I. The Asymmetry Latour's Vocabulary Can't Hold
Bruno Latour's The Pasteurization of France is arguably the richest existing account of exactly what Living Value Theory calls mesocosmic reach: how a laboratory fact stops being a laboratory fact and becomes woven into the ordinary coordination of an entire society, its farms, its hospitals, its national self-image, its colonial administration. A disagreement with a book this good has to be earned rather than asserted, and the disagreement I want to develop here is narrower than it might first sound. It is not that Latour is wrong to grant microbes independent causal behavior, to insist they resist, cooperate, and act in ways no scientist simply dictates to them. That insistence is one of the strongest features of the book, and a purely social-constructivist account of Pasteur, in which microbes are just whatever scientists say they are, would be considerably worse.
The disagreement is this. In Latour's vocabulary, an actant is enrolled, stabilized, resisted, allied, tested in trials of strength. These verbs apply without modification to a bacillus, a comet, a fossil, a proton, or, to push the ontology past the book's own examples, a black hole. The vocabulary is deliberately symmetrical: nothing is, by itself, reducible or irreducible to anything else in advance of a trial, and any typing of actants prior to that trial is exactly the move Latour's whole method is built to forbid. And yet a specific bacillus, isolated in a specific French laboratory in the 1870s, reorganized national public health law, family conduct, colonial hygiene policy, and international philanthropy within a single decade, while the overwhelming majority of microbial species, no less real, no less resistant to trials in Latour's own technical sense, have never done anything of the kind and never will. A black hole, imaged, measured, and confirmed beyond any reasonable dispute, still hasn't. Something is producing that asymmetry, and it is not simply that some actants win more trials of strength than others, because that just restates the asymmetry in different words without explaining it.
My claim is that Pasteur's actual problem was never to enroll or reveal microbes as such. Enrolling and revealing, in Latour's sense, is something a laboratory can do for almost anything, and mostly nobody notices. Pasteur's problem, for a small number of specific organisms that happened to sit at points of enormous pre-existing human concern, was to demonstrate that those organisms' behavior could be made reliable, describable, and controllable enough to act on. Everything else the book describes so well, the laboratory as fulcrum, the displacement onto the hygienist movement, the staged public trial, the provincial franchising of local labs, the international philanthropic institute, is the machinery by which that demonstration achieved mesocosmic reach. What follows tries to show why this reframing is not just a relabeling of Latour's story but a correction of a specific, locatable failure in his ontology, one that Latour's own data exposes more sharply than I think he ever noticed, and one his later work, and indeed his entire career, kept circling without quite landing on.
II. What Latour Gets Right, Briefly
The book opens by refusing hagiography, and it is worth granting this in full before disagreeing with anything else. Latour's use of Tolstoy against the Napoleonic historians, refusing to let "Pasteur" absorb the work of everyone whose prior interest, labor, and movement made his laboratory relevant, is exactly right, and exactly the same move Living Value Theory has made elsewhere against theories that are credited to a single genius rather than to the recursive ecology that received them. No idea travels from a laboratory to a nation by its own force. Someone on the receiving end has to seize it, want it, redirect it for their own purposes. This is not a minor concession. It is the correct starting point, and Latour's insistence on it, against the French hagiographers of his own account, is one of the book's genuine achievements.
Equally important, and equally worth conceding without qualification, is Latour's refusal of naive social constructivism. The microbe is not a passive screen onto which Pasteur's laboratory projects whatever meaning it likes. It has its own virulence, its own behavior under specific culture conditions, its own capacity to kill a guinea pig in two days or five depending on how it has been handled. This resistance is real, and any account that denied it would be worse than Latour's, not better.
