Making Compact Symbolic Structures

Abstract

This chapter provides readers with an insight into the basic processes of an evidentially driven idea system. It uses a concrete case analysis to explain how social actors create and assemble information, codes, preexisting ideas, and other elements from visibly extraordinary occurrences and correlations and other signs that are ultimately useful for witchcraft accusations.

…we have demonstrated through divine and human authorities, and by the evidence gathered throughout antiquity, and by divine along with human laws, experience, judgements, convictions, confrontations and confessions that the transportation of sorcerers is therefore possible; it follows that both sterility and storms are achieved through their [the sorcerers’] influence.

—Jean Bodin, Of the Demon-mania of the Sorcerers (1580) (A different translation of the passage was cited in Virginia Krause, “Witchcraft Confessions and Demonology,” in The Witchcraft Reader, 2nd ed., ed. Darren Oldridge (London and New York: Routledge, [2002]2008), 307. For original text, see Jean Bodin, De la démonomanie des sorciers [Of the Demon-mania of the Sorcerers] (Paris, France: Chez Jacques Du-Puys, [1580]1587), Book 4, Chapter 3, 464).¹

The Great European Witch Hunts refers to the substantive waves of witch trials and prosecutions during the early modern period, from aapproximately1450 to 1750, with a variation of 50 years.² Even though witch prosecutions took place during the medieval period, it is during the early modern period that the number of trials and prosecutions reached its highest point and then subsided.

Roughly 90,000 trials were held throughout Europe between 1500 and 1800, with around a 50% execution rate; this means about 45,000 people were actually penalized by death.³ Around half of the witch trials took place in what is now Germany. There were also additional deaths due to imprisonment, suicide, torture, or illnesses from the harsh and unsanitary conditions during incarceration.

While the scale of death is insubstantial compared to numerous wars, plagues, famines, and natural disasters, especially for the entire European continent across a time span of around 300 years, the witch hunts nonetheless continue to form a substantial part of today’s consciousness for qualitative reasons—in the distinct dynamics, environments, and events they galvanized. This fact notwithstanding, numbers (quantitative aspects) do matter if they are densely concentrated in particular historical locations or against specific populations.

Some cities had seen mass trials of several hundred people—involving thousands—that surely impacted an entire district or region. Quedlinburg, Saxony-Anhalt, in 1589, saw “133 witches burnt in one day.”⁴ Historians report that in certain towns and villages, the entire adult female demographic was at risk of eradication, as women were dying at such a fast rate after several waves of prosecutions.⁵ Even in places where few prosecutions ultimately resulted, the feeling of continuous fear and uncertainty was repeated in many communities when vivid accounts were being circulated, with prominent people participating as witnesses. And if a smaller-sized community conducted unofficial “swimming tests” on suspects, even though there may only have been a few lives at stake, the spectacle could still involve half of all adult villagers.

These events dramatically introduce new phenomena, new and contested collective representations, and involve the emotional, mental, and physical energy of many community members to negotiate new norms and eventual outcomes. What is so phenomenal about witch hunts that warrants historical examination is not the quantity of people they involved, but rather the quality of the events, the dynamics, and the environments they created that were at once unique and abhorrent—yet so eternal to human societies that they transcended these particular historical locations.

The forms, causes, processes, effects, and scales of witch hunts were diverse in nature. Taken as a whole, the “Great Hunt” overwhelmingly affected more women than men, but mixed-gender composition was not infrequent. In many instances, witch prosecutions were isolated and never caused a “panic” in a community. The variations in execution rates, political intervention, and legal regulations that have been meticulously analyzed in historical research are too numerous to recount here.⁶ Still, it is worth noting that half of all witch executions in Europe took place in German territories—which was dubbed “mother of so many witches”—though the exact reasons might not be entirely linear or clear.

