The wreck was first discovered in the Java Sea by an Indonesian fisherman, 90 nautical miles north of the town of Cirebon on the north coast of Java. The directors of the company Cosmix very quickly obtained an excavation licence and dived on the site, where they established that this was an untouched wreck, protected from looting both by its distance from the coast and by the depth of 57 metres in which it lay. The first objects recovered from around the tumulus suggested that this was the wreck of a merchant ship from the end of the 10th century.
The excavation took place from April 2004 to October 2005, with a break from November 2004 to February 2005. A private operation financed by the company Cosmix Archaeological Underwater Research and Recovery and directed by Luc Heymans, it was executed under the aegis of the Indonesian Ministry of Fisheries, in collaboration with the Indonesian Ministry of Culture. Archaeologists and professional divers of several nationalities, all highly qualified, trained Indonesian divers to assist in the operation. In all thirty-six divers were involved in the excavation, working in turns due to the sea depth which meant that each person could only work on the bottom for 45 minutes a day, each making two dives of 25 and 20 minutes respectively and taking 1.5 hours to ascend in stages.
Being too large and too high for the divers to be able to mark out a grid without risk to themselves, the wreck tumulus was surrounded by a chain labelled every metre. Another chain was set up perpendicularly to this to give the coordinates of the objects and the wreck itself. In all a total of 24,000 dives were required to complete the operation. Each dive was made by three men to a specific spot, each working at a rate of 1 m²/day (= 3 m³). A large ballasted metal basket containing smaller plastic baskets was lowered to the site on a windlass. One diver operated the vacuum nozzle to clear the sand and mud off the objects; the pieces were laid in the smaller baskets, and these were then placed in the large metal basket and brought back up to the surface. Thus the plan of the wreck was gradually drawn up.
When they arrived on board the excavation ship MV Siren, moored over the site, the objects were cleaned with sea water over a sieve. They were then numbered and labelled with plastic tags and packed in nets.
It was not possible to desalinate the objects on site in the excavation ship. The main problem faced by the team was that of preventing them from drying out too quickly. So they were wrapped in cotton cloth and placed in plastic baskets covered in bubble wrap, and then stored in containers which were sprayed regularly to keep them damp. Every two months the filled containers were despatched to the excavation warehouse in Jakarta.
There the items were put into desalination baths in 100% fresh water, changed every five days and metred regularly, with filters to remove the sediment. One of the Indonesian team members suggested putting small fish in the baths to prevent an invasion of mosquitoes, an idea as simple as it was effective! Once desalinated, the items could be drawn and photographed.
The whole cargo of around 500,000 objects was listed by the excavation team, but only 250,000 items were recovered. The "magma" of different ores identified in the bottom of the hull was also drawn and samples taken, but it was left in place due to its size and weight.