Over the course of a year, the 246 private and NHS hospitals in Britain equipped to carry out CSF taps had declined to touch them, usually on the grounds that the test amounted to human experimentation, not treatment. In November 2002 one hospital briefly assented before putting the matter before its ethics committee, which decided four months later not to proceed for the same reason: the children were being used as guinea pigs.

It was an arguable point. Before an illness can be treated, it must be fully understood, and the root of these children's problems hadn't been ascertained. By the time a hospital outside Detroit agreed to accept them in March, their parents and advisers were worrying that the tests would never take place. They were nearly proved right.

On the night before the children arrived at the hospital, lawyers acting for GlaxoSmithKline (GSK), Merck and Aventis Pasteur MSD, manufacturers of the MMR triple vaccines that have been used in the UK since 1988, approached a High Court judge in London for an injunction to prevent the CSF taps going ahead. Two of these combination jabs had been called into question before: Pluserix, by Smith Kline (pre-Glaxo), and Aventis Pasteur's Immravax were withdrawn in 1992 after the "urabe" strain of mumps virus used in them was deemed responsible for a meningitis outbreak by the health authorities in Canada. That strain was replaced and M-M-R II, patented by Merck but licensed to GSK, became the triple jab most often offered in the UK. Now the possible misbehaviour of the measles component was at issue. The drug companies wanted a delay because their medical representative needed to be present at the procedure, but couldn't get to Port Huron, Michigan, in time. The injunction, however, was denied.

The children were the claimants in a "class action" — legal-speak for a case launched jointly by victims with the same grievance. If successful, it would validate the claims of 1,300 other British families and trigger international damages awards that could top $1 trillion. The proposed test, to look for traces of measles-vaccine virus in the children's CSF, could provide evidence that it can pass from the gut's lining into the brain, where measles is known to affect cerebral processes.

This is one of the most contentious issues in the row about what, if anything, brings on a disease described, but not universally accepted, as "autistic enterocolitis". In the UK, the condition was first identified by Dr Andrew Wakefield, but scientists in Japan, Norway, Ireland and the US (including Buie, Winter and Kushak, based at Harvard) have also published research supporting a link between intestinal disease and autism.

The theory that a malfunctioning or "leaky" gut sends partially digested food — in the form of opioid compounds known as peptides — up to the brain is one of the less controversial aspects of the hypothesis under investigation. Whether measles vaccine is what gives rise to the gut disease in the first place is the trillion-dollar question. So far, the sum of Wakefield et al's discoveries has not met the exacting medical standards that establish causation. All it points to is an "association". But the importance of the spinal-fluid link was well understood by the defendants in the class action. Merck's QC had recently referred to it in court as "a significant result when trying an issue as to whether or not MMR vaccine causes autism".

Time was running out for the claimants. Their action was being financed by the Legal Services Commission (LSC), a successor to the Legal Aid Board, which had set a July deadline for the submission of expert medical evidence, after which funding would be reviewed. Having lost a year trying to get the CSF samples in the UK, they now had to fly seven severely autistic, occasionally violent children — most of whom had never been in a plane before — halfway round the world.

Another bid by the defendants to secure an injunction, this time in the US, also failed. Then the hospital called the British party in Detroit to cancel their appointment.

Although lumbar taps on autistic children are common in the US, this batch, Lansing hospital now felt, constituted unwarranted human experimentation.

But the children's camp had an undisclosed back-up plan. They had made an arrangement with another hospital in Port Huron, two hours along the shore of Lake Michigan, and this time, despite further delaying tactics from the lawyers in London, the CSF taps went ahead. One of the seven children reacted badly to the anaesthetic and couldn't be tested; the other six were fine.

Now the party and the fluid samples had to be flown home for analysis. There was bedlam on the bus as the anaesthetic wore off: one child tried to exit the moving vehicle by the back door, while another was restrained by his mother in the toilet. At the airport, the container of dry ice carrying the CSF was deemed too large to be carried on as hand luggage, and another business-class seat had to be specially purchased for it.

After the KLM flight had boarded, five US customs officers arrived to take the lawyers and doctors off the plane — the only passengers they apprehended — for separate, 30-minute taped interviews. They weren't asked any questions pertaining to passenger safety and their large container: the issue was why the children hadn't been tested back in the UK. In transit at Schiphol airport in Amsterdam, they were again singled out for more questioning.