But it is worth being exact about what kind of resistance this is, because getting it wrong here distorts everything that follows, and it is a mistake this essay's own earlier draft made. A microbe is, in its own biological terms, a living, self-maintaining, adaptive entity, and in its relation to a human population it is a full party to something genuinely interrecursive: human hygiene and medicine exert selective pressure on microbial populations, and those populations evolve in response, which reorganizes human institutions, quarantine practice, and drug regimens in turn, a real, mutually responsive, coevolving relation rather than a one-way use of a passive object. None of this is nonrecursive in Living Value Theory's technical sense. What makes the anthrax bacillus behave, for Pasteur's purposes, as if it held still is something narrower and more precise: it does not respond to how it is described, named, classified, or theorized. Its virulence does not change because Pasteur calls it dangerous, and no laboratory can talk it out of killing a guinea pig. In this one respect, and only this one respect, a selfrecursive, interrecursively entangled organism functions exactly like the sun. The sun is genuinely nonrecursive, with no interior, adaptive, or self-maintaining relation to anything, and people have still coordinated entire agricultural, religious, and daily rhythms around its rising for as long as there have been people to coordinate. The microbe is not nonrecursive in that sense at all, but for the purposes of a laboratory trying to describe it reliably enough to act on, its refusal to negotiate with description makes it operationally indistinguishable from something that is. What follows depends on keeping these apart: what an entity technically is, and how far its behavior reaches into human concern, are two different questions, and Latour's vocabulary has no way to ask them separately.
So the disagreement is not about whether nonhuman entities act independently of human intention. It is about whether all such independent action belongs to a single explanatory register once it enters what Latour calls a trial of strength. That is the question the rest of this essay tries to answer by looking closely at two specific organisms Pasteur happened to work on, one of which was enrolled about as thoroughly as Latour's own vocabulary allows, and one of which barely was.
III. Two Actants, Two Fates: Anthrax Versus Rabies
Anthrax is the case Latour's book, and most accounts of Pasteur, treat as paradigmatic. The bacillus was isolated, cultured, attenuated by controlled exposure to oxygen, and demonstrated in a staged public field trial at Pouilly-le-Fort on June 2, 1881, in front of skeptical veterinarians and journalists, with vaccinated and unvaccinated sheep set side by side to live or die in full view. It is, by any reasonable measure internal to Latour's own vocabulary, a fully enrolled and fully stabilized actant: visible under the microscope, reproducible in culture, demonstrably controllable, publicly witnessed. Farmers in anthrax-stricken regions were, in Latour's own account, impatient to bring their livestock in for vaccination within weeks of the demonstration. This is exactly the story the book tells best, and it is a genuine case of a laboratory achievement translating almost immediately into mass uptake.
Rabies is a different case, and Latour reports it with a candor that undercuts his own framework more than I think he registers. The rabies virus was never isolated, cultured, or visualized by any method Pasteur developed; Latour states plainly that its causative agent "was to remain invisible until the 1930s and could be cultivated by none of the methods that he himself had developed." The vaccine that Pasteur administered to Joseph Meister in July 1885 was produced empirically, by attenuating infected nervous tissue through a drying process, without the causative organism ever having been seen, cultured, or demonstrated in the way the anthrax bacillus had been. Latour, citing the historian Dagognet, is explicit that none of the steps that followed was necessary or inevitable, that contemporaries as significant as Louis Pasteur's own rivals Peter and Koch objected that treating two patients, Meister and one other, was scientifically inadequate grounds for the sweeping claims that followed, and that Pasteur "could have, ought to have" declined to treat a human being on the strength of results that thin.
By any measure internal to Latour's own vocabulary, then, anthrax is the stronger actant. It was more thoroughly enrolled, more stably networked, more publicly demonstrated, more scientifically secure. And yet rabies produced the larger mesocosmic explosion by an enormous margin. Latour's own words for what happened next are worth sitting with: the treatment of two cured patients was converted into what he calls millions of gold francs, raised through international public subscription, which became the permanent endowment of the Institut Pasteur as a global research institution. Two clinical cases, involving a virus nobody had seen, produced a bigger and more durable transformation of the mesocosm than a fully cultured, fully demonstrated, publicly staged bacterial vaccine. If mesocosmic reach tracked the strength of enrollment, this result runs backward. On the account developed below, it is not backward at all. It is exactly what the theory predicts.
