°24’E) listed in al‑Iklīl giving the name of orientation for the gates.

  • 2 K, M, and B refer to the three manuscripts which N.A. Faris consulted for his study of al‑Iklīl.

Gate name after Faris

Gate name after al‑Akwa‘

Description of place

Direction of place

Km from Ẓafār

Coordinates, centre

Other sources

a. Walā’


no similar place‑name

b. Aslāf


near Yarīm



14°18’N; 44°17’E

c. Kharaqah


near Ibb



Khalaqah: 14°25’N; 44°28’E

K, M, B2 Ṣifah p.  68 line 4

d. Man’ah





14°17’N; 44°16’E

K, M, B

e. Hadwān


in mountains of Iryāb



K, B

f. Khubān





c 14°15’N; 44°30’E

K & M, Buldān ii, p.  397, 373/4, al‑Shamāri 2004

g. Ḥawrāh





14°09’N; 44°21’E

cf. Buldān ii, p.  359

h. Ṣayd = Sumārah

Ṣayyid (Ṣa‘ad) = Sumārah




14°11’N; 44°16’E

Ṣayd: K & B, see Buldān ii, p.  441; Ṣifah, p.  125; Sumārah: see Buldān iii, p.  441;

i. Ḥaql





14°12'N; 44°20'E

see Buldān ii, p.  299, line 11

Archaeological evidence for gates at Ẓafār


18Although none of the city gates are intact, remains of the fortification walls shed some light on where these once lay. The main entry complex to the city lay to the south and remained in service, although in a very much changed and reduced form, until recently. In Ẓafār South on the slopes known as al‑Ḥayfah and al‑‘Uwār, present‑day roads communicate with the west, the north‑east and south‑east (Fig. 5). Other potential access points for the main gate make less sense in terms of the rough topography. Roads follow topographically defined ways to communicate with the outlying settlements. In Ẓafār South, a stone‑paved sloping corridor (Fig. 3) which leads to the present‑day village was widened in February 2004 and in February 2006 and its archaeological vestiges thereby denatured. Its pavement stones seem to have been antique (Fig.6). Other road widening activities in 2000 and 2005 here took place where the main entrance complex is expected. Although no gates as such exist anymore, ancient walls and inscriptions still do, partly not yet excavated. In the 200 m distance from the slope from al‑Ḥayfah into the valley of al‑‘Uwār, where the roads to the north, east and south come together, is the most likely place for the main gate complex.

3. Quickbird satellite image 01.01.2004 of Ẓafār showing the suspected positions of the city gates. The four names in the lower left (Hadwān, Ḥawrah, Kharafah and Mābah) are not precisely assignable to the gates.

3. Quickbird satellite image 01.01.2004 of Ẓafār showing the suspected positions of the city gates. The four names in the lower left (Hadwān, Ḥawrah, Kharafah and Mābah) are not precisely assignable to the gates.

19Building inscriptions came to light here (ZM 2263+2262+2264, Sima, 2002; Sima, in press), which describe the reinvesting of the city defences (see above), « …in the year 347 of the era of Mabḥash, of the son of ‘Abḥūsh (=232 CE)... (in) the two months dhMdh’n ». This text came to light in al‑‘Uwār, 200 m south of Qarīyat Ẓafār around 1980 (personal communication C. Robin). What archaeologists dub « Ẓafār South », contains the main entry complex described in the inscriptions. Enormous stones in the southern slope of al‑ Ḥayfah bear witness to the once‑great casemate walls. Underneath one of these, wall z189 (Fig. 5), a 14C determination (1596+37 BP) of cal 392–550 CE at 95.3% probability (lab. no.: Hd‑25322) suggests major expansion of the southern city defences toward the end of the empire period (270–523 CE).

4. Geomagnetic prospection at Ẓafār.

4. Geomagnetic prospection at Ẓafār.

20In 2004 geomagnetic survey (Fig. 4) was conducted in al‑‘Uwār in the hope of elucidating the position of the city defenses (Yule, in press 2). Unfortunately, these yielded no concrete data to corroborate the visible negative patterns of growth of the vegetation here. Gates probably pierced the heavy walls which flank both sides of the E–W road in al‑Ḥayfah. Fig. 5 shows the situation as mapped in 2002, since then the entrance to the village and the downward slope toward the east was widened in 2003 and 2006. Recognisable are three ruined fortifications, two above in al‑Ḥayfah (context z184 and z185) and one below in al‑‘Uwār (z179). ENE‑oriented walls connected these tower‑like structures, which were still barely visible in 2002 before the first bulldozing took place. In 2006 the road was paved, the ruins covered and destroyed. In the valley, the ENE–WSW walls have been considered a dam, which they seem to have become once their original function was lost. This fortification / dam is the only part of Ẓafār which seems plausibly immediately post‑Himyarite, to judge from the interpretation of the architecture (Yule, in press 1). Three of the gates must have been located here in order to communicate with the nearest settlements (gate nos. 1‑3).