Keeping in mind the incredible varieties and intricacies of patterns and causations, our focus is on the cognitive dimension associated with the Great European Witch Hunts. The main phenomenon of our interest is this: again and again, believable ideas of extreme crimes, at first with minimal material evidence, were being propagated in the community, which then evolved into actual charges in court systems. Suspects often found themselves unable to fend off these aggressive ideas, as the accusations were incredibly hard to resolutely disprove using counterevidence. Subjects who had neither financial nor reputational backing were especially vulnerable. Furthermore, the empirical information and evidence leveled against them had a tendency to multiply upon itself, actualizing, elaborating, and even expanding the original allegations. Occasionally, these charges grew contagiously. One incident would breed additional allegations of witchcraft involving greater numbers of people. Embodying with great momentum and energy, these dynamics eroded barriers of trust in the community, standards of morality, and miscellaneous protections that came with status, knowledge, historical religious devotion, political positions, and even family ties. These events originate from a reconfiguration in the idea system pertaining to malicious witchcraft.

Preexisting Knowledge

To explain what people were seeing and thinking, it is important to first acknowledge that participants encountering incidents and events did not start with a cognitive “blank slate.” People already possessed certain “stocks of knowledge” regarding witchcraft when they looked at their world. The educated elites “from above” had developed stocks of concepts, facts, and ideas that were significantly different to those of the illiterate poor “from below.”⁷

To briefly summarize this difference, the elites had built their stocks through various printed texts. These include the early confirmations of the Devil from monks who wrestled with strange experiences, illnesses, and temptations; theologians who rigorously theorized and debated the characteristics of the Devil, often based on closed readings of biblical passages; and the opinions of uncountable numbers of respected scholars, educators, judges, and writers—including among them Thomas Aquinas, St. Augustine, and many popes.⁸ The general questions around the 1450s were not whether the Devil or witches existed but about their specific properties. For example, how much power did the Devil actually have? Could the Devil physically transport witches?

As the early modern period progressed, two additional sources came to influence the elites’ stocks of knowledge regarding witchcraft. First were the court records and documented statements by confessors and denouncers—which were meticulously compiled because of the shift toward a “modern” legal system guided by an inquisitorial legal doctrine, a point I will explain later. Another new source, related to the first, was the published books and writings devoted to the “science” of demons and the Devil; the corpus is generally known today as European demonology. Demonological writings often included summaries and analyses of case histories—and they were generally written by theologians, physicians, and jurists who had access to the sources of witch prosecution cases. From these sources, the knowledge and ideas of witchcraft evolved from disparate pieces of information in the Bible, classical literature, scholarly writings, and the Canon Episcopi (a legal document in the Middle Ages, part of which addressed the topic of witchcraft) to much more synthetic theories and elaborate ideas.

Noted examples of the demonology corpus include the Malleus Maleficarum (usually translated as Hammer of Witches or Witches’ Hammer), authored by Heinrich Kramer, a professor of theology at the University of Salzburg, Austria, and a Dominican inquisitor. The book had been reprinted in over twenty editions between 1487 and 1620.⁹ Other titles include De la demonomanie des sorciers (On the Demon-Mania of Sorcerers) (1580) authored by the French jurist Jean Bodin, Daemonolatreiae libri tres (Demonolatry) (1595) by the French magistrate Nicolas Remy, Disquisitionum Magicarum Libri Sex (Investigations into Magic) (1599–1600) by the Belgian Jesuit Martin Del Rio, Discours exécrable des Sorciers (An Examen of Witches) (1602) by the French inquisitor Henry Boguet, and Tableau de l’inconstance des mauvais anges et demons (On the Inconstancy of Witches) by the French judge Pierre de Lancre (1612). These texts were used extensively among witch believers and prosecutors. The authors notably express diverging opinions on specific points, but they all contribute to the corpus’ development by integrating new, successful cases in purging demons and witches as well as making personal, synthetic observations.¹⁰

