Curare and Carbon Monoxide: Claude Bernard’s Poison Experiments

Two poisons made Claude Bernard’s name as an experimenter. The first was curare, the South American arrow poison that kills by paralysis. The second was carbon monoxide, the colourless gas of charcoal fires and faulty stoves. In the 1840s and 1850s the French physiologist used each of them as a tool: give the poison, then check, one by one, which parts of the living body still worked and which had failed. Curare led him to the narrow gap where a motor nerve meets a muscle. Carbon monoxide led him to the red pigment of the blood and to the discovery that the gas takes the place of oxygen and will not let go.

This page tells both stories mainly in Bernard’s own words, taken from his 1865 book An Introduction to the Study of Experimental Medicine, where he used them as worked examples of how an experiment is reasoned out. It also covers what later physiologists changed in his explanation of curare, the 1856 note he read to the Paris Academy of Sciences, and the 1857 lectures in which he gathered his poison work. The dates differ between sources — sometimes between Bernard’s own accounts — and where they do, this page says so. His life, the plants curare comes from, and what became of his findings in twentieth-century anaesthesia each have their own pages in this wing.

Table of Contents

  1. Poisons as Instruments of Physiology
  2. A Gift from Pelouze: Curare Arrives (1845)
  3. The Frog and the Physiological Autopsy
  4. Muscle Alive, Nerve Silent
  5. Locating the Block Between Nerve and Muscle
  6. Vulpian and the Motor End-Plate Debate
  7. Scarlet Blood: Carbon Monoxide (1846)
  8. A Hypothesis That Failed
  9. Oxygen Displaced Volume for Volume (1856)
  10. The 1857 Lectures and the Disputed Dates
  11. Key Research Papers
  12. Connections
  13. Featured Videos

1. Poisons as Instruments of Physiology

By the time curare reached him, Bernard had spent several years in the laboratory of François Magendie at the Collège de France, where, according to Jensen and Puissant’s 2026 history of his glycogen work, he became Magendie’s assistant in 1841. Magendie’s school worked on living animals and asked direct questions of them: cut this nerve, block that duct, and see what changes. Bernard carried the method a step further. A poison, he realised, could do something a scalpel could not. It could reach every tissue through the blood and knock out one property while leaving the others untouched. If the experimenter then looked carefully, the pattern of what had stopped and what still worked would show where the poison struck — and so would reveal something about how the healthy body is put together.

That is why, decades later, Bernard said he had studied curare “not for the sake of the substance itself.” In 1864–65 he gave his entire course at the Collège de France to curare, and he described the aim in the Introduction: to show how “the lesion of a terminal motor nerve” leads, step by step, to secondary effects that end in death. The poison was the probe; the nervous system was the subject.

The same thinking runs back to an older idea in the history of pharmacology — that what separates a poison from a medicine is the dose and the target, an idea associated with Paracelsus. The title of Bernard’s 1857 lectures reflects the same outlook, treating “toxic and medicinal substances” together as one subject. Both curare and carbon monoxide later became examples in the Introduction, Part II, chapter 1, where Bernard numbered his case histories: curare is his “fourth example” and carbon monoxide his “fifth.” He used them to show how an experiment can begin with an observation and no theory at all (curare), or with a theory that the experiment then overturns (carbon monoxide).

Back to Table of Contents

2. A Gift from Pelouze: Curare Arrives (1845)

Bernard’s own account opens plainly: “In 1845, Monsieur Pelouze gave me a toxic substance, called curare, which had been brought to him from America. We then knew nothing about the physiological action of this substance.” Théophile-Jules Pelouze was a Paris chemist, and the sample had reached him from South America, where peoples of the Amazon basin and the Guianas used curare on hunting arrows and blowgun darts.

Europe was not wholly ignorant of the poison. Bernard credited what was known to “the interesting accounts of Alex. von Humboldt and of Roulin and Boussingault.” From those travellers’ reports he knew two things: that its preparation was complicated, and that it “very speedily kills an animal if introduced under the skin.” Earlier English experimenters had also shown that small curarised animals could be kept alive if their lungs were inflated with bellows — a clue that the poison stopped breathing rather than poisoning the whole body. The story of those explorers, the plants and the chemistry is told on the companion page Curare: The Arrow Poison, Its Plants and Its Science.

