Sewn ships to Canton (c. 700) — the monsoon route and its massacres

Before: an ocean of relays

In the middle of the seventh century the sea between the Persian Gulf and south China was already busy, but no ship sailed the whole of it. Goods moved in relays. Red Sea and Gulf merchants sailed to the ports of western India and Sri Lanka; Indian and Sri Lankan merchants sailed on to the Malay Peninsula and Sumatra; ships of the archipelago, which Chinese writers called Kunlun ships, carried cargoes the last stretch to Guangzhou. Each leg had its own sailors, its own lingua franca and its own season. A bale of Chinese silk reaching the Gulf, or a sack of Persian frankincense reaching China, had changed hands several times on the way, and its price had risen at every hand. George Hourani's Arab Seafaring in the Indian Ocean (1951), still the founding study, describes the early centuries of the ocean as exactly this kind of chain: real, long-distance and lucrative, but segmented ¹. Philippe Beaujard's two-volume Les mondes de l'océan Indien (2012) places the same picture inside a longer world-system history, in which the first-millennium ocean was already a single economic space, yet one stitched together by intermediaries rather than by through-voyages ¹⁵.

The intermediaries were not passive. They were the reason the system worked. A relay trade spreads risk: no single crew has to survive five months at sea, and no single ruler has to be trusted from one end to the other. It also spreads profit, and that profit built the harbour polities of the first millennium, from the Tamil ports to Funan on the Mekong delta. This record is about what happened when that chain was replaced, for two centuries, by one ship.

The hinge of the relay system in late antiquity was Sri Lanka. In the sixth century the Alexandrian merchant-turned-monk Cosmas Indicopleustes described the island as a central market of the whole ocean: "The island being, as it is, in a central position, is much frequented by ships from all parts of India and from Persia and Ethiopia, and it likewise sends out many of its own" ²⁴. Silk and other goods from the far east, he wrote, came to Sri Lanka from "Tzinista", China, and passed from there to the western ports ²⁴. In Cosmas's ocean the Persians already had a resident church on the island and a strong position in the trade, but the ships from China and those from Persia met in the middle. Nobody in his account sails from one end to the other. That is the baseline against which the change of the late seventh century has to be measured: an ocean whose trade was already continuous, but whose voyages were not.

Winds that everyone knew

The physical engine of the ocean is the monsoon. From roughly November to March the north-east monsoon blows from the Asian landmass towards the sea; from roughly May to September the south-west monsoon blows back, wet and hard. A sailor who knew the calendar could cross the Arabian Sea in weeks rather than creeping along the coast for months, and could plan a return before he left.

This was not new knowledge in the seventh century. The Periplus of the Erythraean Sea, a Greek merchant's handbook of the mid-first century CE, already describes ships leaving Egypt's Red Sea ports "about the month of July" to catch the south-west wind for India, and credits a pilot named Hippalus with first working out the direct crossing ². The south-west wind itself came to be called by his name, and Lionel Casson's commentary on the text treats the crossing as routine by the author's day ². Persian and Indian sailors almost certainly knew the same winds earlier still. What the monsoon did not do, by itself, was join the western ocean to the eastern one. The wind systems of the Bay of Bengal and the South China Sea run on different timings. A ship that wanted to go from the Gulf to China had to string together several monsoon windows, wait out the gaps in a harbour, and still arrive before the wind reversed. That was a problem of organisation, not of geography.

Relay trade and the Kunlun ships

The eastern half of the ocean, in the centuries before the through-voyage, belonged substantially to Southeast Asian shipping. Pierre-Yves Manguin's study of the region's trading ships, published in 1993, assembled the Chinese descriptions of these vessels and set them beside the archaeological finds ³. Chinese writers of the third century already describe large ships of the southern seas carrying several hundred people, with multiple masts and sails of woven matting; Manguin identifies them with the "lashed-lug" hull tradition of the archipelago, in which planks were edge-joined with wooden dowels and lashed to internal frames through carved lugs ³. These were the Kunlun ships of Chinese texts. Wang Gungwu's classic study of the early "Nanhai trade" (1958) shows how much of China's southern commerce before the Tang ran through such intermediaries, with Chinese merchants themselves rarely going to sea ¹⁹.