Ideas about activities during the sabbath, the special powers and idiosyncrasies of the Devil and his witches, and methods to detect and combat witchcraft became accepted knowledge that trickled down in different forms in different societies and institutions.¹¹ Around 1900, German archivist Joseph Hansen termed the synthetic understanding of witchcraft, circulated in medieval and early modern Europe, as the “cumulative concept of witchcraft.”¹² This cumulative concept was posited as follows: There were witches who made pacts with the Devil during the sabbaths, which were held in secret places and attended by other witches, and these witches in turn caused harm and impairment upon the world. ¹³ Local and regional deviations from this conception, however, differed widely. We can therefore find multiple cumulative concepts with overlapping similarities when we look into each specific case.

What we want to focus upon here is the developmental (“cumulative”) aspect. These concepts were not hasty, impromptu creations springing suddenly from the mind. The idea systems surrounding witchcraft had reached a previously unseen level of complexity, refinement, and elaboration precisely because they had gone through centuries of theological dialogues, scholastic debates, legal experiences, and other demonstrations. The idea system could easily withstand normal refutations voiced by random individuals. Supporters of witchcraft accusations could readily draw on a rich collection of authorities, interpretations, rulings, and records to organize their claims. And there were regular additions of new “empirical” evidence, and a new roster of elite epistemic authorities—building upon the old—promised to further develop and refine the paradigm of thinking.

Average peasants generated and circulated stocks of knowledge of a different kind. Such stocks consisted of local history, rumors, rich folklores, folk knowledge, and folk beliefs. Throughout Europe, people routinely sought out magical practitioners (shamans, healers, “witches,” and most commonly “the cunning folk”) to cure sickness, solve problems, predict the future, and help them fall in love, hunt treasure, and recover stolen items.¹⁴ To deal with curses and malevolent spirits, they may use “amulets, protective symbols on roofs, an iron nail above a baby’s cot, or an iron horseshoe outside the door” to ward off the “evil eye.” ¹⁵ They may, for example, offer a minor gift in the hope of its acceptance by a widely suspected witch (i.e., a process of “reconciliation”). Many ritualistic testing and extralegal trials of witches—such as “swimming,” “weighing,” or “scratching” the witches—contain practices that stemmed from folk beliefs.

These two stocks of knowledge inspired but also clashed with one another. Elites ridiculed peasants’ superstition constantly, deeming the common folk to be ignorant and misguided. Elites often denounced the ideas that witches could fly to the sabbaths at night, transform into animal forms, and possess abilities as powerful as causing hailstorms. Church authorities strongly condemned the use of “counter-magic” and methods of testing witches, deeming some to be bordering on the heretical. At the same time, the average peasants, who were exposed to official pronouncements and verdicts, would insert their own creativity and synthesis into the dominant ideas, adding “local flavors” to the ideas pertaining to witchcraft.¹⁶ Occasionally, they even managed to prosecute members of the elite class via village committees and local courts. Demonologists, several of whom were jurists, were also profoundly developing their own thinking by sorting through and organizing local cases.¹⁷

Compact Symbolic Structures

Witchcraft as a concept has its preexisting epistemic bases; these bases then dynamically interacted with the influx of new information. Overall, ideas pertaining to witches may be analyzed as a set of compact symbolic structures, or a set of mutually shared and enacted symbolic structures made up of coded information.

Seven Domains of Components

I hereby posit a model that distinguishes seven, loosely connected domains of components for idea systems to come into formation: (1) happenings (or raw reality); (2) pre-coded information; (3) codes; (4) coded information (or specific mental objects, or coded things); (5) data; (6) ideas; and (7) idea systems. This scheme can be used as a starting instrument to analyze a broad range of idea systems, whether they are evidentially or ideationally driven.

These components are arranged in a loose hierarchy—in the sense that the components in each domain could potentially assert a primary role in organizing ideas, violating the normal hierarchical order. Still, the stereotypical order is arranged according to how raw reality is asserted as the basis to cognitively inform the formation of ideas or even the development of whole idea systems. According to such assertions, individual ideas or the integrity of whole idea systems can be validated or invalidated by deductively tracing back to the best information people could gather about raw reality. ¹⁸ Disputes over filtering and codification processes are often based on the assumed, supposed interrelations (of correspondence) that exist between these domains.

