Introduction

1This paper1 explores relations between southern Arabia and China through the trade of a specific commodity: frankincense. Frankincense is the resin exuding from Boswellia trees. Twenty-two species are known, but the most famous is Boswellia sacra which grows in the south of the Arabian Peninsula in the regions of Hadramawt, Mahra and Dhofar.2 Other species growing in the Horn of Africa (e.g. B. carterii) give resins that have been traded until now (fig. 1).

Fig. 1 – Map showing the distribution area of Boswellia, Commiphora and Hymenae verrucosa (S. Le Maguer‑Gillon).

2Frankincense from South Arabia has been traded since ca the 9th century BC. It is closely associated with polytheistic religions in Arabia, Egypt, Mesopotamia, the Levant, Greece and Rome. Trade between South Arabia and the Mediterranean increased thanks to the domestication of the camel to carry goods at the turn of the 1st millennium BC.3 From the 2nd century BC, maritime trade developed significantly as the construction of ports shows.4 Qāni’ (modern Bi’r ‘Alī, in the province of Hadramawt, Yemen) was founded in the 1st century BC and the oldest reference to Aden (Yemen) dates to the 1st century AD in the Periplus Maris Erythraei, a navigation handbook.5 Frankincense trade with the Mediterranean world declined in the 3rd‑4th centuries AD as the use of incense was condemned by Christian authorities. The idea of burning incense as a sacrifice to God was strongly condemned by Early Christian writers.6 It was finally reintroduced in funerary practices in the 4th century, and the earliest archaeological evidence for the use of censers in churches dates from the 6th century.7

3Although the demand for frankincense decreased in the Mediterranean world as it was not used by the elites anymore, the trade in frankincense remained of economic importance to southern Arabia throughout the Islamic period. Indeed, this product was traded to the Far East. Frankincense demand in China increased as Buddhism penetrated during the Tang dynasty. Many Buddhist rites require the use of incense, to delimit a sacred space or to accompany the prayers.8 The emperor himself needs the scented smoke before making any decision. It symbolises the divine presence and the rising smoke from an incense burner allows the communication with the gods. Trade with the frankincense-producing countries therefore thrived. From the 7th century onwards, frankincense remained an important aromatic product in the Islamic world, used as incense and in perfumery, but it was considered inferior to scented product from South-East Asia such as camphor and musk.9 Meanwhile, China continued to exert huge demand for frankincense.

4The period between the 8th and 14th century corresponds to the rule of the Abbasid caliphs (750‑1258) and the rise of Turkish dynasties such as the Saljūq and the Mongols in the medieval Muslim world. The Mongol conquests in the 13th century overthrew the Abbasid dynasty and opened a new political era, although the ‘Abbasid power was diminished since the 10th century. In China, the Tang (618‑907) and the Song (960‑1279) periods are characterised by a flourishing trade, thanks to a peaceful policy and development of maritime trade. But in 1433, this policy ended when the Chinese expeditions were stopped.10

5Did frankincense trade play an important role in the development of ports or other maritime infrastructures in southern Arabia between the 8th and the 14th century? As I will show, the study of the trade of this particular good offers a different perspective for the understanding of the development of ports along the southern coast of Arabia from Aden in Yemen to Ḥāsik in the Sultanate of Oman.

6A harbour is defined in the Oxford English Dictionnary as a place where vessels can be stored or take shelter. They can be natural and surrounded by land, or artificial, constructed with breakwaters, sea wall or jetties. According to dictionaries, a port is an artificial construction on a sea, river or lake shore where vessels are loaded and unloaded. An anchorage is a shelter where ships are anchored. Each term refers to the concept of protection and shelter, independent of hierarchy.11 The port is an interface between land and sea for men and goods. The infrastructure is more or less developed depending on the level of attention paid to ensure the loading or unloading of goods. The medieval Arabic words for port are “mīnā’”, “marsā”, and “furḍa”. The first, “mīnā’”, is the generic word for port or harbour.12 It is a non-Arabic word in origin, probably derived from the Greek “limen” or from ancient Egyptian. This word is broadly used in Arabic texts from the 9th century onwards. “Marsā” is generally translated by anchorage and can refer to a natural harbour. “Furḍa” refers to the economic activities of a port as it also means customs. It is difficult from these definitions alone to organise these maritime structures into a hierarchy. However, the descriptions given by Arabic geographers help us to understand their level of integration in the Indian Ocean trade. In addition to these descriptions, it is interesting to look at the Chinese textual sources which describe the trade relations with these ports. The fact they are mentioned in these texts testify to their role in the long-distance trade. Finally, the archaeological evidences help us to determine the nature and the importance of these ports (size of structures and imported material).