The western ocean, meanwhile, was the domain of the sewn ship: a hull whose planks were not nailed but stitched together edge to edge with cord of coconut fibre or hemp, then framed and caulked. Sewn boats are attested on the Arabian and Indian coasts from antiquity; the Periplus itself mentions them ², and Dionisius Agius's survey of the ships of the Islamic world traces the technique through the medieval sources ²⁰. Sewn hulls flexed in a seaway and could be repaired anywhere with local fibre. They also leaked, and had to be re-stitched every year. The Franciscan John of Montecorvino, who sailed from the Gulf to India around 1291, described the problem exactly: "The ships in those parts are mighty frail and uncouth with no iron in them and no caulking. And so if the twine breaks anywhere there is a breach indeed. Once every year, therefore, there is a mending of this, more or less, if they propose to go to sea" ¹⁰. A ship built that way could still cross an ocean. It needed harbours along the route where it could be unstitched and stitched again.

What the ocean had, and what it lacked

Before the late seventh century, the ocean had most of the ingredients of a single trading system and lacked a few that turned out to matter:

Within about a century and a half every item in the right-hand column existed. Several items in the left-hand column were damaged in the process.

The transmission: one ship, one season-pair, five months

The change can be dated by a passenger. In the eleventh month of 671 the Chinese Buddhist monk Yijing, bound for India to collect texts, boarded a ship at Guangzhou that he described as belonging to a Bosi — a Persian — owner ⁵. Twenty-odd days later he landed at Bhoga, the Srivijayan capital on the Musi river near modern Palembang in southern Sumatra. He was not the first Chinese pilgrim to go by sea, but he is the first to tell us that the ship which carried him from China was a western one.

Persian merchant communities were by then already resident in the Chinese south. The Sasanian empire had fallen to the Arab conquests between 636 and 651, but its Gulf ports, its shipowners and its sailors did not stop working; they passed under a new government and, over the next two centuries, largely into a new religion. Hourani's central argument is that the Islamic conquests did not create Gulf seafaring but enlarged its market: the new caliphate unified the Mediterranean, Iraq, Iran and Arabia into a single zone of demand, and after 762 that demand had a capital at Baghdad, on the Tigris, upriver from the Gulf ¹. Gulf ships went further east because there was now more money at home to pay for what they brought back.

Srivijaya: a harbour state built for the wind

The Persian ship that carried Yijing stopped at Bhoga because that was where the wind obliged it to stop. Between the end of the south-west monsoon in the Bay of Bengal and the north-east monsoon that would carry it on to China, any ship coming through the Malacca Strait had to wait somewhere. A ruler who controlled the waiting place controlled the trade.

The earliest dated Old Malay inscription, found at Kedukan Bukit near Palembang and published by George Coedès in 1930, records that in 682 a ruler with the title Dapunta Hyang set out on a "sacred journey" (siddhayātrā) with an army it numbers at twenty thousand, and founded a settlement ⁶. Further inscriptions of 683–686 threaten curses on anyone who rebels against "Śrīvijaya" ⁶. O. W. Wolters's Early Indonesian Commerce (1967) argues that Srivijaya's power was precisely the power to make the strait the obligatory halt for the China trade, using its fleet both to protect ships that called and to punish ports that competed ⁴.

Gunawan Kartapranata. Kedukan Bukit inscription (682), National Museum of Indonesia, Jakarta, 2017. CC BY-SA 4.0 via Wikimedia Commons. · CC BY-SA 4.0

Srivijaya was also a centre of learning. Yijing spent six months there studying Sanskrit grammar on his way out and returned to write there for years. In the fortified city of Bhoga, he wrote in a passage translated by Junjirō Takakusu in 1896, "Buddhist priests number more than 1,000, whose minds are bent on learning and good practices" ⁵. He advised Chinese monks bound for India to stop at Srivijaya first and study for a year or two. The through-voyage did not bypass the archipelago. It made one archipelagic harbour much richer than the rest.