• (1) Happenings. Happenings is a domain that includes numerous matters and dynamics—physical or otherwise—beyond the scope of humans’, animals’, or computers’ ability to perceive or cognitively comprehend. This domain may alternatively be defined as “raw reality.”¹⁹

Suppose, for a moment, maleficent witchcraft actually exists. Then whatever the actual events and nonevents that occur in the supernatural realm, and whatever magical power the Devil uses, would be within the domain of happening. This raw reality may not feasibly be comprehended by humans, even if we may seek to explain it by the means of physics, chemistry, biology, and so on.

• (2) Pre-coded Information. This domain consists of potentially perceivable and processable bits of information existing in an unprocessed, “undigested” state by the percepts and cognition of biological or mechanical mechanisms. It either exists outside the field of attention, or it receives only an elementary level of attention—and therefore limited consideration—by biological or mechanical mechanisms. ²⁰

Continuing with the previous scenario, pre-coded information corresponds to the ability with which humans could potentially comprehend the properties of Satan. Although we may not understand the intricate processes, we could see certain clues—such as the effects of Satan’s power in the form of witchcraft. If there are multiple instances of ostensible manifestations, then there is a pool of potentially codable information about the Devil and witchcraft that could be compiled. This is not very different from a three-year-old child who wants to comprehend the properties of fire by observing many fires.

It is also worthy to note that the prefix “pre-coded” does not literally mean that the information has not been coded by others. It primarily means that a social actor has not registered the information at the cognitive level—or at least not at the perceptive level. The dossiers of distant witchcraft cases inaccessible to local judges clearly contain previously coded information, but it is “pre-coded” by those judges, as the information has never registered to their fields of attention and precept.

• (3) Codes. This domain is essentially a reservoir of symbols—in the forms of colors, sounds, words, letters, and numbers—that are systematically organized, so that biological or mechanical mechanisms could process information using symbolic means. The color red on a traffic light is a color code that conveys the message “stop” to people who are familiar with the symbolic system and can quickly convert the color code into cognitive meaning.

As stated in the introduction, in this book the word “code” refers to a conventionalized set of symbols. Relating to our previous example, the choice to call a female a “witch”—instead of a “herbalist,” “heretic,” or “criminal”—involves a process code selection. These codes are categories that embody symbolic meanings, with different implications that come from the characterization. They are also embedded in different network of symbols. A “witch,” for example, would evoke religious concepts like “Satan,” “God,” the Devil,” and so on. A “criminal” more likely evokes a language related to the court of law and the state. Different codes situate the characterizations in different cognitive frameworks, and different frameworks often have a system of vocabularies and signs used for classification and depiction purposes.

• (4) Coded information. Coded information (interchangeable with coded things and specific mental objects²¹) refers to a reservoir of a unified form of code and concrete content. Whereas codes could be applied to a range of possible information, coded things are formed after codes are applied to particular sets of information. For humans, “specific mental objects” are yielded. The word “happy” is different in the English dictionary (merely as a code) than it is when we apply it to our own experience (which forms unique mental objects). Suppose the color “blue” is a code, a particular “blue jacket” one is wearing is a specifically coded thing. For computers, coded information consist of aggregate units of coded patterns that are to be processed as a package, as a whole arrangement.²²

The phrase “diabolic animal” is an abstract category; but after it is mentally applied to the “likeness of a cat owned by Mary”—a concrete piece of pre-coded information—then a specific mental object is created. The cat, or likeness of a cat, becomes merged and unified with “diabolic animal” and becomes perceived, interpreted, mentally processed as such.