The frankincense trade between Arabia and China in the written sources

The ports involved in the frankincense trade according to Arabic texts

7In the medieval Arabic written sources, “lubān” and “kundur” both refer to frankincense. The first word comes from the ancient South Arabian word “l‑b‑n”, while the second is a Persian word from the Sanskrit “kundurika”. Information regarding the origin and trade of frankincense can be found in different types of Arabic medieval texts, from the lexicographical works to geographical works.13 Most agree that the main area for frankincense production and trade was al‑Shiḥr, a port-city located in the coast of the Hadramawt province in Yemen. The other ports or harbours involved in frankincense trade according to the textual sources are Aden and Mirbāṭ (fig. 1).

8From the 9th century, al‑Shiḥr is generally referred to as “the land of frankincense” in Arabic written sources. Abū Ḥanīfa al‑Dīnawarī says in his Kitāb al‑nabāt wa‑l‑shajar (The book of plants and trees, 9th century) that “frankincense only grows in the land of al‑Shiḥr”.14

9The Akhbār al‑Ṣīn wa‑l’Hind (Accounts of China and India) is the oldest travel book written in Arabic that has reached us.15 It gathers two works written by different authors. The first, written in 851‑852, is anonymous, although attributed to Sulaymān al‑Tājir (Sulayman the Merchant), who was at least one of the author’s informants.16 The author of the second book, written in 915‑916, is identified as Abū Zayd al‑Sīrāfī.17 This work represents a very interesting source regarding the trade in the 9th and 10th centuries as it is a first-hand documentation from traders themselves. The first book focuses on the trade with China and offers valuable information about the Chinese society in the 9th century. There is little information about frankincense trade, except that frankincense was among the commodities exported to China.18 The second book gives more information regarding the products and places of trade and about frankincense. Abū Zayd al-Sīrāfī writes: “On this sea that extends to starboard of India, from the view point of a ship leaving Oman, lies the land of al‑Shiḥr. In it are the places where frankincense grows […]”.19

10In the second half of the 9th century, Ibn Ḫurradāḏbih describes in his Kitāb al‑masālik wa’l‑mamālik (Book of itineraries and kingdoms) the coastal stations of the route leading from ‘Umān to Mecca: “Farak; ‘Awkalân; the anchorage of Habât [?], as‑Schihr, the land of frankincense. A poet said: ‘go to the Schihr and leave Omân. If you don’t find dates there, you’ll find frankincense’”.20

11In the 10th century, according to al‑Mas‘ūdī (d. ca 956‑957) in his Kitāb Murūj al‑ḏahab (The golden meadows), frankincense (“kundur”) is one of the only resources in the region of al‑Shiḥr.21

12At the end of the 10th century, al‑Muqaddasī quotes al‑Shiḥr several times in the book Aḥsan al‑taqāsīm fī ma‘rifat al‑aqālīm (The best division for knowledge of the regions). Al‑Shiḥr appears there as an important port for fishing and a frankincense production center. He uses the word “mīnā’”: “Al‑Shiḥr is a town on the sea, an important port [mīnā’] for enormous fishes, which are exported to ‘Umān and ‘Adan, thence to al‑Basra, and to the towns of al‑Yaman. Here are trees of which the resin is frankincense [kundur]”.22

13The geographer Ibn Ḥawqal (second half of the 10th century) describes in his Kitāb Ṣūrat al‑arḍ (The face of the Earth) al‑Shiḥr as an important center of the production and trade of frankincense: “The Mahra region has Shiḥr as its capital […]. Incense used in all parts of the world comes from there: the homes of the inhabitants are full of them”.23 Here, al‑Shiḥr is the name for both the port-city and the area surrounding.