The route in 851

The best description of the voyage at its height is a short Arabic book completed in 851, the Akhbār al-Ṣīn wa-l-Hind (Accounts of China and India), compiled from the reports of merchants including one named Sulaymān, and supplemented around 916 by Abū Zayd al-Sīrāfī, a man of Siraf ⁷⁸. Gabriel Ferrand's French translation of 1922 and Tim Mackintosh-Smith's English edition of 2014 are the standard versions ⁷⁸. The book begins with the trans-shipment at Siraf: goods were brought "from Baṣra, from ʿOmān and from other ports" and loaded there onto the China ships, because the upper Gulf was too shallow and rough for them ⁷. Ships watered at Masqaṭ and then struck out across the open sea.

Ferrand reconstructed the stages and their durations from the text ⁷:

That is roughly five months from end to end, against monsoon windows that had to be met at each stage. At Kūlam the China ships paid a transit duty of a thousand dirhams each ⁷. The text records exactly the logistics a single crew needed: where to water, where the reefs were, where the duties fell.

The same text records what the merchants found at the other end. It contains what is probably the earliest description in Arabic of Chinese porcelain: "There is in China a clay of superior quality from which they make bowls as thin as glass bottles; one sees the liquid they contain through them" ⁷. It notes the fires that regularly destroyed the warehouses of Khānfū, built of wood and split bamboo, and the price the merchants paid for being so far from home ⁷. The book is not a romance of the sea. It is a merchants' manual, written by people who had made the voyage and wanted the next crew to come back alive.

The Belitung ship

For half a century the claim that Gulf ships sailed all the way to China rested on texts. In 1998 Indonesian sea-cucumber divers found a mound of ceramics on the seabed four nautical miles north of Tanjung Pandan on Belitung Island, between Sumatra and Borneo, in seventeen metres of water ¹⁰. The site was excavated in 1998 and 1999 by a commercial salvage company under an Indonesian licence, with the marine archaeologist Michael Flecker directing the second season ¹⁰.

The hull answered the old question. "Every timber was fastened with stitching, with no sign of wooden dowels or iron fastenings," Flecker wrote; the keel was 15.3 metres long and the whole ship perhaps 18 metres overall ¹⁰. In 2007 timber samples were identified as Afzelia africana, an African hardwood, for the stempost, frames and planks, with teak from India for the through-beams and probably African juniper for the ceiling ¹⁰. Flecker's first publication had cautiously called it "Arab or Indian" ⁹; the timber analysis excluded an Indian yard, and he concluded that it "likely was constructed in the Near East, perhaps in the region of Oman, Yemen, or Iran" ¹⁰. Its cargo and its location, he wrote, provided "irrefutable archaeological evidence that there was direct trade between the western Indian Ocean and China during the latter part of the first millennium" ¹⁰.

The cargo dated the voyage. One bowl carries an inscription giving a day in the seventh month of the second year of the Baoli era — 826 — so the ship sailed in or after that year ²². Its contents, now in Singapore's Asian Civilisations Museum, show what the eastern leg of the route was for:

• Ceramics: some 60,000 pieces recovered, perhaps 70,000 originally, about 25 tonnes ¹⁰; the National Heritage Board counts about 57,500 from the Changsha kilns of Hunan, around 55,000 of them bowls ²².
• Lead: about ten tonnes of lead ingots stowed as paying ballast ¹⁰.
• Spices: storage jars still full of star anise ¹⁰.
• Precious metal: gilt silver boxes and gold vessels, possibly private trade or a diplomatic gift ¹⁰.
• Southeast Asian objects: scale weights and a cut gold coin from the archipelago, which show the ship had called at an Indonesian port, probably Palembang ¹⁰.