What nobody knew was where in the body curare acted. Did it stop the heart? Spoil the blood? Paralyse the muscles themselves? Act on the brain or spinal cord? Bernard stressed that he had no hypothesis to test. He would simply give the poison and look. In the Introduction he presents this as an experiment made “to see” — the kind that starts with an observation rather than an idea.

Which year?

The year 1845 is Bernard’s own, written about twenty years after the event. Other accounts differ. James Black’s 1999 BMJ article, translating Bernard’s popular essay “Du curare” (Revue des Deux Mondes, 1 September 1864), dates a frog experiment to June 1844. Raghavendra’s 2002 history of neuromuscular blocking drugs gives 1846. The safest summary is that Bernard began working with curare in the mid-1840s, with 1845 the date he gave in his best-known account.

Back to Table of Contents

3. The Frog and the Physiological Autopsy

The first experiment was as simple as an experiment can be. In Bernard’s words: “First, I put curare under the skin of a frog: it died after a few minutes; I opened it at once, and in this physiological autopsy I studied in succession what had become of the known physiological properties of its various tissues.”

The phrase “physiological autopsy” is the key to the method. An ordinary autopsy looks at the body after death for visible damage — a clot, a tumour, an inflamed organ. Curare leaves nothing to see. Bernard’s autopsy instead tested functions while the tissues were still fresh. Does the heart still beat? Does the blood still look and behave like blood? If a muscle is touched with an electric current, does it twitch? If its nerve is stimulated, does the muscle respond? Each tissue had a known property — the heart’s rhythm, the muscle’s contractility, the nerve’s excitability — and the question was which of them the poison had destroyed.

The frog was the standard animal of nineteenth-century nerve and muscle physiology. Its tissues keep working for a long time after removal, and its leg nerve and calf muscle are easy to reach and to stimulate with the simple electrical apparatus of the day. That made it ideal for a fast, systematic check of one property after another.

A note on the experiments themselves

Bernard’s work was done on living animals, in an era before anaesthesia was routine in the physiology laboratory. His own household opposed vivisection, as the Life and Career page describes. Those facts are part of the historical record of these experiments and are reported here as such.

Back to Table of Contents

4. Muscle Alive, Nerve Silent

The result surprised him. Almost everything still worked. In the Introduction he lists it: “the heart maintained its movements, the blood was apparently no more changed in physiological properties than the muscles, which kept their normal contractility” — but “when the motor nerves were stimulated directly, they no longer caused any contraction in the muscles.”

Put simply: the heart went on beating, the blood looked normal, and the muscle twitched perfectly well when stimulated itself. Only one link had broken. The nerve that ordinarily commands the muscle could be stimulated as much as Bernard liked, and the muscle took no notice.

He did not stop at frogs. He repeated the experiment in mammals and in birds, and the result was the same. This mattered: a peculiarity of frog tissue would have told him little, but a pattern shared by frogs, birds and mammals pointed to something basic about how motor nerves and muscles work in every backboned animal.

His conclusion, in his own words

Bernard drew a sweeping conclusion: “curare causes death by destroying all the motor nerves, without affecting the sensory nerves.” Two parts of that sentence proved lasting. Curare acts on the motor side — the nerves that carry commands outward to muscles — and it leaves the sensory side, the nerves that carry feeling inward, alone. The word “destroying” was the part later physiologists would revise (see section 6).

The explanation of death followed from the pattern. The muscles of breathing are ordinary skeletal muscles driven by motor nerves. Cut off their commands and breathing stops, even though the heart is unharmed. That fitted the earlier observation that curarised animals could be kept alive with artificial inflation of the lungs: the poison killed by suffocation, not by attacking the heart or blood.

His conclusion that sensation was spared implied that a curarised animal is paralysed, not unconscious. The same point became a serious question when curare entered surgery in the 1940s, as described on the Legacy and Later Research page.

Back to Table of Contents

5. Locating the Block Between Nerve and Muscle

Knowing that the nerve failed while the muscle worked did not yet say where the failure lay. A motor nerve runs a long way: from the spinal cord, along the limb, to fine branches that end on individual muscle fibres. Was the whole nerve poisoned, or only part of it? Bernard answered with experiments on limbs.

