One, I think that science is just really hard. When Leeuwenhoek saw microbes under his microscope, he had no idea that they caused disease. How could he have guessed? It's easy for us to see in hindsight that he had the first clue, but it was very nonobvious at the time.
And the truth can be weird. Imagine it's the 1850s and you're thinking about the germ theory. So… where do these germs come from? Are you telling me that microbes are everywhere, all the time? They're on our tables, in our food and our water? They're just floating around in the air, ready to land on us if we get an open wound? Are they in our bodies already? They're *everywhere*? That could be hard to believe. Indeed, Pasteur did a bunch of experiments to prove this.
Experiments and data can be hard to interpret. There are always confounding factors. Semmelweis had clear data to point to in the disease rates between the two wards, but as I recall one doctor pointed out that the ventilation systems were also different, so if you believed the miasma theory you could think that was the culprit. Florence Nightingale swore that she had seen one disease morph into another as it spread through a hospital ward—something the germ theory said couldn't happen.
Again, science is just really hard.
Another thing is that scientists have to not only discover knowledge but convince their peers of it. These are separate skills. Some people are good at both, like Pasteur. Some are not. I suspect Semmelweis was just bad at this. That doesn't excuse the doctors who refused to listen to him—and of course they were blinded by their own self-image and false pride—but it might help to explain what happened.
I do agree – science can lead us to conclusions that are both weird and nonobvious without the benefit of hindsight. A couple of things on Semmelweis though:
1. It is true that, unlike Pasteur, he was not particularly gifted at scientific communication or presenting his ideas in a way that did not readily antagonize his contemporaries. This definitely made things more difficult for him – thanks for bringing this up!
2. The point re: ventilation systems in the hospital acting as a confound is slightly more complicated. It turns out that the hospital did have data on patient deaths that pre-dated both Semmelweis and the installation of new ventilation that helped make the causal relationship between disease and hand-washing a little bit clearer.
Context from Sherwin B. Nuland’s book on the saga: “Upon becoming its director in 1789, five years after it opened, Boer [the old hospital director]had introduced the methods he learned from his period spent studying with English obstetricians. The hallmarks of their approach were cleanliness, gentleness, and patience. In addition, he refused to use cadavers in the teaching of students, employing an anatomical model instead. Of the 65,000 patients delivered during his thirty-three-year tenure, only 1.3 percent died, whether of puerperal fever or any other cause; in his final year, the mortality was 0.84 percent. It was now completely comprehensible that [Johann] Klein [the new hospital director who was in-charge during Semmelweis's tenure], who taught on cadavers and used far less gentle techniques during labor and delivery, should have had a high mortality during the entire period of his directorship. *Moreover, it had long been known that the incidence of puerperal fever declined when the amount of teaching declined.* That all of this information was readily available was due to the careful record keeping that had characterized the Allgemeine Krankenhaus since its beginnings.”
I think a real challenge here is that miasma theory (while completely false), like Ptolemaic theory, is actually fairly good at making practical and actionable predictions, especially pre-penicillin. If you take it seriously, you strive for good ventilation, you have open windows, doctors and nurses wear masks when treating infectious outbreaks, you prescribe exercise and good air for chronic conditions, people with allergic reactions should move around, etc.
Florence Nightingale's research demonstrated the importance of open windows and good sanitation, which is a result that is, of course, fully consistent with miasma theory.
If you look into ancient miasma theory, it's clear that they had derived some useful empirical observations about infection from reality, but their theory was only directionally correct. The big question is, "how do I operationalize this," and leaning into nonsense with some overlap into reality was often more operationaizable.
"Just as certain diseases spring from contact with the body, so the mind also transmits its own evil to those nearest. A single person who bids us waste our time, or encourages our desires, or provokes our anger, is enough to infect us.”
Great article! I think the main answer, as you did allude to but I don't think explore as fully as you mightve done, is the causation/correlation problem. It's just really difficult to prove a causation. Even today in clinical medicine, we struggle to decide whether a bug grown in culture is cause or contaminant. The human nody is just crawling with microbes, most of which are harmless. To complicate matters, many of these commensal bacteria are harmless...until they are not! Certain condition changes can cause overgrowth of 'normal' bacteria which causes disease (e.g. acidity changes in bcterial vaginosis). 'Opportunitistic' infections are also common in hospital i.e. the idea that certain germs seek diseased tissue is actually very correct! It's just not the only correct idea about germs! And to complicate matters further, sometimes its not the overgrowth of 'notmal' bacteria causing disease, but our own immune reaction to it (e.g. marginal keratitis). Many diseases today are still stuck in this causation/correlation pickle. Rosacea MIGHT be caused by overgrowth of Demodex mites, or it MIGHT just be a correlation. In some people it definitely is a cause and killing the mires sorts it out, but not everyone with rosacea has this problem. The dury is out. Another case is the correlation between Alzheimers and a bacterium called Porphyromomas gingivalis, found in the brains of most Alzheimers patients bu not healhthy brains. This bug is normally found causing gum disease and theres a known link between gum disease and Alzheimers. But again, correlation or causation? Pathogen or contaminant? We still dont know! But perhaps in 20 years someone will write a similar article to yours, questioning why it took us so long to realise. The answer will, of course, be that 'doctors are slow to catch on'. The real answer though, is that, as another commenter has said, science is just really difficult.