Jacklee. Changsha bowls from the Belitung shipwreck (c. 826–850), Shipwrecked: Tang Treasures and Monsoon Winds, ArtScience Museum, Singapore, 2011. Collection of the Asian Civilisations Museum, Singapore. CC BY-SA 3.0 via Wikimedia Commons. · CC BY-SA 3.0

Hourani's verdict, quoted by Flecker at the end of his report, held: "The sea route from the Persian Gulf to Canton was the longest in regular use by mankind before European expansion in the 16th century" ¹¹⁰.

Who sailed

The 851 text names the human types of the route as precisely as its harbours. The nākhudā (from the Persian for "ship-master") owned or commanded the ship; the rubbān piloted it; merchants travelled as passengers with their own goods ⁷. Crews were mixed. Flecker suggests that the Belitung ship probably took on Southeast Asian sailors and pilots to replace men lost to disease or accident on the way out, as European ships would later do on the same route ¹⁰. Its Chinese personal objects — an inkstone, a lacquer dish, cymbals — suggest that a Chinese merchant may have been on board ¹⁰.

At the Chinese end, the community was large and organised. The Accounts report that at Khānfū "the sovereign of China has conferred on a Muslim the administration of justice among his coreligionists," who led Friday prayers and read the sermon in the name of the caliph, and whose judgements the merchants of Iraq always accepted ⁷. Chinese sources call the foreign quarter of Guangzhou the fanfang. Kuwabara Jitsuzō's study of the Song-era Muslim superintendent Pu Shougeng (1923), the foundational Japanese work on Arab and Persian merchants in China, traces the quarter and its headman back to the Tang ¹⁴.

What changed and what was replaced

The through-voyage did not invent Indian Ocean trade. It changed its scale, its goods and its institutions, and it displaced several older ways of doing business. K. N. Chaudhuri's Trade and Civilisation in the Indian Ocean (1985), which runs from the rise of Islam to 1750, treats the period after the seventh century as the point at which the ocean's emporia became a single interdependent system ¹⁶. The changes can be grouped under five heads.

Bowls by the tens of thousands

The most visible change is in what was carried. Before the through-voyage, the eastward trade from China ran on silk and other light luxuries: high value, low volume. The Belitung cargo is dominated instead by stoneware bowls, made in Changsha in Hunan, hundreds of kilometres inland, and shipped down the river systems to the coast. Some 55,000 bowls in one hull is not luxury trade but mass manufacture for export ²². Many carry decoration that seems designed for western buyers, and the scale implies kilns producing for an ocean they never saw.

The ceramics arrived. Flecker lists Tang ceramics excavated at Siraf, at Sohar in Oman, inland at Samarra, the Abbasid capital, at Nishapur in Khurasan, at Fustat in Egypt, and as far south as Zanzibar and the Comoros ¹⁰. At Siraf, Changsha stoneware and white porcellaneous wares appear in levels below the great Friday mosque, which was built around 820 ¹⁰¹¹. David Whitehouse's excavations at Siraf, published in final form in 2009, showed a port city of stone houses, warehouses and mosques strung along a barren shore with almost no agricultural hinterland: a city that existed for the sea trade and for nothing else ¹¹.

What the bowls displaced is the more interesting story. In ninth-century Iraq, potters at Basra answered the import with an imitation: a white tin-opacified glaze that copied the look of Chinese white wares on local earthenware. Robert Mason's study of Islamic lustreware traces how this imitation became the base for the new decorative traditions of Abbasid ceramics ²¹. A Chinese export reorganised a Middle Eastern craft from the outside.

A harbour state at the strait

The through-voyage concentrated power at the waiting points. Srivijaya's rise from the 680s is the clearest case. Wolters argues that the older pattern of many competing harbour principalities in the western archipelago gave way to a single dominant harbour that could compel ships to call ⁴. The cost of that concentration fell on the rivals. The Srivijayan inscriptions of 683–686 are full of threats against those who rebel, and Wolters reads the polity's maritime power as both protection and coercion ⁴⁶.