5. Plan of Ẓafār South, state: 2000 (M. Schupp & K. Thiele).

5. Plan of Ẓafār South, state: 2000 (M. Schupp & K. Thiele).


21Ẓafār South includes the anciently fortified Jabūbat al‑Lagīyah slope (Fig. 3), which would not require a major gate of its own. In the saddle between Ẓafār South and the Ḥuṣn Raydān would be a logical place for a gate which leads to the east (gate no. 4). At mid length of the north‑south oriented settlement a way is visible here which may correspond with a Himyarite one. Glaser noted this « chaussée » in the late 19th century. Just below the eastern slope, fields have been built. But the road proceeds east directly to the neighbouring al‑‘Arāfah. Just where the gate lay is debatable. But a well cut from the living rock just at this place where it is expected is probably Himyarite work (Fig. 7). Wells are a common feature beside roads and gates near cities.

22Further north, between the Ḥuṣn Raydān and Raydān North (al‑Gusr) lies another saddle which appears to contain the remains of a large gate structure (Fig. 3). While the western access appears to lead to roadways, in the other direction no trace of such can be found. If a gate existed here, then it was a small one (gate no. 5).

6. Entrance to Qarīyat Ẓafār (context z190) to the north‑west before its destruction as a result of road‑widening in February 2004.

6. Entrance to Qarīyat Ẓafār (context z190) to the north‑west before its destruction as a result of road‑widening in February 2004.


23At the north‑eastern end of the al‑Gusr defences two stone abutments have been fashioned into a gate post some 4 m in width (Fig. 3). Again here, unfortunately there are no obvious traces of a large road either to the settlements to the north and north‑west, nor to those to the north‑east (gate no. 6).


24Gate no. 7 (Fig. 3) is not excavated but appears on the surface to be a gate. It faces to the west and leads to a way which zig‑zagged down the hill. Traces of this are visible, mostly stone debris (Yule et al., in press 3).

25Opposite the proposed gate on Ẓafār eastern flank lies an area with the name Bāb Sūq al‑Layl (gate no. 8, Fig. 9) which means the gate of the night market. Today a stairs winds up the slope into what is taken to be the main part of the ancient city, al‑Jāḥ, a common place‑name in the Yemen. Given the long distance between the gates which more clearly existed (especially gates 1–3, 6 and 7), this seems a likely place for a gate. Extensive cultivation has removed all traces of an east–west access until about 500 m to the west in al‑Danān. Here scraps of a paving associated with water troughs and graves suggest traces an access point.

7. Well z057, on the eastern part of Ẓafār.

7. Well z057, on the eastern part of Ẓafār.

26There are no other obvious sites at Ẓafār where one might expect a gate. Over long distances the walls are steep and unsuited for an entrance. A small entrance (z015) is preserved on the eastern flank of the Ḥuṣn Raydān, but it is hardly a major entrance (Yule, in press 1) mentioned in the description.


27To reconstruct the positions of Ẓafār’s city gates, one must reconcile textual and archaeological evidence which do not concur at all points. A first difficulty is that nine city gates named in the text but only seven or eight are archaeologically manifest at the present stage of research. While none of these are excavated, those especially in the south are partly visible on the surface. The clearest, least difficult gate name equates with the al‑Ḥaql plain, which lies a few kilometre directly to the west. It might correspond to the gates 1, 7 or 8, but for one consideration: Since the Walā’ gate is mentioned first, this is probably the most important one. It probably lies in a part of site today known as al‑Ḥayfah, which itself is a well‑known place‑name in the Yemen, for example the present‑day seat of the district of Arḥab to the north of Sanaa. The Sabaic place‑name ḤYFN (al‑Sheiba, 1987, 26) shows the root. Also today, Ḥayfān is a village name in Taez governorate. Thus, this is a potentially early name for the area in which the main gate complex is located. Partial confirmation of the position of the ancient main gate complex comes from the position of the main road and main access to present‑day Ẓafār which proceeds from the south gate toward the west, to judge from the preserved road remains. The other directions of the southern gate complex lead to lesser settlements and correspondingly are less suited for this gate attribution.

8. Ẓafār/al‑Sha‘ūb on the left and Bāb Sūq al‑Layl on the right, view to the NE.

8. Ẓafār/al‑Sha‘ūb on the left and Bāb Sūq al‑Layl on the right, view to the NE.

28The Khubān area lies to the east of Ẓafār and could correspond with the gates 2–6 on this side of the settlement. Since the appearance of the inconspicuous and unexcavated ‛gate’ 5 is essentially unknown, this structure is questionable, and we can discount it as a possibility. Gates 3 and 4 also face in this general direction are more likely candidates for the Khubān gate.

Gate 7 faces al‑‘Aslāf, which lies to the NW.

29The gate names Mābah, Hadwān, Ḥawrah and Kharafah are names which can be found in the vicinity, but which do not correspond with the positions of the city gates. The reasons for the contradictions are unknown. It is clear from different contexts that place‑names and different language fossils are uniquely conservative in the Yemen. At least several of the gate names can be clearly identified.

