Here’s my second piece for La Recherche’s special issue in August on scientific controversies – this one on the ‘memory of water’.
So far, “The Memory of Water” has been used as the title of a play, two movies, a collection of poems and a rock song. When the French immunologist Jacques Benveniste proposed in 1988 that water has a memory, he gave birth to a catchphrase with considerable cultural currency.
But Benveniste, who died in 2004, also ignited a scientific controversy that is still simmering a quarter of a century later. While most physicists and chemists consider Benveniste’s original idea – that water can retain a memory of substances it has dissolved, so that they can display chemical effects even when diluted to vanishing point – to be inconsistent with all we know about the properties of liquid water, Benveniste’s former colleagues and a handful of converts still believe there was something in it.
The claim would be provocative under any circumstances. But the dispute is all the fiercer because Benveniste’s ‘memory of water’ seems to offer an explanation for how homeopathy can work. This ‘alternative’ medical treatment, in which putative remedies are so diluted that active ingredients remain, has a huge following worldwide, and is particularly popular in France. But most medical practitioners consider it to be sheer superstition sustained by ignorance and the placebo effect.
Yet while there seems no good reason to believe that water has a ‘memory’, no one is quite sure how to account for the peculiar results Benveniste reported in 1988. This episode illustrates how hard it is for science to deal with deeply unorthodox findings, especially when they bear on wider cultural issues. In such cases an objective assessment of the data might not be sufficient, and perhaps not even possible, and the business of doing science is revealed for the human endeavour that it is, with all its ambiguities, flaws and pitfalls.
Rise and fall
Benveniste did not set out to ‘discover’ anything about water. As the head of Unit 200 of the French national medical research organization INSERM in Clamart on the edge of Paris, he was respected for his work on allergic responses. In 1987 he and his team spotted something strange while investigating the response of a type of human white blood cell, called basophils, to antibodies. Basophils patrol the bloodstream for foreign particles, and are triggered into releasing histamine – a response called degranulation – when they encounter allergy-inducing substances called allergens. Degranulation begins when allergens attach to antibodies called immunoglobulin E (IgE) anchored to the basophil surface. Benveniste’s team were using a ‘fake allergen’ to initiate this process: another antibody called anti-IgE, produced in non-human animals.
The researchers sometimes found that degranulation happened even when the concentration of anti-IgE was too low to be expected to have any effect. Benveniste and colleagues diluted a solution of anti-IgE gradually and monitored the amount of basophil degranulation. Basic chemistry suggests that the activity of anti-IgE should fall smoothly to zero as its concentration falls. But instead, the activity seemed to rise and fall almost rhythmically as the solution got more dilute. Even stranger, it went on behaving that way when the solution was so dilute that not a single anti-IgE molecule should remain.
That made no sense. How can molecules have an effect if they’re not there? Benveniste considered this finding striking enough to submit to Nature.
The editor of Nature at that time was John Maddox, who often displayed empathy for outsiders and a healthy scepticism of smug scientific consensus. Rather against the wishes of his staff, he insisted on sending the paper for peer review. The referees were puzzled but could find no obvious flaw in Benveniste’s experiments. After they had been replicated in independent laboratories in Canada, Italy and Israel, there seemed to be no option but to publish Benveniste’s paper, which Nature did in June 1988 [E. Davenas et al., Nature 333, 816 (1988)] – accompanied by an editorial from Maddox admitting that “There is no objective explanation of these observations.”
Hope for homeopathy?
The Nature paper caused pandemonium. It was clear at once that Benveniste’s results seemed to be offering scientific validation of homeopathy, the system of medicine introduced in the early nineteenth century by the German physician Samuel Hahnemann, in which the ‘active’ ingredients, already diluted to extinction, are said to get even more potent as they get more dilute.
Advocates swear that some clinical trials support the efficacy of homeopathy, but most medical experts consider there to be no solid evidence that it is effective beyond what would be expected from placebo effects. Even many homeopaths admit that there is no obvious scientific way to account for the effects they claim.
Not, at least, until the memory of water. “Homeopathy finds scientific support”, proclaimed Newsweek after Benveniste’s paper was published.
But how could water do this? The French team evidently had no idea. They suggested that “water could act as a ‘template’ for the [anti-IgE] molecule” – but this made no sense. For one thing, they evidently meant it the other way round: the antibody was acting as a template to imprint some kind of molecular structure on water, which could then act as a surrogate when the antibody was diluted away. But why should a negative imprint of the molecule act like the molecule itself? In any case, the properties of antibodies don’t just depend on their shape, but on the positions of particular chemical groups within the folded-up protein chain. And most of all, water is a liquid: its H2O molecules are constantly on the move in a molecular dance, sticking to one another by weak chemical bonds for typically just a trillionth of a second before separating to form new configurations. Any imprint would be washed away in an instant. If Benveniste and colleagues were right, shouldn’t water show the same behaviour as everything it has ever dissolved, making it sweet, salty, biologically active, toxic?
But data are data. Or are they? That’s what Maddox had begun to wonder. To get to the bottom of the affair, he launched an unprecedented investigation into INSERM Unit 200. Maddox travelled to Clamart to watch Benveniste’s team repeat their measurements before his eyes, accompanied by American biologist Walter Stewart, a ‘fraud-buster’ at the National Institutes of Health who had previously investigated allegations of misconduct in the laboratory of Nobel laureate David Baltimore, and stage magician James Randi, a debunker of pseudoscientific claims like those of the ‘psychic’ Uri Geller. “So now at last confirmation of what I have always suspected”, one correspondent wrote to Nature. “Papers for publication in Nature are referred by the Editor, a magician and his rabbit.”