Srivijaya also reorganised language. The inscriptions are in Old Malay, written in a South Indian–derived script, at a time when the court's sacred language was Sanskrit. The harbour that the wind made important made Malay the language of the strait, and Malay later became the trade language of the archipelago ⁶¹⁵.

Merchant law without an empire

The through-voyage created something new at its far end: a resident community of foreign merchants inside a Chinese city, living under its own law. The Accounts describe a system in which the Chinese authorities impounded the cargoes of arriving ships in warehouses, guaranteed them for six months until the last ship of the monsoon had arrived, then took thirty per cent in kind as an import duty and returned the rest ⁷. What the court wanted it bought at the highest price, and paid at once ⁷.

Several distinct institutions grew out of that arrangement:

• The foreign quarter (fanfang). A defined residential district for foreign merchants at Guangzhou, later formalised under the Song ¹⁴.
• The community judge. A Muslim appointed by the Chinese ruler to judge disputes among Muslims according to Islamic law, recognised by the Iraqi merchants as legitimate ⁷.
• The trading-ship commissioner. A Tang official, the shibo shi, charged with registering and taxing foreign ships; Wang Gungwu traces the growth of this office in the south ¹⁹.
• The warehouse and the season. The six-month holding period, keyed to the monsoon, which turned Chinese customs practice into a function of the ocean's calendar ⁷.

This was a model of merchant self-government that the Indian Ocean's Muslim diasporas would reproduce from Calicut to Quanzhou over the next seven centuries, and that Kuwabara traced into the Song and Yuan ¹⁴. It replaced a looser older practice, in which foreign traders came and went through local intermediaries without a standing legal identity of their own.

Segmentation after 878

The through-voyage itself did not last. Abu Zayd, writing around 916, says plainly that "new events" in China had "interrupted all maritime relations with China, ruined that country, and made its laws disappear" ⁷. After the rebel Huang Chao sacked Guangzhou in 878–879 (see the cost section below), western ships stopped going to China. Abu Zayd describes Kalah, on the Malay Peninsula, as the market where aloes, camphor, sandalwood, tin and spices were now gathered, and the port to which "the ships of ʿOmān" now went and from which they sailed home ⁷⁸. The western ships had stopped halfway.

The pattern of the ocean thus returned partly to relays, but a different kind: a segmented network with two great hinges, the Malacca Strait and the Indian west coast, and with Chinese shipping, which under the Song (960–1279) began to sail the eastern legs itself. Siraf, whose shipowners had built the through-route, was badly damaged by an earthquake in 977 and lost its primacy to the island port of Kish ¹¹. The western ships' two centuries were over. The institutions they had planted, the foreign quarter, the community judge and the monsoon-keyed customs regime, outlasted them.

The ship type outlasted them too. Sewn hulls continued to be built on the Omani, Yemeni and Indian coasts into the twentieth century. Between 2008 and 2010 shipwrights at Qantab in Oman built a full-size reconstruction of the Belitung ship, the Jewel of Muscat, from the excavation drawings, stitching its planks with coir rope; in 2010 it sailed from Oman to Singapore, where it was put on permanent display. Tom Vosmer and his colleagues, who documented the project, describe a hull that worked at sea as the evidence predicted, and that needed constant attention to its seams ²⁵.

What was replaced

Taken together, the through-voyage displaced or demoted:

1. The relay intermediaries of the first millennium, especially the Sri Lankan and smaller archipelagic harbours, whose share of the long-distance trade fell as ships ran past them to Srivijaya and Kalah ⁴¹⁵.
2. The archipelago's own long-distance shipping on the China leg, as western ships carried cargoes Southeast Asian ships had once carried ³.
3. Chinese luxury-only exports, as mass-produced stoneware joined silk in the holds ¹⁰²².
4. Local ceramic traditions in Iraq, remade around imitations of the imported white wares ²¹.
5. Many small sovereignties at the strait, replaced by one dominant harbour state that lived by compelling ships to call ⁴.

