Case Study 1 — Secretin, 1902, and the Invention of the Hormone

The experiment that proved chemical messaging exists

Type: Real, public, historical · Tier 1 facts · Relevance: §3.1, §3.7

Background: a world where only nerves carried messages

In 1900, physiology had one explanation for how one part of the body told another part to do something: nerves.

This was not a naive position. It was extremely well supported. Nerves had been traced, stimulated, cut, and mapped. Cutting a nerve reliably abolished a function; stimulating it reliably produced one. The entire discipline of experimental physiology had been built on this, and it worked.

So when William Bayliss and Ernest Starling, working in London, turned to a question about digestion, they were working inside a nervous-system framework.

The question: when acidic material from the stomach enters the small intestine, the pancreas responds by secreting alkaline fluid to neutralize it. How does the pancreas know?

The obvious answer was a nerve reflex. Acid stimulates sensory nerves in the intestinal lining; the signal travels to the nervous system; a motor signal travels back to the pancreas. This had in fact been the accepted account, associated with Pavlov's laboratory, which had done extensive and excellent work on nervous control of digestion.

The experiment

Bayliss and Starling did what the framework demanded: they cut the nerves.

In an anesthetized dog, they severed the nerve supply to a loop of small intestine, leaving its blood supply intact. Then they introduced acid into that denervated loop.

The pancreas secreted anyway.

This should not have happened. With the nerves cut, there was no path for a signal — under the prevailing model, the pancreas could not have learned that acid had arrived.

Then they did the experiment that mattered. They scraped the lining from a piece of intestine, ground it up with acid, filtered it, and injected the filtrate into a vein.

The pancreas secreted.

No nerve. No intestine involved at all. A chemical extract, injected into the bloodstream, produced the response. Something in the intestinal lining was released into the blood by contact with acid, and that something travelled to the pancreas and told it what to do.

They named the substance secretin. And a few years later, Starling proposed a general term for such substances, from a Greek root meaning "to arouse" or "to set in motion":

hormone.

🔬 Read the Study — the denervated loop experiment

text FIGURE 3.CS1 — "The pancreas that could not have known" [real published work] THE STUDY Physiological experiment in anesthetized dogs. Bayliss and Starling, University College London, 1902. Intervention: denervation of an intestinal loop, then introduction of acid; and separately, intravenous injection of an acid extract of intestinal mucosa. THE QUESTION Is pancreatic secretion in response to intestinal acid controlled by a nerve reflex? WHAT IT SHOWS No. Secretion occurred in the denervated loop, and an injected mucosal extract reproduced it entirely. A chemical messenger released into the blood is sufficient. Chemical signaling between organs exists. WHAT IT DOESN'T It does not identify secretin's structure (that took decades), quantify it, establish that nerves play NO role in digestion (they do), or show that other organs use the same strategy — though it made that question askable, which was the point. THE VERDICT Foundational. This is the experiment that created endocrinology as a field, and it is the ancestor of every drug in Part II of this book. THE LESSON The decisive experiment was a NEGATIVE control done properly. They removed the mechanism everyone believed in and the effect persisted. A result that survives the removal of its supposed cause is worth far more than a result consistent with it.

Why this is the right case study for Chapter 3

It established the category this entire book is about. Before 1902, "chemical messenger travelling in blood" was not a thing that existed in physiology. Every hormone in this book — insulin, GLP-1, growth hormone, oxytocin — is a member of a class that this experiment created.

It found the gut first. The very first hormone was a gut hormone, released in response to nutrients arriving, acting on a distant organ. That is precisely the architecture of §3.7 and precisely the architecture of GLP-1. The incretin story is a direct descendant of this experiment, and it is not a coincidence: the gut is a large sensory surface with a real need to communicate quickly with organs it is not connected to.

And it is an unusually clean example of a paradigm being overturned by an experiment rather than by an argument. Bayliss and Starling did not out-reason the nervous-control model. They removed the nerves and watched what happened.

The part that is usually left out

There is a coda worth knowing, because it says something about how science actually proceeds.

Pavlov's laboratory had, some years earlier, come very close to this result. Reports suggest that a researcher there had observed pancreatic secretion in circumstances difficult to explain by nerve reflex — and that the observation was not pursued, in a laboratory strongly committed to nervous control and enormously productive within that framework.

Take this as characterization rather than a documented verdict on anyone's judgment; the historical record on precisely who saw what and when is not fully settled. But the general pattern it illustrates is very well documented across the history of science: an anomalous result inside a successful framework is much more likely to be attributed to experimental error than to a new mechanism.

This is not stupidity. It is usually correct. Most anomalous results are experimental error, and a laboratory that chased every one would never finish anything. The judgment call — is this noise or is this a door? — is genuinely hard and is made under uncertainty by people who cannot know the answer in advance.

It is worth holding onto when Part III examines compounds where the evidence is anomalous and incomplete. The correct response to an unexplained result is neither to dismiss it nor to build on it. It is to run the experiment that would distinguish the two. That is what Bayliss and Starling did, and it is what has not been done for several compounds in this book.

Discussion questions

1. The decisive step was cutting the nerves — removing the mechanism everyone believed in and seeing whether the effect persisted. Why is that more informative than an experiment showing the effect is consistent with a chemical messenger? Relate your answer to §2.9's asymmetry between evidence for and evidence against.

2. Bayliss and Starling injected a crude extract, not a purified molecule. Secretin's structure was not determined for decades. What could they legitimately conclude from a crude extract, and what could they not?

3. The first hormone discovered was a gut hormone released in response to food arriving. Given §3.7, is that a coincidence? What is it about the gut that makes chemical signaling especially necessary there?

4. Nerves do participate in digestive control — the nervous-system model was not simply wrong, it was incomplete. How should a field respond when a new mechanism is found that does not eliminate the old one? Can you think of a current example in peptide science?

5. Consider the coda. Suppose you are running a productive laboratory and a junior researcher brings you an anomalous result that does not fit your framework. What would you actually do? What should you do? Are those different, and why?

6. Starling coined "hormone" from a root meaning "to set in motion." Chapter 2 argued that a peptide does not act but rather informs. Is Starling's word well chosen? Would you pick a different one?