14Abū ‘Ubayd al‑Bakrī (11th century) is the author of a Kitāb al‑masālik wa’l‑mamālik (Book of itineraries and kingdoms) written around 1068 and in which he reports the specificities of the Arabian Peninsula. Among the remarkable products he describes is frankincense (“lubān”) which “is only found in Yemen, Shiḥr and Ḥaḍramawt, from where it is exported among others to India, China and Khurāsān”.24 He also mentions myrrh which, similarly, “exists only in Yemen and the best comes from [the island] of Suqutrā”.25

15Ibn Khaldūn’s (1332‑1406) Kitāb al‑‘ibar (Universal history, started in 1375) compiles information from the works of al‑Mas‘ūdī and Ibn Ḥawqal, among others. He presents al‑Shiḥr as a territory and not only as a city. He describes the resources of the region, among them frankincense and ambergris:

Ash‑Shiḥr is, like Ḥijāz and Yaman, one of the kingdoms of the Arabian Peninsula. It is separate from Ḥaḍramaut and ‘Oman. Ash‑Shiḥr is so named after its capital. […] The country is also known as that of the Mahrah […]. Ash‑Shiḥr is sometimes conjoined with ‘Oman, but it is contiguous to Ḥaḍramaut and it has been described as constituting the shores of that country. It produces frankincense [lubān; oliban] and on the sea-shore Shihrite ambergris is found.26

16The lexicographical and geographical works concur that frankincense originated in Yemen, and more precisely al‑Shiḥr. They also tell us that it was the main source of income of the region of al‑Shiḥr, and of the port-city itself, along with myrrh and ambergris. These aromatic goods were exported to eastern Asia, especially China. Administrative documentation from the Rasūlid period (1228‑1454) confirms that frankincense was an important source of income for the city.27 The Kitāb al‑irtifā’ (Book of income), a manuscript dated 1296, indicates in the list of income of al‑Shiḥr (f^o 89, v^o) that the price of frankincense (“lubān”) was fixed by the Rasūlid administration for 7 ½ dinars the “bahār” (=120 kg).28 This administrative document confirms that the trade of frankincense was closely monitored by Rasūlid rulers.

17Aden was an active port involved in the spice trade long before Islam known as Eudaimôn Arabia.29 The medieval city of Aden is only known thanks to the study of written sources. G. Ducatez carried out a historical study of it from the account of Ibn al‑Mujāwir in the 13th century which he partially translated.30 Two recent theses have shed important light on the topography of the city and on the organization of the trade. The first is that of R.E. Margariti (2007) who, using documents from the Cairo Genizah, reports on commercial life in the port of Aden between the 11th and 12th centuries. The second is the result of research by E. Vallet on Rasūlid administrative archives between the 13th and 15th centuries.

18In the Abbasid era, Aden appeared to be an important port in Yemen, involved in the Indian Ocean trade up to China. This is what al‑Ya’qūbī tells us in the 9th century: “Aden, port of Sanaa, where ships from China, Salāhiṭ and Mandeb come from”.31 At the end of the 10th century, al‑Muqaddasī qualifies Aden as “furḍa” of Yemen and presents this port as the junction point between the commercial route from China and the Maghreb: “Adan is a splendid, flourishing, populous town, fortified, and with a pleasant climate. It is the corridor of al‑Sīn, the seaport [furḍa] of al‑Yaman, the granary of al‑Maghrib, an entrepôt of various kinds of merchandise”.32

19The word “furḍa” has several meanings. It can be synonymous with port: in this case, al‑Muqaddasī insists that Aden was the most important in Yemen. Or he used the word “furḍa” in the sense of administrative or customs, highlighting Aden’s administrative and commercial role in Indian Ocean trade.33

20The “furḍa” building (customs office), erected in the 12th century, was located by the sea to facilitate the transport of goods from ships to this warehouse. There, the products were consigned, and taxes levied and then the goods were either sold in town or exported again.34 Near the “furḍa”, markets allowed the numerous merchants, coming from Egypt or India in particular, to exchange the contents of their cargoes. Dār al‑Sa‘āda and Dār al‑Ṭawīla probably received most of the flow of goods.35 This complex reflects the image of an important commercial and administrative district in Aden, centralising goods from the Indian Ocean and the Red Sea. Among these goods were spices in general and frankincense, sold not only by Arab merchants but also by Jewish merchants. The latter played an important role in the trade between Egypt and India via Aden. These transactions reached us through the texts of the Cairo Genizah. The merchant Abraham Ben Yijū was from al‑Mahdiyya in Tunisia.36 He has traveled to Egypt, Aden and to India. He specialized in import-export and owned a bronze tableware factory, activities which enriched him. He was also considered by his contemporaries to be a scientist. One of his letters (III, 40a), probably written in 1152, lists the customs duties paid to ‘Ayḏāb (Egypt) on his goods transported from Aden:

