In 1988, an unusual scene unfolded at the French National Institute of Health and Medical Research Unit 200 in Clamart, France. The then editor of Nature, John Maddox, arrived at Jacques Benveniste’s laboratory accompanied by investigator Walter Stewart and the professional magician and skeptic James Randi. Their task was not to perform tricks but to examine an extraordinary scientific claim: that biological activity might persist even when no molecules of the original substance remained. Earlier that year, Jacques Benveniste and colleagues had reported experiments suggesting that water might retain a form of “memory” of anti-immunoglobulin E antibodies, capable of triggering basophil degranulation even after extreme serial dilution of the antibody solution. The implications were striking, not least because the findings appeared to echo ideas long associated with homeopathy.
Seen in retrospect, the episode is more than a curious scientific controversy. It shows how experimental claims, expectations within a field, and standards of evidence can interact when results challenge established knowledge. Science advances when curiosity pushes beyond what seems settled, and genuine discoveries sometimes arise from surprise. Difficulty arises, however, when extraordinary claims are presented as proven even though they conflict with well-established principles of chemistry, molecular biology, and sound experimental practice. I first learned of the Benveniste episode as a doctoral student in biomedical research, when the controversy was still fresh in scientific conversation. Nearly four decades later, I return to it as a working scientist, with a deeper appreciation for how rigor and restraint protect inquiry from illusion.
The arithmetic of extreme dilutions
In his now-infamous experiments, Benveniste used anti-immunoglobulin E antisera and subjected the preparation to serial dilutions, 1:10, 1:100, 1:1,000, and so on, eventually reaching concentrations at which no meaningful number of molecules of the original antibody could be expected to remain. Avogadro’s number, a concept most of us were introduced to as high school students, matters here. Beyond roughly 10-23 molar (assuming an initial antibody concentration near 5 x 10-6 molar to 1 x 10-5 molar, corresponding to about 16 or 17 serial 1:10 dilutions), the expected number of molecules in a typical sample falls below one. In practical terms, most aliquots contain no molecules at all, while a few may contain a single molecule purely by chance. At that point the system behaves less like chemistry and more like probability, where apparent signals require particularly careful scrutiny.
Yet Benveniste carried the dilutions far beyond this threshold, more than 50 serial steps, reaching concentrations on the order of 10-56 molar or lower, and still reported reproducible basophil degranulation. At such levels, chemistry predicts nothing more than noise and routine assay variability. Central to Benveniste’s interpretation was the claim that biological activity did not diminish monotonically with dilution. Instead, the responses appeared and disappeared in a reproducible, oscillatory manner across successive dilution steps. According to the published data, these effects recurred at defined intervals of dilution, producing patterns often described as sinusoidal rather than stochastic.
Crucially, these oscillatory patterns were reported to occur at dilution levels where, as the arithmetic of Avogadro’s number makes clear, no molecules of the original antibody would be expected to remain in solution. The proposed explanation therefore shifted away from molecular presence toward an unspecified non-molecular mechanism.
The investigation and its fallout
These claims nevertheless drew the attention of the journal Nature, and under its then editor, John Maddox, the paper appeared in mid-1988 following peer review, accompanied by an editorial note highlighting the extraordinary nature of the findings and the need for independent confirmation. Many scientists were alarmed, and heated discussions soon spread internationally. The publication, accompanied by a cautiously worded editorial, triggered widespread criticism. Letters to the journal reflected concerns about methodological rigor. The implications, if taken seriously, were difficult to reconcile with established physicochemical laws.
Carried to its logical conclusion, especially given the explosion of monoclonal antibody therapies in the decades since, the reasoning would suggest that vanishingly small quantities of monoclonal antibodies targeting molecules such as interleukin-5, interleukin-6, or tumor necrosis factor-alpha could, in principle, be diluted to yield biological “activity” on an unlimited scale. If correct, the implications would extend far beyond a single experiment, reaching into the foundations of pharmacology and molecular biology.
In response to the mounting controversy, Maddox took the unusual step of visiting Benveniste’s French National Institute of Health and Medical Research laboratory. He was accompanied by Walter Stewart and James Randi. Their purpose was straightforward: to examine experimental procedures, blinding, and controls in a system where subtle effects and strong expectations could introduce bias. What they found raised serious concerns. Laboratory records were disorganized, blinding procedures inconsistent, and controls inadequate. When proper blinding was introduced, the reported effects disappeared. The observed signals fell within ordinary assay variability, yet they had been presented as meaningful phenomena at dilution levels where any apparent effect would most likely arise from chance or uncontrolled noise.
Lessons for modern science
The episode became a cautionary lesson that remains relevant today. It reminded the scientific community that belief, however sincerely held, cannot override physical law, and that extraordinary claims require not only extraordinary evidence but also disciplined experimental design. In retrospect, the episode proved to be about more than one research group, one journal, or a single extraordinary claim. It showed how belief, when allowed to override method, can masquerade as discovery.
Today, the Benveniste episode still resonates because the pressures that shaped it remain familiar. Scientific ideas can sometimes spread faster than the experiments needed to confirm them, and enthusiasm can outrun verification. The episode remains a reminder that skepticism is not hostility but an essential part of scientific self-correction. The lesson endures. Science does not fail because it asks difficult questions; it fails when belief, however sincere, is mistaken for evidence. That boundary, the discipline between conviction and demonstration, is one science cannot afford to lose.
Suggested readings
- Davenas, E., et al. 1988. Human Basophil Degranulation Triggered by Very Dilute Antiserum against Immunoglobulin E. Nature 333: 816-818.
- Maddox, J. 1988. When to Believe the Unbelievable. Nature 334: 287.
- Maddox, J., J. Randi, and W. Stewart. 1988. “High-Dilution” Experiments a Delusion. Nature 334: 287-290.
- Chaplin, M. F. 2007. The Memory of Water: An Overview. Homeopathy 96: 143-150.
- Ioannidis, J. P. A. 2005. Why Most Published Research Findings Are False. Public Library of Science Medicine 2:e124.
Rao M. Uppu is a professor of environmental toxicology and chemistry.