As Raghavendra summarises them, “curare injected into a limb prevented the muscle contraction in response to nerve stimulation; the muscle continued to respond when stimulated directly.” By controlling which parts of the animal the poison could reach, Bernard could compare poisoned and unpoisoned stretches of the same pathway. The pattern pointed away from the spinal cord and the trunk of the nerve, and towards the far end of the pathway — the place where the nerve hands its signal to the muscle.

The verdict of later pharmacology is clear on this point. W. C. Bowman, in a 2006 review in the British Journal of Pharmacology, writes that the curares’ “site of action in producing neuromuscular block was determined by Claude Bernard in the mid-19th century.” Bernard did not use the modern terms. He spoke of the “terminal motor nerve,” the nerve’s last branches. But the place he identified is the one now called the neuromuscular junction, which the site’s neuromuscular junction animation shows in action.

The 1856 note to the Academy

Bernard set out the analysis formally in a short paper in the Comptes rendus of the Paris Academy of Sciences in 1856: “Analyse physiologique des propriétés des systèmes musculaire et nerveux au moyen du curare” (“Physiological analysis of the properties of the muscular and nervous systems by means of curare”), volume 43, pages 825–829. Its title states the method in a phrase: curare as a means of analysis. The paper is remembered for its demonstration that muscle has an excitability of its own, independent of its nerve. With the nerve silenced by curare, the muscle still contracted when stimulated directly, so contractility had to belong to the muscle itself.

What he could not yet know

Bernard located the block, but he could not explain its chemistry. The idea that a nerve passes its message to a muscle by releasing a chemical substance belongs to the twentieth century. In 1936 Henry Dale, Wilhelm Feldberg and Marthe Vogt published evidence for the release of acetylcholine at voluntary motor nerve endings, and curare was later understood to block acetylcholine’s action at the muscle. That story belongs to Otto Loewi and Henry Dale and to the companion page on curare’s science. A 1976 paper by W. D. Paton in the Journal of Physiology looked back at one of Bernard’s curare experiments and its place in the origins of the Physiological Society, a sign of how long the experiment stayed in physiologists’ minds.

Back to Table of Contents

6. Vulpian and the Motor End-Plate Debate

Bernard’s word was “destroying”: in his view curare destroyed the motor nerves, or at least their endings. Other French physiologists were not convinced. The point where a motor nerve fibre meets a muscle fibre carries a specialised structure, the motor end-plate. Alfred Vulpian, a younger physiologist in Paris, proposed that this end-plate, rather than the nerve itself, was where curare acted.

A 2002 letter in Anesthesiology by M. T. Cousin makes the historical point in its title: it was “Vulpian and not Claude Bernard” who first proposed the motor end-plate as curare’s site of action. The historian Jean-Gaël Barbara, writing in 2009, describes about thirty years of argument over curare at the Société de Biologie in Paris. In his account, Vulpian’s theory was that curare blocked transmission between the end-plate and the muscle, and it was younger physiologists of the next generation who eventually convinced Bernard. Black’s 1999 BMJ piece likewise notes Vulpian’s later end-plate idea alongside Bernard’s original account.

The difference matters. If curare destroyed the nerve endings, the paralysis would be a kind of damage. If it blocked transmission at the end-plate, the nerve could be perfectly healthy and the effect could, in principle, wear off. Twentieth-century pharmacology settled the question in Vulpian’s direction: curare’s active compound, tubocurarine, is a reversible blocker of acetylcholine at the muscle’s receptors, and Bowman’s review describes the nondepolarising blockers descended from it in exactly those terms. The nerve still fires; the muscle simply does not receive the message.

None of this lessens Bernard’s finding. The pattern he discovered — heart, blood and muscle intact, nerve-to-muscle command lost, sensation spared — was correct, and it located the problem in the right place. What changed was the interpretation of what happens there. The same junction is where the disease myasthenia gravis does its damage, which is why the history of curare and the history of that disease keep crossing.

Back to Table of Contents

7. Scarlet Blood: Carbon Monoxide (1846)

Carbon monoxide was a known killer in the nineteenth century, given off by burning charcoal and coal, but how it killed was not understood. Bernard’s account in the Introduction begins: “About 1846, I wished to make experiments on the cause of poisoning with carbon monoxide.”