Again, I agree that science can lead us to absurd conclusions. Knowing this, it's worth thinking of ways to reach consensus around them faster. In very general terms, this seems to be a two-step process: firstly we need to have good scientific tools and methodology (experiments that are well designed and can help make cause clearer, like the work that Pasteur and Koch did), if these lead to results that fall outside the "scientific overton window" then it is worth factoring in this new information with cautious openness.
As always, all of this is much easier said than done but showing that it is worthwhile to consciously move in this direction is part of what makes historical accounts useful!
I agree, which is why I think its important that historical accounts discuss the most salient points. In this case, its not necessarily a problem with concensus, as I've said. It's more the experimental design, except experiments alone can never determine causation, so I have to disagree with you on what the problems here are. The solution is improving the science of causation. One correlation does not imply causation, but with a good enough model of the relevant causal network (i.e. a directed acyclic graph showing the interactions between exposures, outcomes, and covariates), multiple correlations can certainly imply causation. So again, it's not really a problem of 'overton windows' as is commonly assumed by historians. This is a simplification. If somebody provided strong, definitive evidence of something (as Doll did with his smoking studies), most scientiss tend to accept the conclusion anyway. Icould give you several examples in recent memory where findings that go against the grain have been quickly accepted and changed clinical guidelines. One of my favourites is how the RECOVERY trial showed that steroids helped in covid. This is completely the opposite of what was expected by many people. Steroids suppress the immune system! And yet almost overnight, clinical guidelines up and down the country changed to recommend steroids in moderate/severe cases of covid.
Not when the mortality rate is annualized (as in the graph I have shown). But when we look at monthly mortality, it does get there: for example, in October and December 1842 fatality was as high as 29.33 and 31.38 percent respectively.
Great questions in here. A couple of thoughts.
One, I think that science is just really hard. When Leeuwenhoek saw microbes under his microscope, he had no idea that they caused disease. How could he have guessed? It's easy for us to see in hindsight that he had the first clue, but it was very nonobvious at the time.
And the truth can be weird. Imagine it's the 1850s and you're thinking about the germ theory. So… where do these germs come from? Are you telling me that microbes are everywhere, all the time? They're on our tables, in our food and our water? They're just floating around in the air, ready to land on us if we get an open wound? Are they in our bodies already? They're *everywhere*? That could be hard to believe. Indeed, Pasteur did a bunch of experiments to prove this.
Experiments and data can be hard to interpret. There are always confounding factors. Semmelweis had clear data to point to in the disease rates between the two wards, but as I recall one doctor pointed out that the ventilation systems were also different, so if you believed the miasma theory you could think that was the culprit. Florence Nightingale swore that she had seen one disease morph into another as it spread through a hospital ward—something the germ theory said couldn't happen.
Again, science is just really hard.
Another thing is that scientists have to not only discover knowledge but convince their peers of it. These are separate skills. Some people are good at both, like Pasteur. Some are not. I suspect Semmelweis was just bad at this. That doesn't excuse the doctors who refused to listen to him—and of course they were blinded by their own self-image and false pride—but it might help to explain what happened.
I do agree – science can lead us to conclusions that are both weird and nonobvious without the benefit of hindsight. A couple of things on Semmelweis though:
1. It is true that, unlike Pasteur, he was not particularly gifted at scientific communication or presenting his ideas in a way that did not readily antagonize his contemporaries. This definitely made things more difficult for him – thanks for bringing this up!
2. The point re: ventilation systems in the hospital acting as a confound is slightly more complicated. It turns out that the hospital did have data on patient deaths that pre-dated both Semmelweis and the installation of new ventilation that helped make the causal relationship between disease and hand-washing a little bit clearer.
Context from Sherwin B. Nuland’s book on the saga: “Upon becoming its director in 1789, five years after it opened, Boer [the old hospital director]had introduced the methods he learned from his period spent studying with English obstetricians. The hallmarks of their approach were cleanliness, gentleness, and patience. In addition, he refused to use cadavers in the teaching of students, employing an anatomical model instead. Of the 65,000 patients delivered during his thirty-three-year tenure, only 1.3 percent died, whether of puerperal fever or any other cause; in his final year, the mortality was 0.84 percent. It was now completely comprehensible that [Johann] Klein [the new hospital director who was in-charge during Semmelweis's tenure], who taught on cadavers and used far less gentle techniques during labor and delivery, should have had a high mortality during the entire period of his directorship. *Moreover, it had long been known that the incidence of puerperal fever declined when the amount of teaching declined.* That all of this information was readily available was due to the careful record keeping that had characterized the Allgemeine Krankenhaus since its beginnings.”