(1) In Your Name, O Merciful,
(2) Customs for “qāṭir”, three and a sixth and a half of a “qirāṭ”, half for it; galangal, one and a third and a quarter; (3) frankincense [lubān], five and half; aloe vera [ṣabr], seven and two thirds and a “qirāṭ”, half for it; its third we (4) still owe; five [pieces of] China, a dinar. Transport fees on our portion of the aloe vera and customs for it on the way // [five (?) And a half and a quarter] four and a sixth //; transport fees for the frankincense (5) and the “qāṭir” // [… and a fourth] a third and a fourth //; customs on the frankincense in ‘Aydhāb [‘Adhāb’], a third; the “qāṭir”, 23 “manns”…37

21Aden thus appears as a transit port for aromatics in the Middle Ages. These products came from India as well as from the regions of Dhofar and Ḥaḍramawt where frankincense is produced, or from the island of Socotra known for its aloe (“ṣabr”). This port had a dimension both regional as the goods of Yemen arrived at Aden, and international since the merchants coming from India resold their products there or were satisfied to pay customs duties at the “furḍa”.

22Ẓafār (modern site of al‑Balīd in Salalah, Sultanate of Oman) was an important port in the frankincense area of Dhofar (Sultanate of Oman), but Arab geographers such as Ibn Baṭṭūṭa in his Rihla (Travels, ca 1340) only mention horse trade to India.38 The references to frankincense trade are related to Mirbāṭ, a port located about 70 km east to Ẓafār. Al‑Idrīsī (1100‑1166), in his Kitāb Nuzhat al‑mushtāq fī ikhtirāq al‑āfāq (The pleasure excursion of one who is eager to traverse the regions of the world, also called The book of Roger) described Mirbāṭ as a city (“madīna”) and a trading hub, adding that frankincense trees grow in the hills behind.39 Following al‑Idrīsī, Yāqūṭ (1179‑1229) wrote in his Mu‘jam al‑Buldān (Dictionary of countries) that Mirbāṭ is the only port of the coast of the region of Ẓafār, has an independent “sulṭān” and that frankincense is produced on the hills of its hinterland.40 Thus, according to Arabic written sources, the main port involved in the frankincense trade in Dhofar was Mirbāṭ and not Ẓafār.

23Maritime relays were involved in the incense trade within the framework of secondary networks between the major ports such as Aden, al‑Shiḥr or Ẓafār. Ḥayrīj lies on the Mahra coast (Yemen), about 170 km east from al‑Shiḥr. Ibn Ḫurradāḏbih describes Ḥayrīj as an anchorage (“marsā’”), and al‑Hamdānī (Ṣifa, 10th century) indicates that the place served as a stopover between al‑As‘ā (al‑Shiḥr) and Dhofar.41 Finally the Kitāb ‘ajā’ib al‑Hind (The book of the wonders of India, 10th century) situates the distribution limit of the incense trees in the surroundings of Ḥayrīj.42 Ḫalfāt is mentioned by al‑Idrīsī and later by Ibn al‑Mujāwir who indicates its role as a relay between Dhofar and Aden.43 Although the material collected attests to exchanges with India, Africa and the Gulf, it is unclear whether Ḫalfāt played the same role in the Islamic period from the 11th century onwards as did the ancient “Syagros” described in the Periplus.44 Further east is the town of Ḥāsik, about 120 km from Salalah. It corresponds with the “‘Asíkhôn” of the Periplus.45 According to al‑Idrīsī, Ḥāsik is a small village (“qariya”).46 Ibn Baṭṭūṭa landed here on his way from Ẓufār to ‘Umān. He says that the frankincense trees are located in this area and that the resin is taped.47 However, he does not give information about its trade, but we can conclude that Ḥāsik was an anchorage visited by boats navigating along the coast.

Arabia in Chinese textual sources

24According to the Arab geographers such as al‑Ya’qūbī and al‑Bakrī, ports such as Aden and al‑Shiḥr had important trade relations with China. What do the Chinese written sources tell us about these ports and the trade of frankincense?

25The rise of the Tang dynasty in the early 7th century corresponds with a political and social revolution. The rulers gave lands to farmers, a new system of taxes was set up and many people migrated to central and southern China in order to improve their wealth.48 Moreover, Indian culture penetrated China: mathematics, medicine, astronomy, philology helped improve China’s quality of life. A taste for luxury and exotic products developed in the court before spreading to the wealthy and urban classes. Many merchants from Asia settled in this prosperous region. In Guangzhou, Muslims, Jews, Christians and Mazdeans settled in the foreigners’ district. In Chang’an (Xi’an), located in north-eastern China, a Nestorian church was built in 628 and a Mazdean temple was rebuilt in 631.49 The wealth and security of such a vast territory under Tang rule stimulated a large-scale trade that would assure a regular flow of goods. The ports of Guangzhou, Hangzhou, Ningbo, Quanzhou or Jiaozhou (Hanoï) welcomed ships from Srivijaya, Ceylon, India, Iran, and Byzantium (fig. 2).50

Fig. 2 – The frankincense trade according to Zhao Rugua’s Zhufanzhi (after Wheatley 1959, p. 16).