He poisoned a dog with the gas and opened it after death. The striking thing was the colour of the blood: “its blood was scarlet in all the vessels, in the veins as well as the arteries.” Normally, blood in the arteries, freshly loaded with oxygen from the lungs, is bright red, while blood in the veins, having given up much of its oxygen to the tissues, is darker. In carbon monoxide poisoning the veins were as bright as the arteries.

He repeated the observation in rabbits, birds and frogs, and found the same scarlet blood each time. As with curare, a result shared across very different animals pointed to something basic — this time about the blood itself.

The site’s page on carbon monoxide poisoning covers the condition as medicine understands it today, and the Carbon Monoxide page covers the gas and its sources.

Back to Table of Contents

8. A Hypothesis That Failed

The scarlet veins seemed to point to an obvious explanation, and Bernard took it. If venous blood was as red as arterial blood, perhaps it still held its oxygen. Perhaps carbon monoxide stopped the tissues from taking oxygen out of the blood, so the oxygen simply stayed where it was and the body suffocated in the middle of plenty.

He returned to the problem a decade later. “In 1856,” he wrote, “in my course at the College de France on toxic and medicinal substances, I again took up the study of poisoning by carbon monoxide which I had begun in 1846.” This time he set out to test his idea directly. If the poisoned blood still held its oxygen, it ought to be possible to drive that oxygen out and measure it. The method of the day was to expose the blood to another gas — here hydrogen — which would displace the dissolved oxygen so that it could be collected.

The test failed. Hydrogen could not liberate oxygen from the poisoned blood: the oxygen he expected was not there. Bernard drew the conclusion in a short sentence that has become one of the best-known in the Introduction: “My preconceived idea was therefore false.”

He used this episode to teach a point about method. An experimenter needs ideas to guide him, but must hold them loosely and accept the verdict of the experiment when it goes against him. The failed hypothesis was not wasted; it narrowed the problem. The red colour did not mean the blood was full of oxygen. Something else was making it red. The Legacy page returns to how this lesson shaped the Introduction as a whole.

Back to Table of Contents

9. Oxygen Displaced Volume for Volume (1856)

The decisive experiment turned the question around. Instead of looking for oxygen left behind in poisoned blood, Bernard started with normal arterial blood, rich in oxygen, and added carbon monoxide to it, working over mercury so that every bubble of gas could be collected and analysed. He shook the blood with carbon monoxide in a closed tube and then examined the gas above it.

The air in the tube, he reported, had become “remarkably enriched with oxygen.” The carbon monoxide had gone into the blood, and oxygen had come out. He described what he saw as “an exchange, volume by volume, between the carbon monoxide and the oxygen of the blood.” For each volume of carbon monoxide the blood took up, it gave out a volume of oxygen.

The second half of the finding explained why the poisoning was so deadly. Once the carbon monoxide had taken oxygen’s place, it “remained chemically combined in the blood and could no longer be displaced either by oxygen or by other gases.” Oxygen could not push it back out. The blood was red because it was full — but full of the wrong gas, held in the very place where oxygen is normally carried.

Haemoglobin and the firmer bond

Elsewhere in the same book Bernard put the conclusion in terms of the blood’s red pigment: carbon monoxide, he wrote, “is deadly when uniting more firmly than oxygen with the hemoglobin.” In modern words, carbon monoxide and oxygen compete for the same carrier in the red blood cells, and carbon monoxide binds far more firmly. Breathnach’s 2014 review summarises the finding the same way: carbon monoxide paralyses the carriage of oxygen by taking its place on haemoglobin. Sternbach and Varon’s 2003 profile in Resuscitation treats this experiment as the origin of the modern understanding of carbon monoxide poisoning, and Otto Warburg, writing in 1928, credited Bernard with discovering the compound now called carboxyhaemoglobin in the middle of the nineteenth century.

This was a new kind of explanation of a poison. The tissues were not attacked directly; the blood simply stopped delivering oxygen, so the body suffocated with its lungs full of air. The site’s animation of the haemoglobin–oxygen curve shows how haemoglobin loads and releases oxygen normally.

Putting numbers on it

Bernard’s result was qualitative: a one-for-one exchange and a firmer bond. Measuring how much firmer took another half-century. John Haldane’s 1895 paper “The Action of Carbonic Oxide on Man” and the 1912 paper by C. G. Douglas, J. S. Haldane and J. B. S. Haldane on the laws of combination of haemoglobin with carbon monoxide and oxygen turned Bernard’s competition into measured quantities. That work, and carbon monoxide as a clinical problem today, are covered on the Legacy page.