The chapter Latour titles "You Will Be Pasteurs of Microbes!" supplies a further piece of evidence worth noting here, because it shows the same asymmetry operating across the whole of Pasteur's career rather than only in the anthrax and rabies cases. Latour traces a repeated pattern: Pasteur moves sideways, again and again, out of crystallography, out of fermentation chemistry, out of the dispute over spontaneous generation, out of silkworm disease, carrying his laboratory methods each time into a new domain and abandoning the previous one to specialists like Koch to classify and describe in detail. What determines which domain he stays in only briefly and which one detonates into provincial franchises, local laboratories, veterinary training, and eventually national legislation, is never the rigor of the enrollment. Crystallography was rigorously enrolled and stayed a matter for a few colleagues. Fermentation chemistry, by contrast, put him at the center of the wine, beer, and vinegar industries, whose economic stakes, in Latour's own words, were out of all proportion to a handful of crystallographers. The pattern repeats with mechanical regularity: whichever organism or process happened to sit inside an existing structure of economic or mortal concern is the one around which local hygienists and veterinarians were trained, in Latour's memorable phrase, to become Pasteurs of microbes in their own right, reproducing the laboratory's authority across the whole of provincial France. This is multimateriality doing exactly the work Living Value Theory would expect it to do, translating a demonstrated, reliably describable finding into distributed material and institutional practice, but it only happens to the organisms that were already relevant. Crystals never got their own provincial franchise.
IV. The Tell: Audacity and Genius
What makes this worth dwelling on is not just the inversion itself but what Latour does when he has to explain it. Having spent the entire book dismantling the hagiographic vocabulary of Pasteurian genius, insisting that no idea moves by its own force and that "Pasteur" must always be read in quotation marks to separate the man from the myth, Latour arrives at the rabies case and needs an explanation for why a scientifically premature, methodologically incomplete, and at the time actively contested decision produced the single greatest expansion of Pasteurian reach in the whole story. His own vocabulary of trials of strength, allies, and displacement, built for exactly this kind of explanatory work everywhere else in the book, goes quiet. What replaces it is the word audacity, and then, offered almost apologetically, the word genius, the same word the book's opening pages had set out specifically to dismantle.
This is not a slip to pass over. It is the moment the flat ontology runs out of resources on its own data, and the author reaches, if only for a sentence, for exactly the explanatory currency his own method exists to refuse. A theory that cannot explain its own best case except by falling back on the vocabulary it was built to replace has identified, without quite naming it, the location of its own limit. That limit is the subject of the rest of this essay.
V. Recursive Relevance, Not Type
The variable that explains the anthrax and rabies asymmetry, and the deeper asymmetry between the handful of microbial species anyone has heard of and the billions nobody has, is not enrollment strength, and it is not recursivity type either. It is what Living Value Theory calls recursive relevance: the degree to which a given process, independent of anyone's interest in it and independent of what kind of process it technically is, already intersects the selfrecursive and interrecursive coordination through which people organize their lives and their relations with one another.
This has to be stated carefully, because it would be easy to hear the previous section as having simply reclassified microbes into a special nonrecursive category and to think recursive relevance is just another name for that reclassification. It is not, and the distinction matters more than the correction of a single term. The overwhelming majority of microbial species are, in exactly the loose biological sense established above, every bit as selfrecursive and interrecursively entangled as the anthrax bacillus or the rabies virus: metabolically active, adaptive, capable of evolving in response to selective pressure, embedded in relations with hosts, soils, and other organisms that are every bit as mutually responsive as the relation between a farmer's herd and an epidemic. What distinguishes anthrax and rabies from the billions of microbial species no one has ever heard of is not that they are differently typed. It is that their behavior happened to intersect an existing structure of selfrecursive dread and interrecursive economic stake, livestock death threatening a farm's survival, a mad dog's bite threatening a child's life, while soil nitrogen fixation, marine archaea, and the overwhelming majority of the human gut microbiome do not, regardless of how genuinely selfrecursive and interrecursively active all of these organisms equally are in their own right. Recursive relevance cuts across recursivity type rather than tracking it. A nonrecursive entity like the sun can carry immense recursive relevance, nobody negotiates with sunrise, and entire agricultural, religious, and circadian orders of human life are built around its schedule regardless. A selfrecursive, interrecursively entangled entity like a microbe can carry equally immense relevance, as anthrax and rabies did, or carry none at all, as most of the microbial world does. What predicts mesocosmic reach is never the type. It is always the relevance, and the two are independent variables that happen, in Pasteur's case, to have been confused with one another because his most famous organisms behaved, for the narrow purposes of laboratory description, as though they held still.