26During the Song period (960‑1279), the trade network established during the previous era grew thanks to maritime trade improvements in China. These can be shown through two major books dealing with trade and economic geography written during the late southern Song dynasty (1127‑1279). First, the Lingwai daida was compiled in 1178 by Zhou Qufei, then assistant to the deputy prefect of Guilin.51 Second, Zhao Rugua, superintendent in the merchant marine of the province of Fujian, wrote his Zhufanzhi in 1225.52 The first part describes the regions trading with China, from Japan to Morocco. The second part details 43 products imported into China through maritime trade: frankincense, myrrh, styrax and ambergris are among the 28 products exported from “Dashi” (Arab lands) to China.53

27Medieval China had an important need in perfumes and aromatic products for domestic use, in medicine, and for religious rituals.54 Taoist and Buddhist legends richly describes different perfumes and scents, whether locals or exotics.55 Among the elites, exotic and expansive products were preferred. Agarwood, in Chinese “chenxiang”, was the most important aromatic substance of vegetal origin.56

28In China, frankincense is named “ruxiang” and “xunluxiang”. The first one is a descriptive locution, “ruxiang” meaning aromatic nipple or milky incense referring to the shape and colour of the resin.57 It is attested since the 2nd century BC.58 The second one is attested from the 3rd century AD and is a deformation of the Sanskrit word “kundurika”.59 This etymology demonstrates the role of Indian merchants in the trade of frankincense in China.

29In the Tang textual sources, the link between Boswellia sacra’s resin and its geographical origin, that is to say southern Arabia, is not obvious. However, the Jiu Tangshu (History of the Tang) mentions that the city of al‑Shiḥr, called “Shuego”, was on the maritime trade route from Guangzhou at the end of the 8th century.60 By the beginning of the 13th century, the frankincense production area is better known to the Chinese. Zhao Rugua indicates that frankincense comes from the land of the Arabs and more precisely from the southern coast of Arabia:

[Ruxiang, milk incense], or [xunluxiang], come from the [Dashi] countries from [Maluoba, Mirbāṭ], [Shihe, al‑Shiḥr], and [Nufa, Ẓafār], from the depths of the remotest mountains valleys. The tree which yields this drug may, on the whole, be compared to the [song, pine]. Its trunk is notched with a hatchet, upon which the resin flows out, and when hardened, turns into incense, which is gathered and made into lumps. It is transported on elephants to the [Dashi, on the coast]; the [Dashi] load it upon their ships for barter against other goods in [Sanfoqi, Srivijaya]; and it is for this reason that the incense is commonly collected at [Sanfoqi].61

30This description fits with the distribution area of frankincense trees in the south of Arabia, from Hadramawt to Dhofar. It also highlights the importance of these three harbours or anchorages in the frankincense trade in the Indian Ocean, although no hierarchy is mentioned.

31Frankincense is not the only aromatic product traded by the Arabs to China. Again according to Zhao Rugua, myrrh comes from the country of “Malomo”, Mirbāṭ of the “Dashi” and aloes from “Nufa”, Ẓafār.62 The other aromatic substances originating from Arab lands are dragon’s blood (Dracaena sp.), sweet benzoin (Styrax benzoin), liquid storax (Liquidambar orientalis or Styrax officinalis), cloves and ambergris.63 Though it is not the main point of this paper, it is worth noting that according to Chinese written sources, cloves were presumably originating from Arab lands (“Dashi”) and Java whereas cloves come from Maluku Islands. It shows the role of Arab-Muslim merchants as intermediaries in the spice trade with China. The origin of ambergris remains unknown and Zhao Rugua explains: “In the Western Sea of the [Dashi] there are dragons in great numbers. Now, when a dragon [long] is lying on a rock asleep, his spittle [xian] floats on the water, collects and turns hard, and the fishermen gather it as a most valuable substance”.64

32Thus, according to these Chinese texts, the main ports in the south of Arabia exporting frankincense to China are al‑Shiḥr, Ẓafār and Mirbāṭ. It is interesting to compare with the written sources in Arabic, which do not mention Ẓafār. However, the Chinese information are indirect as there is no evidence that the Chinese ships visited the ports of Arabia before the 15th century.65 They had their information from traders or intermediaries who may have been imprecise regarding the origin of frankincense to keep their trade secret.