I think a real challenge here is that miasma theory (while completely false), like Ptolemaic theory, is actually fairly good at making practical and actionable predictions, especially pre-penicillin. If you take it seriously, you strive for good ventilation, you have open windows, doctors and nurses wear masks when treating infectious outbreaks, you prescribe exercise and good air for chronic conditions, people with allergic reactions should move around, etc.
Florence Nightingale's research demonstrated the importance of open windows and good sanitation, which is a result that is, of course, fully consistent with miasma theory.
Great piece.
If you look into ancient miasma theory, it's clear that they had derived some useful empirical observations about infection from reality, but their theory was only directionally correct. The big question is, "how do I operationalize this," and leaning into nonsense with some overlap into reality was often more operationaizable.
"Just as certain diseases spring from contact with the body, so the mind also transmits its own evil to those nearest. A single person who bids us waste our time, or encourages our desires, or provokes our anger, is enough to infect us.”
— Seneca, Letters, 7.2
Great article! I think the main answer, as you did allude to but I don't think explore as fully as you mightve done, is the causation/correlation problem. It's just really difficult to prove a causation. Even today in clinical medicine, we struggle to decide whether a bug grown in culture is cause or contaminant. The human nody is just crawling with microbes, most of which are harmless. To complicate matters, many of these commensal bacteria are harmless...until they are not! Certain condition changes can cause overgrowth of 'normal' bacteria which causes disease (e.g. acidity changes in bcterial vaginosis). 'Opportunitistic' infections are also common in hospital i.e. the idea that certain germs seek diseased tissue is actually very correct! It's just not the only correct idea about germs! And to complicate matters further, sometimes its not the overgrowth of 'notmal' bacteria causing disease, but our own immune reaction to it (e.g. marginal keratitis). Many diseases today are still stuck in this causation/correlation pickle. Rosacea MIGHT be caused by overgrowth of Demodex mites, or it MIGHT just be a correlation. In some people it definitely is a cause and killing the mires sorts it out, but not everyone with rosacea has this problem. The dury is out. Another case is the correlation between Alzheimers and a bacterium called Porphyromomas gingivalis, found in the brains of most Alzheimers patients bu not healhthy brains. This bug is normally found causing gum disease and theres a known link between gum disease and Alzheimers. But again, correlation or causation? Pathogen or contaminant? We still dont know! But perhaps in 20 years someone will write a similar article to yours, questioning why it took us so long to realise. The answer will, of course, be that 'doctors are slow to catch on'. The real answer though, is that, as another commenter has said, science is just really difficult.
Thank you for the thoughtful reply!
Again, I agree that science can lead us to absurd conclusions. Knowing this, it's worth thinking of ways to reach consensus around them faster. In very general terms, this seems to be a two-step process: firstly we need to have good scientific tools and methodology (experiments that are well designed and can help make cause clearer, like the work that Pasteur and Koch did), if these lead to results that fall outside the "scientific overton window" then it is worth factoring in this new information with cautious openness.
As always, all of this is much easier said than done but showing that it is worthwhile to consciously move in this direction is part of what makes historical accounts useful!
I agree, which is why I think its important that historical accounts discuss the most salient points. In this case, its not necessarily a problem with concensus, as I've said. It's more the experimental design, except experiments alone can never determine causation, so I have to disagree with you on what the problems here are. The solution is improving the science of causation. One correlation does not imply causation, but with a good enough model of the relevant causal network (i.e. a directed acyclic graph showing the interactions between exposures, outcomes, and covariates), multiple correlations can certainly imply causation. So again, it's not really a problem of 'overton windows' as is commonly assumed by historians. This is a simplification. If somebody provided strong, definitive evidence of something (as Doll did with his smoking studies), most scientiss tend to accept the conclusion anyway. Icould give you several examples in recent memory where findings that go against the grain have been quickly accepted and changed clinical guidelines. One of my favourites is how the RECOVERY trial showed that steroids helped in covid. This is completely the opposite of what was expected by many people. Steroids suppress the immune system! And yet almost overnight, clinical guidelines up and down the country changed to recommend steroids in moderate/severe cases of covid.
The physicians' section has higher mortality than the midwives', but it doesn't seem like it got anywhere near 30%?
Not when the mortality rate is annualized (as in the graph I have shown). But when we look at monthly mortality, it does get there: for example, in October and December 1842 fatality was as high as 29.33 and 31.38 percent respectively.