In 1936 you observed that runner bean plants continued their leaf movements in constant darkness, with a period close to — but not exactly — 24 hours. You published this as evidence of an endogenous biological clock. Your claim was precise: the rhythm was not driven by external cues. It came from inside.
The response, for the next thirty years, was largely skeptical. The objection was reasonable in principle: how could you be sure? The list of possible environmental cues that a plant might track is long — geomagnetic fluctuations, cosmic radiation, subtle temperature cycles, changes in atmospheric electricity. Your experiments had controlled for the obvious ones, but the skeptics could always propose another. Frank Brown at Northwestern spent much of the 1950s and 1960s arguing that organisms were sensitive to geophysical cues that no laboratory could fully eliminate. Proving a negative — proving that no external signal was responsible — is structurally harder than proving a positive. The burden fell on you.
You spent decades accumulating evidence. Plants grown in spaces shielded from magnetic fields still oscillated. Organisms flown across time zones free-ran at their own period rather than immediately tracking the local light cycle. Organisms in constant conditions kept approximate 24-hour rhythms for weeks, then months. The approximate nature of the period — roughly 24 hours but not exactly — was itself evidence: if an external geophysical cue were driving the rhythm, you would expect it to be exactly 24 hours, locked to the day. The fact that it drifted slightly was the fingerprint of something internal. An endogenous clock that Earth's light cycle could entrain, but that ran on its own when nothing entrained it.
The argument was settled, in most fields, by the mid-1960s. Colin Pittendrigh at Princeton had accumulated enough formal properties of the system — the phase-response curve, the limits of entrainment, the behavior in constant conditions — to treat the clock as a proper object of study. By 1971, Konopka and Benzer had found the first clock gene in Drosophila, calling it period, and the debate moved from whether the clock existed to what molecular machinery ran it.
What the molecular work found, over the next thirty years, was a transcription-translation feedback loop. In mammals: CLOCK and BMAL1 proteins activate genes for PER and CRY. As PER and CRY accumulate, they inhibit CLOCK and BMAL1. The inhibition causes them to degrade. As they degrade, CLOCK and BMAL1 are released and the genes switch on again. The cycle takes roughly 24 hours because protein production, accumulation, and degradation are slow. Hall, Rosbash, and Young worked out the details and received the Nobel Prize in 2017. By that point the feedback loop was the standard answer to the question of what a circadian clock is.
In 2005, Masahiro Nakajima and colleagues dissolved three proteins into a test tube — KaiA, KaiB, and KaiC, from a cyanobacterium — added ATP, and measured what happened. The phosphorylation state of KaiC rose and fell with a period of approximately 24 hours, for days, with no cells, no genes, no transcription, no translation. The feedback loop that Hall, Rosbash, and Young had described was not operating. There was no nucleus, no ribosomes, no DNA being read. Three proteins in a solution, keeping time because of the slow chemistry of their interactions.
I find this result harder to interpret than I expected. It does not overturn the mammalian feedback loop — that mechanism is real and operates in mammals. What it establishes is that the feedback loop is not the only way to build a circadian oscillator. Cyanobacteria and mammals arrived at the same period through entirely different molecular routes. No shared proteins, no conserved mechanism, no homologous clock genes. Two independent solutions to the same problem.
The problem, as the cyanobacterial result makes clear, is not to build a feedback loop. The problem is to produce a chemical oscillation with a period of approximately 24 hours. Everything else — the specific proteins, the mechanism, the number of components — is the solution, not the problem. The problem is set by Earth.
This is where I keep returning to your original claim. You said the clock was endogenous. Your skeptics said it was probably driven by some external signal you hadn't been able to eliminate. Both positions now seem to be pointing at something that is both. The KaiC oscillator is as endogenous as anything biological could be: three molecules in a test tube, ticking away from their own chemistry. But the period it settled on is 24 hours, not because of any intrinsic constraint on protein kinetics, but because organisms whose internal period matched Earth's rotation outcompeted organisms whose didn't. The test-tube clock is endogenous in mechanism and environmental in origin. It carries the signature of the planet's rotation period in its own slow chemistry.
What this means for the original debate is something like: you were right that the mechanism was internal, and Brown was right that the period was set by something external. The argument was about which one mattered more, but the answer is that neither is more fundamental — they're at different levels. The mechanism is internal; the selection pressure that shaped the mechanism over billions of years is external. An organism with an internal clock set to 23 hours loses, in a 24-hour world. An organism with a clock set to 25 hours also loses. The winners are the ones whose internal chemistry happened to resonate with the rotation of the planet.
What you were tracking in your runner beans in 1936 was a record of that selection. Every oscillation of those leaves — slow open, slow close, in the laboratory dark, with no sun, no zeitgeber, nothing to synchronize to — was the expression of a period that had been shaped by billions of years of exposure to the actual day. The endogenous mechanism is the planet's influence made internal. It just took a test tube to see it clearly.