Archaeological evidence

33The archaeological remains from the different ports described above attest to the maritime contacts between the southern shores of the Arabian Peninsula and East Asia. The sites of al‑Shiḥr and Sharma were excavated recently, al-Balīd (Ẓafār) is still being investigated, and Ḥayrīj, Ḫalfāt, Mirbāṭ and Ḥāsik have been surveyed. Aden is the only port that has not been investigated archaeologically due to its ongoing occupation.

34Archaeological research along the southern coast of Arabia, in both Yemen and the Sultanate of Oman, dates back to the late 19th century. J. Theodore and M. Bent, two British explorers, visited Dhofar and the Hadramawt in order to gather information regarding the ancient trade of incense.66 In the late 1980s, D. Whitcomb carried out surveys around Aden and its coastline as far as Abyan and Wādī Ḥaḍramawt.67 In 1993, archaeological research on the Islamic period in Yemen developed mainly on the coast. Surveys were carried out by C. Hardy‑Guilbert and A. Rougeulle from the Tihama to Mahra.68 G. King and C. Tonghini surveyed the districts of Aden and Abyan.69 Surveys by C. Hardy‑Guilbert and A. Rougeulle led to the excavations of two important medieval port sites in Ḥaḍramawt: al‑Shiḥr and Sharma. The former, excavated by C. Hardy‑Guilbert, is a port whose occupation dates back to the pre-Islamic period. However, the excavations focused on the tell of al‑Qariya which has occupation levels ranging from 780 to the present day. Sharma is a site which played an important role in Indian Ocean trade from ca 980 to 1180. Studies of these two sites have provided a better understanding of the integration of Islamic Yemen into the vast trading network of the Indian Ocean. Unfortunately, the research at both had to be stopped after 2007 because of the political situation in Yemen. While the coast of Yemen cannot be investigated further at present, the coast of Dhofar, in the prolongation of the Mahra coast, was recently surveyed by L. Newton and J. Zarins. They investigated eleven sites with evidence of Islamic occupation.70

35The city of al‑Shiḥr is located on the south coast of Yemen, in the province of Ḥaḍramawt, almost halfway between Aden and the current Sultanate of Oman. The Boswellia sacra, the frankincense tree, grows in its hinterland (fig. 1). This strategic position favored the development of maritime trade, which was particularly flourishing from the Abbasid period. Al‑Shiḥr became one of the main links between the Far East, in particular China, India and the Arab and Mediterranean world. The excavated site, containing the first levels of occupation of the medieval city, in the al‑Qariya district, is located 60 m from the shore.71 It is bordered to the west by the Wādī Samūn and is in the form of a tell about 9 m high, occupying mainly an area of one hectare. Six campaigns were carried out between 1996 and 2007 under the direction of C. Hardy‑Guilbert. The city of al‑Shiḥr is mentioned in ancient sources under the toponym “al‑As‘ā”.72 Sometimes called “Lahsā”, the ancient city is said to be outside the present city, in an area called by the excavator “Shiḥr‑est”. If few structures are visible on the surface, on the other hand, diagnostic material was found in significant quantity, dating from the 1st millennium BC until the rise of Islam. The oldest pieces are comparable to the material found from the excavations of Qāni’ (Bi’r ‘Alī, in the Gulf of Aden), a port in activity between the 1st century BC and the 7th century AD.73 The site also delivered Iron Age materials comparable to that found in Oman. The most recent archaeological remains date from the end of the Sassanid period or the beginning of ancient Islam. The absence of classical Abbasid ceramics indicates that the site was abandoned no later than this period, but more likely before. The urban occupation has moved further to the west, located under the present city. According to the tradition, a mosque was built in al‑Shiḥr as early as 632, on the site of the current Friday mosque, in the al‑Qariya district. It is indeed from this district, located just in front of the sea, that the city developed. However, excavations did not attest to occupation until the 8th century. In addition, 15 stratigraphic levels can be distinguished, grouped into 10 chronological phases extending from the end of the 8th century onwards.74 After the initial phase, the city expanded to the north, east and west, before being fortified during the Rasūlid period (626‑858/1228‑1454). As mentioned earlier, its commercial activity is attested in written sources and this is also evidenced by the archaeology as there is evidence for items from other Islamic countries such as Iran or Iraq, India, the Far East (China, Burma) and Africa.75 On the other hand, the incense trade was an important source of income for this port and the discovery of as many as 61 incense burners in all archaeological levels of the site shows the importance of incense in daily life.

