1. Introduction

The case of Ignaz Semmelweis and childbed fever was made famous by Hempel in the Philosophy of Natural Science (Reference Hempel1966) and has been a staple in philosophy of science since. It was significantly expanded upon by Peter Lipton (Reference Lipton1991/2004), and, more recently, has undergone a resurgence at the hands of Alex Broadbent (Reference Broadbent2009), Alexander Bird (Reference Bird2010), Donald Gillies (Reference Gillies2005), and Raphael Scholl (Reference Scholl2013).

While these accounts differ on what exactly the case of Semmelweis shows, they all agree that Semmelweis was an excellent reasoner.Footnote ¹ Because of the widespread agreement on this last point, and also because only brief reflection seems to show that Semmelweis’s conclusions about childbed fever (now known to be a streptococcal infection) were so obviously correct and plausible, there is also a puzzle associated with this case, namely, why his contemporaries rejected his views both vehemently and for a long time, until well after his death.

My goal in this article is to dissolve both the puzzle and the standard view of Semmelweis by showing that, contrary to prevailing opinion, Semmelweis was not the excellent reasoner he has been assumed to be. I will argue that the reason Semmelweis’s views were rejected is that he simply did not provide particularly strong arguments in their favor, and that he failed to address eminently reasonable objections being put to him by his contemporaries. The discrepancy between the standard story and a closer look at this episode casts doubts on existing philosophical accounts that depend on the story of Semmelweis as an important source of support.Footnote ² Given the recent renewed interest in Semmelweis, it also seems important to prevent the mythical version of the Semmelweis story from being invoked as support for further philosophical theses.

I will proceed as follows: I begin with a discussion of the standard view of Semmelweis (sec. 2), and then move on (sec. 3) to a discussion of the alleged puzzle of Semmelweis and the one philosophical solution (Gillies Reference Gillies2005) that is available. I will argue that this solution is untenable but that, however, there is no puzzle that needs solving; the reason Semmelweis’s views were not accepted is that his arguments did not hold up to scrutiny (sec. 4).

2. The Standard Story of Semmelweis

Let us begin with the standard story as it is found in the extant philosophical literature on Semmelweis. According to this story, Semmelweis, a Hungarian obstetrician, was working in the Vienna General Hospital between 1844 and 1848. The hospital contained two different maternity divisions—the first staffed by doctors, the second staffed by midwives. Semmelweis, working in the first division, noticed that its mortality rate from childbed fever was higher than that of the second. Since Semmelweis was interested in finding ways to reduce mortality from puerperal fever, he examined the extant hypotheses about its cause. He found that many were inconsistent with the evidence he had, and so proceeded to exclude them as viable causes. Hypotheses he rejected this way included the view that puerperal fever was caused by atmospheric cosmic-telluric changes, overcrowding, or unhealthy diets; since Semmelweis judged these to be the same in both divisions, they were excluded as causes.

Other hypotheses were not outright inconsistent with the available evidence, and had to be tested first. For example, in the second division the birth position was different, so Semmelweis tried to adopt this in the first clinic, but it did not make a difference. He also found that a priest bearing last sacrament to the dying women would pass through the first division, ringing a bell, so maybe psychological terror was responsible. He rerouted the priest, but, again, no difference was found.

There was, however, a momentous event in 1847. Semmelweis’s colleague Kolletschka was injured on his finger by a student’s scalpel during an autopsy. Kolletschka developed an infection and died an agonizing death. It occurred to Semmelweis that this infection might be the same as puerperal fever, and he realized that cadaveric matter from the scalpel was introduced into Kolletschka’s bloodstream and was responsible for his death. Semmelweis thought that cadaveric material could also be introduced to the women in the maternity wards through doctors who had performed autopsies.

He then tested his hypothesis. He thought that if puerperal fever could be caused by cadaveric matter on the doctors’ hands, and one could destroy this matter on the hands, mortality would decline. He required everyone, after autopsies, to wash their hands with chlorinated lime and mortality fell to the level of the second division, where only midwives, who did not perform autopsies, worked. Later, despite strict hand washing measures, there was another outbreak. Semmelweis realized that this was due to a woman suffering from cervical cancer, and so broadened his hypothesis from cadaveric matter only to any kind of decomposing animal matter.