• (5) Data. The domain of data is a reservoir of coded or pre-coded information that serves “evidential” purposes to support an idea.²³ Used in reasoning, analysis, or imagination mechanisms, data carry a value for subjects to ascertain or disaffirm phenomena. Data that are systematically organized into a reservoir or a more specified collection may be termed a database. 

Seeing the repeated occurrence of abnormal deaths—all followed by successful witchcraft convictions—would first yield pieces of coded information (specific mental objects). Each piece of such coded things—ostensibly “evidencing” something about witchcraft—could then gradually yield a database concerning witchcraft, from which people could observe its patterns and analyze its properties.

• (6) Ideas. This domain of ideas is a reservoir of unified formations constituted by data, codes, and information as its components and then arranged in accordance to structures. These unified formations are evoked, shaped, and suppressed dynamically in the mind. Although all ideas must rely on codes and coded things and must base themselves on a minimal level of data and information, they could also be derived rather purely by psychic processes.

The idea that someone is a witch, for example, could be developed without any affirmed data or even a rich amount of information. However, a minimal level of information must be involved to construct the basics of such conceptions.

• (7) Idea system. This domain refers to a loosely organized set of ideas that form a complex; the ideas and associated components operate according to patterns that affect the gestalt of the whole system. Each idea system has a schematic set of expressions, worldviews, individuated ideas, figures, attitudes, codes, epistemic pathways, and evidentiary bases that facilitate users to generate individual ideas or process information. The corpus of demonological treatises represents a semi-cohesive idea system.

Overall, separating these loosely hierarchical domains can help us to more precisely consider the level at which departures and disagreements may arise. Departures in selecting pre-coded information are different from departures in selecting codes to be applied. Recklessly combining codes and information is different from using data selectively to substantiate ideas.

Pre-coding Assembly

The starting point of a credible witchcraft accusation was an account of an occurrence. Pre-coded information must meet two requirements in order to become useful for the account. One, it could not be readily ‘explained’ (coded) within the structures of other already-existing systems of ideas—such as by “common sense” or by existing frameworks or paradigms in medicine, science, and theology. In short, it needed to be “out-of-the-ordinary” in some way. ²⁴ Two, it must demonstrate a fit—preferably an extraordinary fit—with the code structures supporting and informed by an idea system pertaining to witchcraft. That is to say, it fits in some extraordinary way with the witchcraft idea system.

Pending a basic pool of pre-coded information, coded things could then be formed to substantiate both simple and complex ideas.

Compactness of Idea Sets

An idea system that is truly brought to life, I postulate, is one that exhibits a highly structured set of ideas that coheres with a rich amount of compact, dense information.

Information density (or informational compactness) was often observed in accounts of witch accusations. In many written records, a rich set of sensory or non-sensory details acted as snippets of information.²⁵ Such snippets of information were also quite “compact,” in the sense that even just a few sentences could create a vivid picture. They are somewhat comparable to the “pixels per inch” of a recorded video; a higher-resolution video could give a story, a vignette, or an account in greater clarity and detail.

There were historical actors who mastered the art of highly visualized language. The evidential power in their accounts is intuitive: people knew the coded events to be true because they could see the coded events. The more sense data are presented to support the same visual picture, the more “self-explanatory evidence” a phenomenon contains; thus, the ideas become all the more self-evidentiary. And the more compact or dense the data are, the less effort it would require for an audience to “see” the phenomenon, the less likely it is that the information would become diluted to a point that an audience would have to assemble the patterns themselves.

Structures of Idea Sets

Now, let us consider the structures of information.

How much information goes into supporting an idea certainly matters—but so does the way in which the information is structured (or unstructured). Figure 2.1 marks the basic distinction between a compact versus a non-compact symbolic structure of information. If the alphabet stands for a point of information, the prime symbols (including prime, double prime, and triple prime) stand for derivatives or variants of information, and combined letters (BC) as conjunction, we can see the basic contrast.