36Forty-five kilometers east of al‑Shiḥr, on the ocean coast of Yemen, Sharma is an entrepôt created ex nihilo around 980 and abandoned around 1180. It was excavated by A. Rougeulle between 2001 and 2005.76 It is a transit entrepôt belonging to the vast medieval trading network of the Indian Ocean, undoubtedly established by merchants from the Persian Gulf, at the crossroads of their maritime routes. The site is presented as a fortified complex, comprising 50 massive buildings interpreted as warehouses. The quantity of imported ceramics, particularly porcelain and stoneware from the Song period, is one of the largest and most varied ever found on an Islamic site. It attests to the exchanges maintained with China throughout the period of occupation. The corpus of imported ceramics also includes “sgraffiato” from Iran, Indian red polished ware, African and Yemeni wares, such as so-called “wavy line ceramics”, and production at the nearby site of Yaḍghaṭ. The study of the pottery shows that a large proportion of it was imported, mainly porcelain and stoneware from the Song period. This production made it possible to date the occupation of the site and proves the regularity of the commercial exchanges which were maintained between this entrepôt and the Far East. Resin was found in all the archaeological levels during the excavations, demonstrating that resinous incense was used during the entire period of occupation of Sharma.77 More than one hundred samples were analysed. The chemical analysis revealed that two thirds of them were copal, whereas only 18 % of the residues were frankincense.78 The copal identified comes from Hymenae verrucosa trees growing in East Africa and Madagascar (fig. 1). While we know of no text describing the entrepôt of Sharma as involved in the trade of incense, the substances that were found and identified on the site, along with the twelve pottery incense burners recovered during the excavations, show that frankincense was used there, and that the merchants of Sharma were involved in the import of copal from East Africa or Madagascar to the east.79 This copal may have been imported to the Persian Gulf, as a piece of copal has been found during the excavations led by D. Whitehouse in the medieval port of Sīrāf (Iran).80

37On the shore of the Mahra, between the Ḥaḍramawt and Dhofar, surveys by A. Rougeulle have highlighted the existence of coastal settlements with activity beginning in the pre-Islamic period.81 The sites of Ḥayrīj and Ḫalfāt appear in Arabic written sources and the material collected attests to long-distance exchanges. Ḥayrīj covers around 30 hectares and lies at the mouth of Wādī Ṭīkar. Several buildings are located there, as well as a small mosque. The site also includes an impressive cemetery of several thousand graves. The involvement of Ḥayrīj in the incense trade is strongly suggested by A. Rougeulle, who also indicates the relations of this establishment with the island of Socotra known for its aromatic and tinctorial products, in particular aloe (“ṣabr”) and dragon’s blood. As early as the 10th century, Ḥayrīj played a relay role between Dhofar and Yemen, and was active in international trade.82 The diagnostic material collected comprises fragments of Chinese ceramics from the Song period and Abbasid ceramics. The site then experienced a more limited commercial activity, but it was not abandoned until the 17th century. Ḫalfāt, which was occupied from the pre-Islamic period, is located a few kilometers north of Ra’s Fartak and has been identified with Cape Syagros described in the Periplus.83 The site extends for about 1,000 m at the mouth of the Wādī Ḫalfūt. In the center of the site is the town spread over 1.5 hectare. The port was located at the mouth of the wādī and was protected by two forts. Although the material collected attests to exchange with India, Africa and the Persian Gulf, it is unclear whether Ḫalfāt played the same role in the Islamic period from the 12th century onwards as the ancient Syagros.