Back to Table of Contents

10. The 1857 Lectures and the Disputed Dates

Bernard gathered his poison work in a book of lectures published in Paris by Baillière in 1857: Leçons sur les effets des substances toxiques et médicamenteuses (“Lectures on the effects of toxic and medicinal substances”). The lectures came from his course at the Collège de France, the same course in which, by his own account, he took up carbon monoxide again in 1856. Curare and carbon monoxide both appear in it.

The book’s title made a quiet claim that shaped later pharmacology. Poisons and medicines were not two separate subjects, one for toxicologists and one for physicians. Both were chemical substances acting on particular tissues, and both could be studied by the same experimental method: find where the substance acts and what it does there. This is the idea behind the modern search for a drug’s “site of action,” and it is why Bernard is counted among the founders of experimental pharmacology as well as of physiology.

Bernard kept returning to curare for the rest of his career. He wrote the essay “Du curare” for the general readers of the Revue des Deux Mondes in 1864, and he devoted his 1864–65 course at the Collège de France to it. In 1865 both poisons took their place as worked examples in the Introduction to the Study of Experimental Medicine, which is why his own account of them is so well known.

Where the dates disagree

Because Bernard told these stories more than once, years after the events, and because historians have used different sources, the dates do not all agree. The main differences:

The broad shape of the story is not in doubt: curare work beginning in the mid-1840s, carbon monoxide observations from about 1846, both analyses completed in 1856–57, and both retold in 1864–65.

Back to Table of Contents

Key Research Papers

  1. Black J. Claude bernard on the action of curare. BMJ. 1999;319(7210):622. PubMed PMID: 10473481
  2. Raghavendra T. Neuromuscular blocking drugs: discovery and development. J R Soc Med. 2002;95(7):363-7. PubMed PMID: 12091515
  3. Cousin MT. Vulpian and not Claude Bernard first proposed the hypothesis of the motor end-plate as the site of action of curare. Anesthesiology. 2002;97(2):527-8. PubMed PMID: 12151956
  4. Sternbach GL, Varon J. The Resuscitation greats. Claude Bernard: on the origin of carbon monoxide poisoning. Resuscitation. 2003;58(2):127-30. PubMed PMID: 12909373
  5. Bowman WC. Neuromuscular block. Br J Pharmacol. 2006;147 Suppl 1(Suppl 1):S277-86. PubMed PMID: 16402115
  6. Barbara JG. [Claude Bernard and his successors on curare: epistemological questions at stake]. J Soc Biol. 2009;203(3):227-34. PubMed PMID: 19833069
  7. Breathnach CS. Claude Bernard and his revelations in physiology. Ir J Med Sci. 2014;183(1):139-46. PubMed PMID: 24297053
  8. Paton WD. An experiment of Claude Bernard on curare: the origins of the Physiological Society [proceedings]. J Physiol. 1976;263(1):26P-29P. PubMed PMID: 796424
  9. Haldane J. The Action of Carbonic Oxide on Man. J Physiol. 1895;18(5-6):430-62. PubMed PMID: 16992272
  10. Douglas CG, Haldane JS, Haldane JB. The laws of combination of haemoglobin with carbon monoxide and oxygen. J Physiol. 1912;44(4):275-304. PubMed PMID: 16993128
  11. Dale HH, Feldberg W, Vogt M. Release of acetylcholine at voluntary motor nerve endings. J Physiol. 1936;86(4):353-80. PubMed PMID: 16994763
  12. Jensen J, Puissant C. Claude Bernard’s route to the isolation of glycogen: the journey that changed scientific views on the physiological role of the liver and animal metabolism. Eur J Appl Physiol. 2026;126(2):629-647. PubMed PMID: 41417056

PubMed Topic Searches

  1. https://pubmed.ncbi.nlm.nih.gov/?term=Claude+Bernard+curare
  2. https://pubmed.ncbi.nlm.nih.gov/?term=Claude+Bernard+carbon+monoxide
  3. https://pubmed.ncbi.nlm.nih.gov/?term=curare+history+neuromuscular+junction
  4. https://pubmed.ncbi.nlm.nih.gov/?term=carboxyhemoglobin+history

Further Reading

Back to Table of Contents

Connections