Fig. 2.1

Non-compact and compact symbolic structures

Non-compact information (see the left side of Fig. 2.1) might be lacking in solid patterns or regularities; it may only be constructed in the form of weak or one-time associations. The points of information have sparse, scattered, uncertain, or changing connections. The relationship between information may also lack any points of “closure” or “closure-like arrangements.”

A structure with many points of closure or closure-like arrangements is more cognitively complex. A story that has more characters and plot twists is more complex than one which is lacking such nuances. We can also imagine the differences between complex beliefs and a simple belief system. When each point we pick randomly is connected to several others, a point of information is “backed” or “related” in some way by other points.

That is to say, there is potentially a synergistic (and often evidentiary) effect of information that cannot be captured by a straightforward 1 + 1 = 2 type of “additive” relationship. In the left box, where there are only three links between information of varying strengths, if one link is broken, then the whole idea is at risk. In contrast, the right box shows many more linkages between information, and a missing link has only a tenuous effect on the overall structure of evidencing an idea. This principle can be applied to a single idea—or to be transfigured to apply to a “big idea.”

Relating to witch hunts, the accounts are often found to be complex and intricately structured. One correlative event—for example, a destructive storm that occurs after a suspect utters a curse—may be connected to authoritative concepts, which are in turn tied to other incidents, testimonies, and textual interpretations. Faced with such compact symbolic structures (modeled by the right side of Fig. 2.1), there is a significant implication for an ordinary person. The total amount of works, skills, and resources needed to “undo” the idea can be onerous; attacking one point may necessitate that all the others must also be unpacked and undone. This effort can be especially overwhelming if new points of linkages far exceed that of undoing them. Alternatively, if the next step involves sensitive subjects or authorities, such as a declaration by the Church or a statement from the Bible, then it would be potentially costly to undertake such an endeavor, even if it is realistic to expect an ordinary person to do it at all.

Diverse, Intricate Justification Structures

There are several common forms that create mutually supporting, cohesive structures. These structural forms link different points of information, codes, and idea units in different ways.

A narrative (or story) would be a structure that has characters, a plot, course of events, and perhaps an ending.²⁶

An argument would be a structure that has main claims, subclaims, evidence, and assumptions that ground claims to evidence.²⁷

Or an idea may merely be posited in the form of a cognizant belief, a structure wherein there is a noticeable gap between evidence and propositions, bridged by a self-aware leap of faith or by a taken-for-granted, unquestioned, or authoritative assumption.²⁸

Other less cohesive structures also could exist, such as a vignette, an underdeveloped thought, a repeated symbolic association, a conjecture, or a partially formed argument.

Each of these structures would have a different “justification” dynamic, that is, have its own criteria to make it cohesive. The personalities of the characters in narratives, for example, should not change too abruptly without explanation. The link between evidence and claims has to be warranted in arguments. A belief tends to survive more on value, vision, and faith as well as on unbroken, incorrigible core propositions well protected by easy-to-make defenses (which are called “secondary elaborations”). There are also intricate ways to “justifiably” derive more ideas and arguments from an original idea using these structures, that is, it carries its own rules, its own integrity, within the context of the case.

Considering that each case of witchcraft can have multiple incidents, witnesses, confessions, and investigations—as well as an array of stories, arguments, and beliefs grounded in rich stocks of preexisting knowledge—all moving dynamically as events go on, it would be close to impossible to actually capture the workings of any complex witchcraft case using a visual diagram. If we only look at the accounts through the lens of argument, we are likely to dismiss many of them as unjustified. And if we look at them merely through the lens of imaginative beliefs, we would overlook how they justified themselves through rigorous procedures.

Flexibility and Restrictedness of Empirical Referents

Keeping in mind the diverse forms of ideas, we first delve into the issue of “evidence.” Whether they support a story or an argument, what materials qualify to be coded into data that service the idea system?