38Still further east, about 500 km from Sharma, the site of al‑Balīd is the medieval port of Ẓafār which gave its name to the Dhofar region (ar. “Ẓufār”) in the present Sultanate of Oman. The site was discovered in the 1950s as part of the archaeological research carried out in Dhofar by an American Foundation for the Study of Man (AFSM) team which opened several soundings. The main trenches were undertaken on the highest part of the site which corresponds with a complex set of buildings traditionally called the “palace of the king”. On the floor of a warehouse was a mass of 108 pounds (ca 49 kg) of what has been interpreted as frankincense, along with the remains of the baskets in which it was kept.84 Excavations were then carried out from 1977 under the direction of P. Costa.85 Today, the site is being excavated by a mission from the University of Pisa.86 The site extends over 1,600 m and is up to 400 m wide. The eastern extension of the site is limited by a T-shaped lagoon, a remnant of an ancient “khawr”. The latest archaeological remains date to the 11th‑12th century. The enclosure was fortified, and three gates were excavated. Finally, the remains of the great mosque bear witness to the richness of the city. It measures 25.5 m by 22.5 m and contained 143 octagonal columns. The numerous mansions and warehouses, only a few of which have yet been excavated, convey the image of a dynamic trading city. Thus, though the written sources in Arabic do not mention the trade of frankincense as the main source of income of the port, archaeological evidence show that incense was traded here.

39The recent survey by L. Newton and J. Zarins along the shores of Dhofar led to identification of eleven sites dating from the Islamic period, according to diagnostic material.87 Among them, Mirbāṭ and Ḥāsik yielded pottery assemblage similar to that found at al‑Balīd, i.e. material from the Late Abbasid period (ca 11th century) and material imported from East Africa, South-East Asia and China.88

Conclusion

40The trade of frankincense involved different types of port structures. There were important ports like Aden or al‑Shiḥr, described by medieval authors as “mīnā’” or “furḍa”. They were directly controlled by a centralised administration, especially during the Rasūlid period. Taxes on frankincense sale were collected there, and ships coming from the Red Sea or the eastern part of the Indian Ocean met in these ports. Moreover, a secondary network of smaller maritime structures (Ḥayrīj, Ḫalfāt, Ḥāsik) also traded frankincense.

41The main ports were Aden, al‑Shiḥr and Ẓafār (fig. 3). However, the port of Ẓafār was more famous for the trade of horses to India. Consequently, the Arab geographers do not mention the frankincense trade from Ẓafār while the Chinese written sources seem to consider Ẓafār to be an important centre for frankincense trade along with Mirbāṭ, located in the same region.89 Moreover, archaeological evidence testifies there were frankincense warehouses in Ẓafār. Mirbāṭ is a port mentioned in both Arabic and Chinese texts. Following Yāqūṭ, this port had an independent position, at least in the 12th century. The presence of secondary ports (“marsā”) also assess the importance of cabotage along the coast up to the 14th century at least. These ports are Ḥayrīj, Ḫalfāt, and probably Ḥāsik, though the last one is of less importance (fig. 3). Other entrepôts like Sharma had a special status. Sharma played an important role in the Indian Ocean trade, but for a rather short period.

Fig. 3 – Map showing the hierarchy of the different ports involved in the frankincense trade in southern Arabia between the 8th and the 14th century (S. Le Maguer-Gillon).

42From the 4th century BC, rulers in South Arabia have established ports on the shores of the Indian Ocean.90 They were founded in strategic locations allowing control of production, the sale of incense and a role in long-distance trade of these spices. It is very likely that the so-called “caravan kingdoms” understood the danger of a trade based solely on two products. By acting as a staging post on the maritime spice route, they were securing some commercial success. This is why the port of Aden saw continuous activity from at least the 1st century AD to the present day. When ports like Qāni’ or Sumhuram were abandoned, new settlements followed, such as al‑Shiḥr about 150 km east of Qāni’/Bi’r ‘Alī and Ẓafār only about 50 km to the west of Sumhuram. Small coastal settlements like Mirbāṭ do not experience a break between the pre-Islamic and Islamic periods. All these ports are located in areas where frankincense grows, and this continuity testifies to the economic importance of these products. However, like their ancient South Arabian predecessors, the Muslim rulers were concerned that these ports were not only involved in the trade of frankincense produced in their hinterlands. The authorities ensured that these commercial establishments captured goods originating in the Far East and India to supply local markets and transport them to the Mediterranean world via the Red Sea. This transit was subject to taxation and ship owners obliged to buy local products.

43These different geographical and historical elements explain why the southern coast of Arabia has continued to play a major role in the aromatics and spice trade from the Indian Ocean to the Mediterranean Sea, despite periods of economic recession.

44This article is a preparatory work leading to a wider research, including archaeological investigations. My goal is to develop further these parallels between Chinese and Arabic texts. Eventually I aim to set up a typology of these structures thanks to both textual and archaeological data in order to give a new insight into the Indian Ocean trade in the medieval period. Finally, it would be interesting to conduct excavations at Mirbāṭ in order to date more precisely its medieval occupation, as well as to understand better its relation to medieval Ẓafār and its trade links with China.