Anthrax and rabies were not selected by Pasteur's laboratory technique out of an undifferentiated field of equally available microbes. They were handed to him already maximally relevant. A disease that kills livestock threatens a farming family's entire livelihood, which is an interrecursive fact about property, credit, and survival that required no laboratory to establish. A disease transmitted by the bite of a household or working animal, with a horrifying and, at the time, universally fatal presentation, frequently striking children, and folklorically loaded for centuries with the image of the mad dog and the specific terror of hydrophobia, was already sitting at a point of maximal selfrecursive dread before Pasteur's laboratory ever touched it. The laboratory's job, in each case, was not to manufacture this relevance and not to change what kind of organism it was dealing with. It was to close the gap between a relevance that already existed, diffusely, in the fear of every farmer and every parent, and a description of that organism's behavior reliable enough to act on. For anthrax, closing that gap required a fully stabilized actant, a bacillus visible, cultured, and demonstrated in public. For rabies, astonishingly, it barely required one at all, because the selfrecursive terror on the other side of the gap was so large that even an incomplete, scientifically contested intervention was enough to detonate it. This is precisely the same variable that, in a very different case, explains why atmospheric chemistry has needed decades of accumulating, diffuse, endlessly deniable evidence to achieve a fraction of the reach a single treated child achieved for Pasteur in a matter of weeks: not because the physics is weaker than the biology, and not because carbon dioxide is differently typed than a virus, but because climate science had to manufacture its own recursive relevance from something people did not yet feel concerned enough to organize around, while rabies needed only to be pointed at a fear that was already universal.
VI. Latour Already Knew: From Matters of Fact to Matters of Concern
What makes this limitation especially interesting is that Latour spent much of the following two decades circling back toward exactly the phenomenon his 1984 ontology could not name. In his 2004 essay "Why Has Critique Run Out of Steam? From Matters of Fact to Matters of Concern," Latour proposes what he calls a second empiricism, one that treats objects not as isolated, indisputable matters of fact but as matters of concern: bundles of relations, contested constituencies, and gathered publics, things that assemble people around them rather than sitting inertly beneath whatever is said about them. The proposal is explicitly a call to trace how facts become entangled with values, stakeholders, and controversy, rather than assuming a bare fact exists underneath the entanglement, waiting to be either accepted or denied.
Read against The Pasteurization of France, this later essay is Latour renarrating, at the level of general method, precisely the asymmetry this essay has been diagnosing in his own earlier case study. The hygienist movement gathering around the anthrax bacillus, the international public gathering around the rabies vaccine, are matters-of-concern stories in every sense the 2004 essay would recognize, twenty years before Latour had that vocabulary available to name the method explicitly. In this sense, Latour's own intellectual trajectory is itself evidence for the argument being made here: he kept returning, book after book, to the observation that some objects gather enormous, diverse, contested publics around them while structurally identical objects, equally real, equally resistant to trials, gather none at all, and he kept reaching for a new vocabulary, actants, then trials of strength, then matters of concern, then later still the modes of existence, to hold onto that observation without ever supplying a typology of what determines how much concern a given matter is capable of gathering, or what kind.
This is the limitation worth stating precisely, because it is easy to think matters of concern solves the problem this essay has been raising, and it does not. It is still a single, undifferentiated category. The 2004 essay asks scholars to notice that a fact gathers concern and to trace that gathering empirically, wherever it happens to occur, but it offers no principled account, prior to the tracing, of why rabies gathered a concern that dwarfed anthrax's, or why the vast majority of microbial species gather none. The same flattening that treated anthrax and rabies as equivalently enrolled actants in 1984 treats them as equivalently rich matters of concern in 2004. The vocabulary became more attentive to publics and gathering without becoming more discriminating about why some things are so much more capable of gathering than others.