Haut de page


Abdullāh Y., Awrāq fi tārikh al‑yaman wa ’athārih, Bayrūt, Dār al‑fikr al‑mu‘āṣir, 1990.

al‑Akwa‘ M., « Abī Muḥammad al‑Ḥasan bin Aḥmad bin Ya‘qūb al‑Hamdānī », al‑Iklīl, vol. 8, Ṣan‛ā’, 2004.

Cameron A., « Wandering Poets. The Empress and the Poet », Yale Classical Studies 27, 1982, 217–289.

Faris N. A., The Antiquities of South Arabia being a Translation... of the Eighth Book of al‑Hamdani's al‑Iklil, Princeton, 1938.

Lewcock R., « Early and Medieval Sana’a.The Evidence on the Ground », Ṣan‛ā’ al‑ḥaḍārat wa‑ al‑tārikh, vol. 2, Ṣan‛ā’, 2005, 71–85.

Robin Chr. & Brunner, Uli, Map of Ancient Yemen, Munich, 1997.

Müller W., « Ancient Castles mentioned in the Eighth Volume of al‑Hamdani’s Iklīl and Evidence of them in pre‑Islamic Inscriptions », in Y. ‘Abdullāh (ed.), Al‑Hamdānī a Great Yemeni Scholar, Ṣan‛ā’, 1986, 139–157.

Müller W., « Himyar », in Reallexikon Christliches Altertum, vol. 15, Stuttgart, 1991, 303–331.

Nebes N (ed.), Katalog der Inschriften aus dem Museum Z.afār, in press.

Sima Al., « Die Geschichte der Stadt Ẓafār », lecture held in Bamberg on 29.06.2002.

al‑Shamāri M. Ḍ., Lahjat Khubān, Ṣan‛ā’, 2004.

al‑Sheiba Abd., « Die Ortsnamen in den altsüdarabischen Inschriften », ABADY 4, 1987, 1–62.

Yule P., « Ẓafār, Capital of Ḥimyar, First Preliminary Report: 1998 and 2000 », in press, in: ABADY 11, 2007, 479–505.

Yule P., « Fourth Preliminary Report: 2004 », in press, in: ABADY 11, 2007, 530–547.

Yule P., « Sixth Preliminary Report, February–March 2006 », in press, Zeitschrift für Orient‑Archäologie 1, 2006.

Haut de page


1 The author thanks Yūsuf ‘Abdullāh (DFG Mercator Professor at the University of Heidelberg), who during the course of several discussions, illuminated the Arabic part of what follows. The author is responsible for the interpretations given below. He also thanks the DFG for their support of our project and the referees of the Chroniques yéménites for numerous corrections and suggestions. Where not otherwise cited, the writer furnished the figures. Finally, Albrecht Berger (Munich University) generously shared his research on the Vita of Grigentius. I thank Cornelia Ruppert for providing information with regard to the present‑day topography of the area. Basic introduction to Ẓafār: Müller, 1991.

2 K, M, and B refer to the three manuscripts which N.A. Faris consulted for his study of al‑Iklīl.

Haut de page

Table des illustrations

Titre 1. Map of places cited (Tactical Pilotage Chart, sheet K‑6A, edition 3‑GSGS).
Fichier image/jpeg, 80k
Titre 2. Ẓafār in the surrounding landscape (N. Carstensen, 2005).
Fichier image/jpeg, 28k
Titre 3. Quickbird satellite image 01.01.2004 of Ẓafār showing the suspected positions of the city gates. The four names in the lower left (Hadwān, Ḥawrah, Kharafah and Mābah) are not precisely assignable to the gates.
Fichier image/jpeg, 80k
Titre 4. Geomagnetic prospection at Ẓafār.
Fichier image/jpeg, 96k
Titre 5. Plan of Ẓafār South, state: 2000 (M. Schupp & K. Thiele).
Fichier image/jpeg, 28k
Titre 6. Entrance to Qarīyat Ẓafār (context z190) to the north‑west before its destruction as a result of road‑widening in February 2004.
Fichier image/jpeg, 40k
Titre 7. Well z057, on the eastern part of Ẓafār.
Fichier image/jpeg, 44k
Titre 8. Ẓafār/al‑Sha‘ūb on the left and Bāb Sūq al‑Layl on the right, view to the NE.
Fichier image/jpeg, 48k
Haut de page

Pour citer cet article

Référence papier

Paul Yule, « The Gates of Himyarite Ẓafār », Chroniques yéménites, 14 | 2007, 23‑35.

Référence électronique

Paul Yule, « The Gates of Himyarite Ẓafār », Chroniques yéménites [En ligne], 14 | 2007, mis en ligne le 08 octobre 2007, consulté le 18 avril 2014. URL :

Haut de page


Paul Yule

Heidelberg University

Haut de page

Droits d’auteur

Tous droits réservés

Haut de page
  • Logo Centre français d’Archéologie et de Sciences Sociales de Sanaa
  • Logo Institut de recherches et d'études sur le monde arabe et musulman
  • Logo Ministère des Affaires étrangères et européennes
  • Logo CNRS - Centre National de la Recherche Scientifique