3. The Puzzle

Why, then, given that Semmelweis’s reasoning seems so obviously correct and plausible, did his contemporaries reject his views for so long? In the philosophical literature, this puzzle has been confronted head on only by Gillies (Reference Gillies2005).Footnote ³ According to Gillies, “the principal reason why Semmelweis’s theory of the causes of puerperal fever was largely rejected by his contemporaries, was that it contradicted the then dominant paradigm in medicine. In other words, Semmelweis experienced difficulties largely because he was a revolutionary scientist” (177).

It is worth discussing Gillies’s view in some detail, both to set up the arguments in section 3, but also because Gillies’s account is the only philosophical account we have available, and his paper is widely cited in the recent Semmelweis literature. My aim in this section is to show that Gillies’s solution is implausible and that, contra Gillies, (i) it is not clear that we can really speak of the paradigm Gillies identifies, and (ii) even if we can, Semmelweis did not contradict it—far from it, as we will see.

According to Gillies (Reference Gillies2005, 174), the dominant paradigm at Semmelweis’s time was a composite paradigm consisting of the miasma theory and the contagion theory. Miasmas were thought to be toxic odors arising from decomposing animal matter, entering the atmosphere, and so causing diseases in people. Contagionism is understood by Gillies as the following: “A contagious disease is one like smallpox which produces a contagion which infects another patient with the same disease. However puerperal fever, according to Semmelweis, is produced by particles from the corpses of people who did not suffer from puerperal fever” (175).

At first glance, Gillies’s argument looks rather convincing: it certainly looks like Semmelweis went against the contagionists, and, in conjunction with the claim that the contagionist view was the dominant one at the time, it appears that Semmelweis did indeed contradict what was considered the mainstream view. However, on closer inspection, it is not clear that it is appropriate to speak of anything like a mainstream contagionist view at this point.Footnote ⁴ While there were some clear-cut cases of diseases that people agreed were contagious (smallpox and measles, for example), most diseases were very much unlike these classic contagious diseases. Most diseases—such as cholera, yellow fever, and typhus—were thought to exhibit some properties of contagious and some properties of noncontagious diseases, and many people adopted a view called “contingent contagionism,” holding that such diseases were only sometimes contagious: they could, for example, become contagious under unfavorable circumstances, such as bad hygienic conditions or overcrowding (thereby also explaining the occurrence of occasional epidemics). In addition, it was thought that the exact form and nature of the disease in a specific person depended on the balance of a variety of additional factors, such as personal morals, the weather, and diet. Here, one can make a distinction between predisposing causes (exposure to cold air, depression, etc.) and exciting causes (miasmas, specific poisons), and, according to some, enough predisposing causes would be sufficient for causing a disease, while according to others, exciting causes were essential (but could be modified by predisposing ones) (see Hamlin Reference Hamlin1992). In fact, it was common for people to hold different combinations of these views for different diseases, so that they might have held a contagionist view of one disease, an anticontagionist one of another, and adopted contingent contagionism about a third.

Given this situation, it is unclear whether it is really appropriate to speak of anything like a contagionist paradigm, either about childbed fever or other diseases. Clearly, if there isn’t such a paradigm, Semmelweis could not have contradicted it. However, for the sake of argument, let us grant this point for now, and assume that talk of such a paradigm is plausible. In this case, did Semmelweis contradict the contagionists? One might be tempted to think so, as is Gillies, because Semmelweis is explicit that he is not a contagionist: “Childbed fever is not a contagious disease. A contagious disease is one that produces the contagion by which the disease is spread. This contagion brings about only the same disease in other individuals. … Smallpox causes only smallpox and no other disease. … Childbed fever is different” (Semmelweis Reference Semmelweis and Codell Carter1861/1983, 117; see also Gillies Reference Gillies2005, 175).

It seems clear from this that Semmelweis is going against the contagionists. However, it turns out that this is not actually the contagionist view. Rather, it had been known since the 1790s not just that doctors and midwives could convey childbed fever to healthy women, but also that there were other diseases, especially erysipelas (a type of streptococcal rash), that could cause childbed fever. Indeed, it was known that this was also possible in reverse: doctors and midwives would often contract erysipelas on their arms after having attended cases of childbed fever. Indeed, examining medical treatises from the 1790s onward, there is an abundance of treatises arguing for a link between these two diseases (see Loudon Reference Loudon1992, chap. 4). Thus, when Semmelweis (and Gillies with him) attributes to the contagionists the view that only childbed fever can cause childbed fever, and that childbed fever can cause only childbed fever, he is making a misattribution. The view that childbed fever was different from smallpox in this respect was not just shared by the contagionists, but had actually been argued for by them since the 1790s,Footnote ⁵ thus undermining Gillies’s claim that Semmelweis contradicted the contagionists’ view.

