In fact, as demonstrated in a two-part investigation published in The Business in May 2004 (see panel), experts have known since the early years of Aids that "HIV" test kits could not be used to diagnose Aids. Delegates at a World Health Organisation meeting in Geneva in 1986 heard that the kits were licensed to protect blood and plasma donations, not as a screen for Aids or people at risk of Aids. But, dictated by public health needs, usage had expanded and "it was simply not practical" to stop this, as Dr Thomas Zuck, of the US Food and Drug Administration, put it.

The 100 experts from 34 countries heard that, though the tests were useful in safeguarding blood supplies, something more was needed to distinguish genuine infection with HIV. Dr James Allen, of the US Centres for Disease Control Aids programme, said studies suggested some people were reacting to components of the cell line used to grow HIV for many of the test kits licensed in America. Other reactions occurred because of antibodies to normal cell proteins, naturally occurring in the body. Allen warned that the problems could be magnified in areas of the world that did not have the sophisticated facilities of America.

The meeting was told that a so-called "confirmatory test", called western blot, relied on the same principle as the test kits it was supposed to be checking and so was liable to the same kind of false-positive reactions. Subsequent research has repeatedly confirmed this problem: more than 60 conditions that cause such false-positives have been documented. One is tuberculosis, which produces symptoms of Aids as defined in Africa and is immensely widespread among impoverished people.

As the HIV/Aids paradigm won worldwide acceptance, increasingly complex procedures for trying to make a reliable diagnosis came into being. But the basic problem – not being able to validate any of these procedures against pure virus taken from patients – still remains.

Harper’s has published pages of letters in the latest (May) issue in response to Farber’s article, which appeared in March. Roughly half are supportive, half against. The first letter is from Culshaw, who writes: "This debate should have happened long ago, before an unproven hypothesis of an immune-destroying retrovirus was thrust upon a vulnerable public, and without being thoroughly critiqued in the scientific literature. Despite the promises made in 1984, there is still no cure and no vaccine. Instead, there has been a fundamental erosion in scientific and clinical-trial standards, with implications reaching far beyond HIV.

"To do the best we can for those affected by Aids – including those in Africa, where Aids presents a clinical picture quite different from that in the developed world – there urgently needs to be an honest scientific debate."

There is an association between testing HIV-positive and risk of developing Aids. This is the main reason why scientists believe HIV is the cause of Aids. But the link is artificial, a consequence of the way the test kits were made.

It never proved possible to validate the tests by culturing, purifying and analysing particles of the purported virus from patients who test positive, then demonstrating that these are not present in patients who test negative. This was despite heroic efforts to make the virus reveal itself in patients with Aids or at risk of Aids, in which their immune cells were stimulated for weeks in laboratory cultures using a variety of agents.