What the cost was

The through-voyage was conducted by people who had no navy. That fact explains most of its bill. A merchant community resident in someone else's city is wealthy, visible and defenceless; it is a natural target in a revolt and a natural scapegoat in a crisis. Along the route between 758 and 1502 the record documents at least four episodes in which foreign merchant communities were attacked in their thousands, and a continuous trade in human beings that ran in the same holds as the bowls.

Guangzhou, 758, and Yangzhou, 760

The first recorded violence at the Chinese end was committed by the merchants themselves. The Jiu Tangshu, the older official history of the Tang, records under the first year of the Qianyuan era (758) that Guangzhou reported that soldiers of the Dashi (Arabs) and the Bosi (Persians) were attacking the city, and that the prefect, Wei Lijian, abandoned it and fled ¹². The later Xin Tangshu adds that they burned the warehouses and houses and left by sea ¹². The circumstances are obscure. The empire was then fighting for its life against the An Lushan rebellion, and the attack shows how heavily armed, and how numerous, the foreign community in Guangzhou had become.

Two years later the violence ran the other way. In 760 the Tang general Tian Shengong, sent to suppress a rebellion in the Huai region, entered Yangzhou, the great inland port at the junction of the Grand Canal and the Yangzi. His biography in the Jiu Tangshu records that his troops "plundered the property of the people and merchants" and that "of the foreign merchants, Persians, those killed numbered several thousand" ¹². They were killed for their goods, by the army of the state that had admitted them.

Khanfu, 878–879

The episode that ended the through-voyage was the sack of Guangzhou by the rebel Huang Chao. Abu Zayd's account, written within about forty years, is the only detailed one, and his figure is the one every later writer repeats. In Ferrand's translation: the city was taken after a long siege "in the year 264 of the Hijra" (878 CE), and "its inhabitants were put to the sword. People who knew of these events report that 120,000 Muslims, Jews, Christians and Mazdeans who were established in the city and traded there were massacred, not counting the Chinese who were killed" ⁷. Abu Zayd explains how the number could be known: the Chinese authorities taxed the foreigners by head count ⁷. He adds that Huang Chao cut down the mulberry trees on which the silkworms fed, which "put an end to the export of silk, particularly to Arab lands" ⁷.

The figure should not be accepted at face value. No Chinese source records a massacre of foreigners on this scale, though the Chinese histories do record Huang Chao's capture of Guangzhou in 879. A resident foreign population of 120,000 would be very large for a ninth-century port, and modern historians generally treat the number as an exaggeration that nonetheless reflects real mass killing ⁸¹. What is not in doubt is the outcome: Abu Zayd records that after the sack, Chinese officials seized the property of the Arab ship-masters, and that "the disaster reached even the pilots and the brokers of Siraf and Oman" ⁷.

People as cargo

The same ships carried people for sale. The trade in enslaved East Africans to Iraq, which produced the great Zanj revolt of 869–883 in the salt marshes near Basra, ran along the western legs of this same network, and is treated in the atlas record on the Swahili coast. At the eastern end, Chinese sources of the Tang and Song speak of Kunlun slaves: dark-skinned people, at first from Southeast Asia and later also from Africa, held in wealthy households as servants, porters and doorkeepers. Don Wyatt's The Blacks of Premodern China (2010) traces how the term shifted from Southeast Asians to Africans as the maritime trade brought more Africans to Guangzhou ¹³. A Song writer, Zhu Yu, recorded around 1119 that the rich of Guangzhou kept "devil slaves" (guinu), and Wyatt shows that the number of people enslaved in this way, though impossible to count, was large enough to produce a literary stereotype ¹³.