One might argue, at this point, that the real importance of Semmelweis lay in the fact that he identified the mechanism through which childbed fever was conveyed (Gillies Reference Gillies2005, 167): Semmelweis realized it could be introduced into the genital tract through the hands of medical staff, whereas others thought it was conveyed through air. This view, however, was not new, either. James Young Simpson, a Scottish obstetrician, for example, was quite specific about how childbed fever could be contracted, publishing the following 11 years before Semmelweis:

One might be suspicious at this point as to whether this is really the view that Gillies has in mind; however, Gillies specifically cites Simpson as someone who rejects Semmelweis’s view: “In effect Simpson misinterpreted Semmelweis as saying that childbed fever was a contagious disease. Such misunderstandings of revolutionary ideas are very common among those who are adherents of a dominant paradigm. It is natural for them to reinterpret the statements of the new theory in terms of the concept of the old paradigm” (Gillies Reference Gillies2005, 178–79).

However, as we have seen, Simpson and Semmelweis actually held virtually the same views. It is true that Simpson calls Semmelweis a contagionist; however, this is not because of a misinterpretation of Semmelweis’s view but, rather, because Semmelweis held exactly the same view that had long been held by people calling themselves contagionists, such as Simpson. Thus, far from being revolutionary, many of Semmelweis’s claims were, in fact, already quite widely accepted at the time.

So, which parts of Semmelweis were rejected then, and why? One claim Semmelweis insisted on, contrary to other accounts we have seen, was that every case of childbed fever was caused in exactly the same way: through resorption of decomposing animal matter (see Gillies Reference Gillies2005, 168 and sec. 5, and, especially, Carter Reference Carter1983). While from the miasma theory’s standpoint there was nothing objectionable about decomposing animal matter as a cause of childbed fever, Semmelweis’s insistence on monocausality was in contrast to the miasma theory, which held that the cause could come from various sources.

So, why was Semmelweis’s claim to monocausality rejected? This is where the puzzle really comes in, since, as we have seen, this is where the standard story outlined in section 1 has Semmelweis so obviously and elegantly deliver the correct answer. What I will argue in the next section is that, in fact, Semmelweis simply did not produce a very good case for the monocausality thesis, something that was pointed out time and again by his contemporaries. Taking this into account, there is no puzzle about why his views were rejected: they simply did not hold up to scrutiny.

4. Dissolving the Puzzle and the Myth

Before looking at any of Semmelweis’s arguments in detail, it is important to understand that he went through three successive versions of his theory: first, a version according to which cadaveric matter from corpses was responsible for childbed fever; second, a version according to which, in addition to cadaveric matter from corpses, people sometimes produced their own cadaveric matter, which could also cause the disease; third, a version according to which any kind of decomposing animal matter could cause the disease. It is also worth pointing out that even though Semmelweis made his “discoveries” in 1847, he did not publish the Etiology until 1861. In the meantime, two of his colleagues, Hebra and Skoda, published accounts of his views; however, their discussions restricted themselves mainly to the first version of the theory, according to which only cadaveric matter from corpses could cause the disease, and it is likely that, partly because of this, quite a large number of people associated only the first version of the theory with Semmelweis.

Since the second version did not play a large role, it is useful to divide up people’s reactions into those to versions 1 and 3, respectively. According to version 1, cadaveric matter from corpses was the only possible way for women to contract childbed fever. It is worth noting that, even with respect to this version, Semmelweis was quite dogmatic and insisted on monocausality, claiming repeatedly that cadaveric matter from corpses was the only possible way for women to contract childbed fever. It is quite easy to see why Semmelweis’s contemporaries did not accept this. Hamernik (1850, quoted in Semmelweis Reference Semmelweis and Codell Carter1861/1983, 216), for example, points out that childbed fever was much older than autopsies, and, in addition, of course, it was known that there was childbed fever both in hospitals in which no autopsies were performed (and, indeed, in the second division of Vienna’s maternity ward), and, of course, also in home deliveries, which could not be linked to corpses.