The pattern continues past 2004. The companion project Latour developed with Peter Weibel, built around the etymological point that the old Germanic and Norse word for a political assembly, a thing, is the same word later narrowed to mean a mere object, again asks readers to notice that objects gather publics, this time under the heading of a politics of things rather than a sociology of facts. And the later inquiry into modes of existence multiplies the categories further still, proposing more than a dozen distinct modes in which beings can be said to exist, reproduction, reference, technology, fiction, law, religion, and so on, each with its own conditions of felicity and infelicity. This is real progress toward finer-grained description, and it shows Latour continuing to feel that a single flat register is not enough. But multiplying modes of existence is not the same operation as typing a given matter by the kind of mediation it stakes a claim to, selfrecursive, interrecursive, or neither, and predicting from that typing alone, before any particular gathering is traced, roughly how far and how fast its concern is likely to spread. Fourteen modes of existence is still, for the purposes of this essay, one axis short of the axis that would have told Latour in advance why two patients would outweigh a public field trial of vaccinated sheep.
What recursivity typing supplies is exactly the distinction matters of concern gestures toward and never draws: not merely that a matter of fact can become a matter of concern, but that the specific kind of recursive mediation at stake, whether a process intersects selfrecursive coordination, interrecursive coordination, both, or neither, determines in advance, before any particular gathering is traced, what kind of concern is even available to accumulate around it and roughly how far it can reach. Latour's career is, on this reading, the record of someone circling a typology for four decades without ever quite drawing it, because his founding methodological commitments, the refusal of stable interests in 1984 and the refusal of a hierarchy among matters of concern from 2004 onward, forbid him the one move that would have let him draw it.
VII. Why the Hygienists Mattered Anyway
None of this makes the century-old hygienist movement Latour reconstructs so carefully irrelevant to the story; it explains why that movement mattered, more precisely than treating it as the whole explanation would allow. France after 1870 was a mesocosm saturated, at mass scale, with exactly the two registers recursive relevance names: selfrecursive concern, in the fear of contagion, the anxiety about personal and familial degeneration, the felt need for what contemporaries called regeneration after military defeat; and interrecursive concern, in national demographic panic, colonial hygiene administration, and the economic stakes of livestock disease. Latour's own diagnosis of the hygienist movement's weakness, a diffuse, decades-old accumulation of causes that could not be hierarchized, dispersed across too many fronts to concentrate its forces, what physicians of the time called morbid spontaneity, is exactly the phenomenology of a population with an abundance of selfrecursive and interrecursive material and no stable, non-negotiable anchor, a fact that behaved reliably enough, the way sunrise does, to organize a movement around.
This is why the general reframing does not need to treat the hygienist movement as a special case invented for this argument. It generalizes past it. Any two organisms, handed to any laboratory, would have produced the same asymmetry Pasteur's own career produced internally, weak reach for the ones that never touched an existing structure of concern, explosive reach for the ones that did, regardless of how skillfully either was enrolled. The hygienist movement is simply the visible, decades-documented case of the concern already being there, waiting, before any bacillus arrived to be pointed at it.
VIII. The Rebuttal, and the Disclaimer That Blocks It
The obvious reply on Latour's behalf is that his own vocabulary already captures this without needing an extra variable. Trials of strength, on this reply, already register that microbes have more available allies to enlist, farmers, physicians, army doctors, worried parents, than a black hole or an obscure soil bacterium ever will; the asymmetry is explained within the existing framework as a difference in the number and eagerness of allies on hand to be enrolled, not as evidence of some further, prior property the framework is missing.