No source quantifies the human cargo of the through-voyage. Its existence is not in doubt.

Ships, sailors and the sea

The 851 text is frank about the ordinary losses of the route. It explains the scarcity of Chinese goods in Iraq by the frequent fires at Khānfū, where houses were built of wood and split bamboo; by shipwrecks "on the outward and the return voyage"; by the pillage ships suffered on the way; and by the long forced stays in intermediate ports ⁷. The Belitung ship itself is one such loss, struck on the reef the Indonesians call Batu Hitam, a hundred and fifty metres from the wreck site ¹⁰. Its crew is not recorded.

The Belitung cargo also carried a modern cost. It was raised by a commercial salvage company and sold as a whole to a Singaporean buyer for a reported US$32 million. When the Smithsonian Institution planned to exhibit it, archaeologists objected that showing a cargo excavated for profit, at speed and in two seasons, would legitimise the commercial salvage of underwater heritage; after a review conference in December 2011 the Smithsonian cancelled the Washington exhibition ²³. The site's full archaeological record was lost to the speed of the work.

Srivijaya, 1025

The harbour states that profited from the route also paid for it. In 1025 the Chola emperor Rajendra I of south India sent a fleet across the Bay of Bengal against Srivijaya. His inscription at Tanjavur lists the harbours sacked — Kadaram (Kedah), Palembang and some dozen other ports of the Malay Peninsula, Sumatra and the Nicobar Islands — and records the capture of the Srivijayan king, Sangrama Vijayottungavarman, and of the jewelled "gate of the Vidyādharas" from his capital ¹⁷. The contributors to Nagapattinam to Suvarnadwipa (2009), edited by Hermann Kulke and others, debate whether the expedition was a war over the China trade or a raid for plunder and prestige ¹⁷. Either way, it was the network's wealth, concentrated at the strait, that made the strait worth sacking.

The Mīrī, 1502

The network remained, for eight centuries after Yijing's voyage, an unarmed one. Ships carried soldiers for defence against pirates, but no power claimed to own the sea. That ended with the Portuguese. On his second voyage to India, in the first days of October 1502, Vasco da Gama's fleet intercepted the Mīrī, a ship returning from the Red Sea to Calicut with Muslim pilgrims, off the Malabar coast. After plundering it, the Portuguese set it on fire with the passengers on board; those who swam were killed in the water. Accounts put the dead at around three hundred, including women and children ¹⁸. The clerk Tomé Lopes, who witnessed it, wrote an account that survives only in an Italian translation published by Ramusio in 1550 ¹⁸. Sanjay Subrahmanyam's biography of da Gama treats the episode as a statement of the Portuguese policy that would follow: passes, convoys, and a monopoly on the sea ¹⁸.

The bill

Those who paid can be named by category, if rarely by name:

• Resident foreign merchants in Chinese ports, killed in their thousands at Yangzhou in 760 and in far greater numbers at Guangzhou in 878–879 ⁷¹².
• The Chinese inhabitants of the same cities, whom Abu Zayd mentions only as the uncounted dead "not counting" the foreigners ⁷.
• Enslaved Africans and Southeast Asians carried in the holds and sold into households and plantations at both ends of the route ¹³.
• Crews and passengers lost at sea, named by the 851 text as a routine cause of loss but never counted ⁷¹⁰.
• Rival harbours and their rulers, coerced by Srivijaya and in turn sacked by the Chola fleet ⁴¹⁷.
• Pilgrims and merchants of the open sea of 1502, the first victims of a system of armed passes that ended the network's openness ¹⁸.

The through-voyage was not itself a war. It was a commercial system that ran for two centuries on wind, timber and trust, and it joined the economies of the Abbasid world and Tang China more tightly than anything before it. Its costs were the costs of wealth carried by the defenceless: periodic massacre of merchant communities at the Chinese end, the sale of human beings in its holds, the coercion of rival harbours by the states that profited from it, and, finally, its seizure by a power that came to the ocean armed.