Semmelweis’s main arguments for version 1 were statistical, highlighting a significant decline in mortality after hand washing was introduced in the first division. However, the mortality rates of childbed fever had always fluctuated quite dramatically, both within hospitals and also within towns more generally. This is, in fact, exactly what people were pointing out in response to hearing of Semmelweis’s views, with many prominent doctors asking Semmelweis for more numbers, claiming that the data put forward were insufficient for drawing Semmelweis’s conclusions. And while Semmelweis did eventually provide more data, this was 14 years after he made his observations in the Vienna hospital.Footnote ⁶ However, even if Semmelweis’s argument had been more decisive, it was still clear that cadaveric matter from corpses could not have been the only cause of childbed fever and that there had to be other causes besides.

Of course, one might point out that even in the absence of more convincing data, Semmelweis presented enough in order for people to take his claim seriously as a possibility and that, therefore, they should at least have considered hand washing, even if just as a precautionary measure. It turns out, however, that this was actually a common reaction. The following quotation from Schmidt, for example, a prominent obstetrician from Berlin, is representative: “The Viennese experiment is totally sufficient to warrant caution. I do not demand details. It may be that this is one path that leads to childbed fever; it is certainly not the only one” (Schmidt, quoted in Semmelweis Reference Semmelweis and Codell Carter1861/1983, 227).

After Semmelweis realized that version 1 could not be quite right, he proceeded to amend his view to claim that, in those cases in which no link to a corpse could be established, the nondiseased body must somehow produce its own cadaveric matter and, in this way, cause the disease. Semmelweis was not very specific about how this might happen; yet, once again, he insisted quite dogmatically that these two routes exhausted the possible causes. Upon realizing that there were problems with this view as well, he finally arrived at his final version, according to which any decomposing animal matter could cause the disease, as long as it was introduced into the genital tract. Once again, he insisted that this was the only cause. In many ways, this last version of Semmelweis’s theory is the most interesting one, since it is the most sophisticated. It is possible, of course, that people were taking his views less seriously at this point, since he had already, unsuccessfully, insisted twice before that he had identified the only cause of puerperal fever. However this may be, let us look at the actual arguments that Semmelweis produced in favor of version 3. Version 3 was actually more controversial than version 1: while quite a few people accepted that cadaveric contamination might well constitute a source of childbed fever, many questioned Semmelweis’s evidence for the claim that the disease could also be caused by noncadaveric decomposing animal matter.

Semmelweis’s evidence for this last view came mainly in the form of two examples: the case of the carious knee, and the case of cancer of the uterus. Semmelweis explains that he arrived at his last hypothesis when, even after careful hand washing, there was another outbreak of childbed fever in the first division. He explains that people contracted the disease from two women on the ward, one sick with a carious knee, the other sick with cancer of the uterus, and that decomposed animal matter from these two women was spread around the rest of the first division by doctors who had examined them. Once again, Semmelweis’s arguments failed to convince people. It was objected that (i) more case studies were required in order to assert anything with the definitiveness of Semmelweis, and (ii) that Semmelweis’s story was implausible because neither of the above-mentioned women contracted childbed fever and because it seemed unlikely that the women should produce the disease in others, yet not succumb themselves (and, in fact, even from a modern-day perspective, this is quite odd). Semmelweis, in The Etiology, claims he will deal with this issue, but he never does (Reference Semmelweis and Codell Carter1861/1983, 93, 185). In view of the fact that this was his main evidence for version 3, leaving the issue unsettled seemed problematic to people who, at this point, might be said to be justifiably skeptical rather than irrationally resistant.