Latour cannot make this move consistently, because he has explicitly forbidden himself the resource it requires. In his own methodological notes to the book, he states that he does not treat interests as stable, or as existing prior to the trial that supposedly reveals them; interests, in his words, cannot explain science and society, because they are what will be explained once the experiment is over. This is not an incidental aside. It is a considered, principled refusal to let any pre-existing distribution of concern do explanatory work in advance of a specific historical trial. But the rebuttal above requires exactly that resource: it requires saying, before Pasteur ever touched anthrax or rabies, that farmers and parents already had more at stake in these organisms than physicists had in an undiscovered black hole. Recursive relevance is precisely a claim about that prior distribution, and Latour's own rule excludes him from having it. He can describe, after the fact, that Pasteur found many eager allies for rabies and few for, say, crystallography, which he also pursued and abandoned. He cannot say, before the fact, why that would be so, because saying why would require exactly the stable, prior account of interest and relevance his method is built to defer indefinitely, always to be explained later, never to be used as a predictor.
IX. The SWIPE, and the Question Latour Never Got to Ask
There is a more fundamental way to state the limitation running through everything above, and it requires stepping back from this one book to Latour's career as a whole. The Enlightenment did not simply divide the world into nature and culture and leave scholars to pick a side. It performed what Living Value Theory calls a SWIPE, a split of worlds paired with incommensurable epistemologies: nature assigned to causal explanation, measurement, and law-like regularity, culture assigned to interpretation, meaning, and historical specificity. Applied to an entity like a microbe, the SWIPE does something very specific. It reduces the entire available debate to a single question: is this organism a natural fact, real and mind-independent, waiting to be discovered and explained, or is it a cultural artifact, a classification produced by discourse, power, and disciplinary convention, waiting to be deconstructed? Realism against constructivism, discovery against construction. A great deal of the science wars, and a great deal of the sociology of scientific knowledge before and after them, is fought entirely on this terrain.
Latour's career, read as a whole, is the record of someone who spent four decades trying to escape this question by refusing its terms rather than by realizing the terms themselves were never adequate to what actually mattered. His early laboratory studies ask whether facts are discovered or constructed and answer, brilliantly, that the opposition is false: facts are built through networks of instruments, allies, and inscriptions, and become real precisely through that construction rather than despite it. We Have Never Been Modern generalizes the move into a historical claim: the purification of the world into a zone of nature and a zone of culture was never actually achieved, hybrids and quasi-objects have always outrun the two purified categories the moderns claimed to sort everything into. The Pasteurization of France stages the same refusal at the level of a single case: actants must not be pre-sorted into natural or social before a trial of strength decides what they turn out to be. Even the 2004 turn to matters of concern, for all its real advance, is still fundamentally an argument about how facts can remain contested and multiple without collapsing into either naive realism or naive relativism, still an answer, if a subtle one, to the SWIPE's own question about the status of the object.
What this means is that Latour spent an entire career deconstructing a binary that recursive relevance never needed deconstructed in the first place. Whether a microbe is a natural fact discovered by science or partly constituted by the laboratory practices that reveal it was never actually the question that determined whether Pasteur became a national hero. The anthrax bacillus is exactly as real, exactly as much a network effect in Latour's own sense, as the billions of microbial species that have never mattered to anyone. Settling realism or constructivism about the bacillus settles nothing about its mesocosmic stakes, because the stakes were never a question of ontology at all. The actual question, the one the SWIPE made almost impossible to ask cleanly because it had already collapsed every inquiry about a nonhuman entity into a referendum on its reality, was always narrower and more tractable: does this entity's behavior intersect the selfrecursive and interrecursive coordination of some population, and if so how far does that intersection reach. That question does not require dissolving the nature and culture binary, hybridizing it, or multiplying modes of existence to answer it. It requires only recursive relevance, a variable orthogonal to the realism debate entirely, indifferent to whether the entity in question is triumphantly real, richly constructed, or, as with most things, simply irrelevant regardless of which of those it is.
This is, I think, the deepest sense in which Latour was circling something he could never quite land. He kept noticing, book after book, that some objects command enormous public stakes and others none, and he kept trying to explain that difference by refining his account of what objects are, hybrid, networked, multiple, existing in different modes, when the difference was never primarily a matter of what they are at all. It was always a matter of how far they reach. The SWIPE bequeathed him a question about ontology. The question he needed was about reach, and no amount of dissolving the ontological binary, however skillfully done, was ever going to produce an answer to a question in a different register altogether.