The Nature team insisting that the researchers carry out a suite of double-blind experiments designed to rule out self-deception or trickery. Their conclusions were damning: “The anti-IgE at conventional dilutions caused degranulation, but at ‘high dilution’ there was no effect”, the investigators wrote [J. Maddox et al., Nature 334, 287 (1988)]. Some runs did seem to show high-dilution activity, but it was neither repeatable nor periodic as dilution increased.
Attempts by other labs to reproduce the results also failed to supported Benveniste’s claims. Although occasionally they did see strange high-dilution effects, it is not at all uncommon to find anomalous results in experiments on biological systems, which are notoriously messy and sensitive to impurities or small changes in conditions. The ‘high-dilution’ claims meet all the criteria for what the American chemist Irving Langmuir called ‘pathological science’ in 1925. For Langmuir, this was the science of “things that aren’t so”: phenomena that are illusory. Langmuir adduced several distinguishing features: the effects always operate at the margins of detectability, for example, and their supporters generally meet criticisms with ad hoc excuses dreamed up on the spur of the moment. His criteria apply equally to some other modern scientific controversies, notably the claim by Russian scientists in the late 1960s to have discovered a new, waxy form of water called polywater, and the claims of ‘cold nuclear fusion’ achieved using benchtop chemistry by Martin Fleischmann and Stanley Pons in Utah in 1989 [coming up next!].
After Maddox’s investigation, most scientists dismissed the memory of water as a chimera. But Benveniste never recanted. He was sacked from INSERM after ignoring instructions not to pursue the high-dilution work, but he continued it with private funds, having attracted something of a cult following. These studies led him to conclude that water acts as a “vehicle for [biological] information”, carrying the signal that somehow encodes the biomolecule’s activity. Benveniste eventually decided that water can be “programmed” to behave like any biological agent – proteins, bacteria, viruses – by electromagnetic signals that can be recorded and sent down telephone wires. In 1997 he set up a private company, DigiBio, to promote this field of “digital biology”, and it is rumoured that the US Department of Defense funded research on this putative ‘remote transmission’ process.
Such studies continue after his death, and have recently acquired a high-profile supporter: the immunologist Luc Montagnier, who was awarded the 2008 Nobel prize for the co-discovery of the AIDS virus HIV. Montagnier believes that the DNA molecule itself can act as both a transmitter and a receiver of ultralow frequency electromagnetic signals that can broadcast biological effects. He believes that the signals emitted by pathogen DNA could be used to detect infection. He maintains that these emissions do not depend on the amount of DNA in suspensions of pathogens, and are sometimes detectable at very high dilution. They might originate, he says, from quantum effects in the water surrounding the DNA and other biological structures, according to a controversial theory that has also been invoked to explain Benveniste’s experiments [E. Del Guidice et al. Phys. Rev. Lett. 61, 1085 (1988)].
“Benveniste was rejected by everybody, because he was too far ahead”, Montagnier has said [Science 330, 1732 (2010)]. “I think he was mostly right but the problem was that his results weren't 100% reproducible.” In 2010 Montagnier began research on high-dilution DNA at a new research institute at Jiaotong University in Shanghai. “It's not pseudoscience, it's not quackery”, he insists. “These are real phenomena which deserve further study.” He is currently the head of the World Foundation for AIDS Research and Prevention in Paris, but his unorthodox views on water’s ‘memory’ have prompted some leading researchers to question his suitability to head AIDS projects.
Meanwhile, the idea that the undoubtedly unusual molecular structure of water – a source of continued controversy in its own right [see e.g. here and here] – might contrive to produce high-dilution effects still finds a few supporters among physical chemists. Homeopaths have never relinquished the hope that the idea might grant them the scientific vindication they crave: a special issue of the journal Homeopathy in 2007 was devoted to scientific papers alleging to explore water’s ‘memory’, although none provided either clear evidence for its existence or a plausible explanation for its mechanism [see here].
Such efforts remain firmly at the fringes of science. But what must we make of Benveniste’s claims? While inevitably the suspicion of fraud clouds such events, my own view – I joined Nature just after the ‘memory of water’ paper was published, and spoke to Benveniste shortly before his death – is that he fully believed what he said. A charming and charismatic man, he was convinced that he had been condemned by the ‘scientific priesthood’ for heresy. The irony is that he never recognized how his nemesis Maddox shared his maverick inclinations.
The “Galileo” rhetoric that Benveniste deployed is common from those who feel they have been ‘outlawed’ for their controversial scientific claims. But Benveniste never seemed to know how to make his results convincing, other than to pile up more of them. Faced with a puzzling phenomenon, the scientist’s instinct should be to break it down, to seek it in simpler systems that are more easily understood and controlled, and to pinpoint where the anomalies arise. In contrast, Benveniste studied ever more complicated biological systems – bacteria, plants, guinea pigs – until neither he nor anyone else could really tell what was going on. The last talk I saw his team deliver, in 2004, was a riot of graphs and numbers presented in rapid succession, as though any wild idea could be kept in the air so long as no one can pause to examine it.
This, perhaps, is the lesson of the memory of water: when you have a truly weird and remarkable result in science, your first duty is to try to show not why it must be true, but why it cannot be.