There was, however, another piece of evidence Semmelweis introduced. This was related to his claim that all childbed fever is really a species of pyemia, a type of septicemia that produces pus in the blood and a variety of metastatic abscesses all over the body. In support of this view, Semmelweis provided two main arguments. The first was an argument relying on the only series of animal experiments he ever performed and in which he introduced pyemic matter into the genital tract of rabbits. Some of the rabbits then contracted pyemia and died, and Semmelweis claimed that this was evidence that childbed fever could be caused by introducing decomposing animal matter into a woman’s genital tract during medical examinations. Moreover, he claimed that the experiments proved this with certainty. The problem with this argument, however, was that Semmelweis’s contemporaries did not think that pyemia and childbed fever were the same disease, even though they granted that they might appear similar on occasion. Thus, people objected that what Semmelweis had shown, at most, was that introducing pyemic matter into rabbits could produce pyemia in rabbits, but that this was not what needed to be shown (it is also worth pointing out that only eight rabbits were involved, so the numbers were not as conclusive as one might have hoped, at any rate). In fact, people were quite specific about outlining the kinds of experiments they would have liked Semmelweis to perform, all of them essentially—and quite reasonably—involving injecting the rabbits with matter from actual victims of childbed fever, not other diseases. Semmelweis’s only response, however, was to claim that, in his opinion, such experiments were superfluous (Reference Semmelweis and Codell Carter1861/1983, 190).

In addition to the above, there were also some more general reasons why people were suspicious of Semmelweis’s claim that decomposing (or even just cadaveric) organic matter was the only cause of childbed fever. For example, it had been known for a long time that childbed fever was seasonal, with its highest mortality rate in late winter, usually February, and a low in August. Semmelweis simply denies that this is the case. He does give what he thinks is an explanation of the apparently seasonal patterns, namely, that in winter “students resume their studies, including obstetrics, with renewed diligence. … In summer the diligence is noticeably less. The charming surroundings of Vienna are more attractive than the reeking morgue or the sultry wards of the hospital” (Reference Semmelweis and Codell Carter1861/1983, 122). However, once again, this was not viewed as a convincing argument by Semmelweis’s contemporaries since, in fact, childbed fever was highly seasonal, not just in teaching hospitals, such as the Vienna General Hospital, which should have been the case if Semmelweis had been right, but also in pure lying-in hospitals, in home deliveries, and also with respect to town epidemics.Footnote ⁷ He also insisted that puerperal fever was not epidemic, which, again, was simply false.

Lastly, there was the problem that it seemed quite clear that not all cases of childbed fever could be caused by decomposing animal matter, since there were quite a few cases, even in Semmelweis’s own hospital, that could not be accounted for in this way: cases that occurred despite the strictest sanitary measures, cases of childbed fever in home deliveries, and sporadic cases of childbed fever that were less deadly than cases occurring in the epidemic form. Semmelweis’s response was to claim that, in these cases, the disease was caused by self-infection, but that, at any rate, these cases were too rare to be of great importance. Aside from the fact that this seemed quite ad hoc, there was also the problem that the fact remained that the hospital mortality rate, even after the strict sanitary measures instituted by Semmelweis, was still about four times higher than the home delivery mortality rate. This, many people thought, seemed to speak in favor of some other, or at least additional, cause in the hospitals, responsible for this difference. Once again, people argued that although Semmelweis might have identified some cause of childbed fever, there were many cases he could not account for, both in general and also specifically related to hospitals, and so it was claimed that it was reasonable to suppose that there were other causes at play.Footnote ⁸

To sum up: while it is true that Semmelweis made a genuinely novel claim—that all cases of childbed fever had one and only one cause—there is no puzzle as to why his contemporaries rejected this claim: Semmelweis simply did not provide any convincing reason to subscribe to the monocausality thesis. When Semmelweis was, reasonably, asked to perform certain experiments that could have supported his thesis, he declined, and, in addition, it was pointed out that the monocausality thesis failed to explain several salient phenomena associated with childbed fever that could be explained on a multicausal view, such as its seasonality, the fact that it went through epidemics, and that hospital mortality rates (even in nonteaching hospitals) were higher than home delivery mortality rates.

In conclusion, I hope to have shown in this article that the standard account of Semmelweis is indeed misleading and that, contrary to prevailing opinion, there is no puzzle surrounding the reception of Semmelweis’s claims. This is problematic since so many philosophical accounts depend on it as support, among them those by Lipton (Reference Lipton1991/2004), Gillies (Reference Gillies2005), Broadbent (Reference Broadbent2009), Bird (Reference Bird2010), and Scholl (Reference Scholl2013). The next goal will be to show in what ways these accounts are undermined by what details of the story, and in what ways they are affected. The one moral that is clear, however, already at this point, is that we need to be more careful about the ways in which we invoke case studies such as that of Semmelweis in support of philosophical claims.