X. Doesn't Typing by Recursivity Just Reimport the SWIPE?
A fair objection follows immediately. If the complaint against Latour is that he stayed trapped inside a binary between natural fact and social construction, doesn't sorting processes into nonrecursive, selfrecursive, and interrecursive simply reinstall a version of the same binary under new labels, nonhuman resistance on one side, human concern doing the real explanatory work on the other?
It does not, and the reason matters for the argument as a whole. The three recursivity types cut across the human and nonhuman line rather than tracking it, which is exactly what a SWIPE, by definition, cannot do, since a SWIPE requires its two zones to align cleanly with nature and culture, causal explanation and interpretive understanding. A selfrecursive process, one that changes through its own recursive relation to itself, need not be human in the biological sense; any sufficiently reflexive, self-monitoring system could in principle qualify, and a microbial population evolving resistance under selective pressure is a defensible instance of exactly this, independent of any human observer. An interrecursive process, one constituted by relations among multiple parties who respond to and reorganize around one another, already includes nonhuman participants as full members as soon as other agents' responses to their behavior become part of the ongoing coordination. Quarantine law that changes in response to microbial resistance, and that in turn shapes the selective pressures the microbe faces, is an interrecursive process with the microbe inside it as an active party, not a passive object being acted upon from outside.
Latour's own anthrax and rabies material supplies the clearest evidence that this typology is not smuggling a human monopoly back in. The anthrax bacillus is fully as nonhuman, fully as resistant to trials in Latour's sense, as any microbe in the vast, mesocosmically silent majority. What made it interrecursively relevant was not anything special about its biology relative to theirs; it was that livestock death was already bound up, before any laboratory touched it, in relations of credit, property, and survival among farmers, buyers, and veterinarians, relations the bacillus itself became a full and active participant in the moment it started killing animals inside them. Recursivity typing does not claim nonhumans are inert until humans notice them. It claims that processes differ, independent of any particular human noticing them, in whether their behavior is of a kind that self-relation or interrelation can organize around at all, which is a question about the shape of a relation, not a question about which side of a nature and culture line an entity falls on. That is the distinction the SWIPE was never built to draw, because the SWIPE only ever had one axis, real or constructed, and this argument has been about a different axis entirely.
XI. Coda: Pasteur's Actual Achievement
Pasteur's achievement, on this account, was not enrolling microbes, in the sense of building networks robust enough to survive Latour's trials of strength. Claude Bernard built comparably robust networks around a comparably rigorous laboratory science and never came close to comparable reach, precisely the control case Latour's own book supplies and precisely why Bernard's laboratory stayed, in Latour's own phrase, a sanctuary of science rather than an obligatory passage point. Nor was Pasteur's achievement, in any simple sense, revealing nonhuman actants, since the actant that mattered most of all, the rabies virus, was barely revealed at all, unseen, uncultured, unvisualized for another half century. And it was never a matter of what kind of entity he was working with, since anthrax and rabies were no more and no less selfrecursive, no more and no less interrecursively entangled with human populations, than the overwhelming majority of microbial species that have never troubled anyone.
His achievement was closing the gap between organisms already sitting at points of maximal selfrecursive and interrecursive concern, a farmer's fear for his herd, a parent's fear for a bitten child, and a description of those organisms reliable enough to act on, sometimes, as with anthrax, by fully stabilizing the actant in question, and sometimes, as with rabies, almost without doing so, because the concern on the other side of the gap was already large enough that a thin, contested, incomplete intervention was sufficient to detonate it. That gap-closing, not enrollment, not revelation, and not any fact about what kind of process a microbe technically is, is what mesocosmic reach names. It is the variable Latour's flat ontology was never going to be able to see, not because he lacked the vocabulary for gathering and concern, he built four decades of increasingly rich vocabulary for exactly that, but because the question he inherited from the Enlightenment SWIPE, whether the microbe was real or constructed, was never the question that determined how far it would travel. The question was always how far, and through which kind of recursive mediation, an entity's relevance could reach into the mesocosm. That is the question this essay has tried to ask in his place.