A straw Jew with a cardboard laptop waits on a Port-au-Prince roof

On the Thursday of Holy Week in 1993, in the back streets of Port-au-Prince, a straw figure sat in a chair on a one storey tin roof. It wore blue jeans, a shirt, a suit jacket, a tie and sneakers, with a pen in the shirt pocket. Across its crossed legs lay a laptop computer cut out of cardboard, and a cord ran from it down into a briefcase standing beside the chair. Its owner, a migrant from the south of the country whose hands showed a life of heavy labour, explained that the figure would stay up for the Rara band to pass and would be burned the following afternoon. Burning a Jwif or a Judas in effigy at Easter had been done in Haiti by every class, from peasant hamlets to the verandas of rich planters. Jean-Claude Duvalier banned the practice in the 1970s, around the time foreign factories and tourists began arriving. People kept burning them.

The court orders that stopped repatriation left the United States holding thousands of Haitians and no plan for them. Within three days of the first restraining order people were being taken to Camp Bulkeley, a remote corner of the naval station at Guantanamo Bay in Cuba, and housed in permanent shelters and tents. Bulkeley was small. Within a week navy construction battalions were putting up a far larger tent city on the abandoned airfield at McCalla for the nearly five thousand others coming off the cutters. Meanwhile the district court turned its restraining order into an injunction, the appeals court dissolved it, and the sequence repeated until the legal shape of the border had been redrawn seven times in a little over a month. Some order or other held the ships back for about six weeks. Guantanamo answered the government's problem. People could be kept there without entering the United States, on ground leased from Cuba, in a place lawyers for the government said American courts could not reach.

In 1992 UNESCO began a project on the routes of the slave trade, on the argument that hiding a history of that size blocks any reconciliation built on top of it. Ceremonies and monuments followed in Benin, Ghana and Ivory Coast. In Ouidah it produced a festival of Vodun arts, and dozens of artists from Africa and the diaspora were commissioned to make sculptures. Their work now lines the road running from the town down to the beach where captives were loaded into canoes. One stop is a memorial said to stand over the common grave of people who died before embarkation. Another is the Tree of Return, where guides say the departing were made to circle three times so their spirits would come home, and another the Tree of Forgetting, which was supposed to strip their memories. Robin Law finds no trace of either tree in any record before the 1990s.

Archaeologists working in the Caribbean have often judged the size of a pre-Columbian settlement from the spread of pottery and stone visible on the ground. Work published by Versteeg in 1993 showed that the measure is unreliable, because what reaches the surface is a small and uneven sample of what was buried and does not mark the edges of the place where people actually lived. Many island villages have therefore been recorded as smaller than they were. The finding carries a second warning. A finished artefact is the end of a process, and reading only the end product hides the labour, the movement of materials and the shifts of style that made it. Caribbean sites were reworked continually, and careful mapping and excavation are what recover that. I keep this on a Haitian timeline because the Taino settlements of Hispaniola were sized by the same rough method, and how many people met Columbus in 1492 has been argued over ever since.