Witchcraft could be evidenced by an enormously flexible set of potential referents. To measure the speed of light, a physicist would be restricted to looking at, say, photons as referents. Witch accusers needed empirical proof, but the proof was not tied to a restricted set of empirical details. Instead, it could be tied to a wide range of empirical referents.

Table 2.1 draws contrasts between a case of murder and of witchcraft. The tangible evidence of a murder accusation is hard to attain. If a person has died from a stabbing, the tangible, material evidence would be the knife that had been inserted into the victim’s body. A direct eyewitness would need to be present at one concrete place and time to witness that one specific act, which had occurred in only one place and at one time. More indirect evidence would involve someone seeing a person around the crime scene, or if the murderer somehow confides to someone about the act.

Table 2.1 The relative flexibility of evidence pertaining to witchcraft cases

The range of possible evidence is much more flexible involving witchcraft accusations, which entail a supernatural crime. If the idea system posits that a witch can cause only one kind of sickness (e.g., vomiting blood), the qualifying phenomena would be rather limited. But if the effects of witchcraft were so varied that they can have many derivatives—including headaches, feelings of discomfort, bad weather, or animal problems—flexible referents offer a considerably easier way to establish the presence of witchcraft.

In witch accusations, circumstantial evidence seems easy to come by. In the hands of a witchcraft theorist, “defective rosaries” found at home could be an inconclusive form of material evidence.²⁹ “Circumstantial evidence” is so called in the legal literature because it is indirect and requires an inferencing process. The “defective rosaries” are a piece of circumstantial evidence about a witch’s identity—the information itself does not automatically prove the witch’s identity, unless by indirect inference; this situation is similar to a victim’s lost property found in the house of an accused thief constituting only circumstantial evidence. Whether circumstantial evidence is inherently worse than so-called direct evidence (e.g., testimonies from credible witnesses, including uncoerced self-confessions) is beside the point here. The key is that the inherent flexible referents of witchcraft accusations enabled many social actors to accrue and organize a large amount of circumstantial evidence against an accused.

The Case of Anna Fessler in Langenburg

One concrete case can help illustrate how a series of abnormal, out-of-the-ordinary coincidences or objects would later be assembled into a formal witch trial. The case was presented by Thomas Robisheaux in The Last Witch of Langenburg.³⁰ In 1672, a German village woman (Anna Fessler) who was in her “lying-in period”—the period where a woman was placed on bedrest for several weeks after giving birth—had eaten a cake delivered by her twenty-two-year-old friend and confidante, Eva Küstner. Eva had taken the cakes from her mother’s kitchen. Shrove Tuesday, as was customary in the region, was believed to be an unusual day when “extraordinary social and even cosmic forces”—including supernatural forces of all kinds, both positive and negative—were more active than normal. The cakes were Shrovetide Cakes, intended for the purpose of celebrating this special day.

After eating one of Küstner’s cakes, Anna Fessler died that night, with foam dribbling from her mouth. This simple event correlation was substantiated by several other “unusual” signs that eventually aided its connection to witchcraft, some of which require elaboration:

• (1) Abnormal-Looking Cake: The basket of Shrovetide Cakes Küstner baked looked abnormal and suspicious. In fact, one of the cakes looked like it “had fallen in and was not at all pretty”; the first people (a wagonmaker and his wife) Eva delivered one of the cakes to, refused to eat it in front of Eva.³¹

• (2) Dog’s Strong Rejection: The wagonmaker and his wife threw a piece of cake to the dog, which also refused to eat it. This was the case even when their daughter tried to beat the dog to force it to eat the cake. The dog was believed to have the ability to sense spirits and demons.³²

• (3) Twilight Appearance: Eva delivered the cake during an odd hour—at twilight. Twilight symbolized a liminal time of danger, when dangerous animals, thieves, spirits, and so forth were more active.³³

• (4) The Abnormal Seventh Cake: Three women were acting as Anna’s lying-in guardians in the house during Eva’s visit. They allegedly saw that Eva first presented six little cakes, but then she seemed to pull out a seventh from a mysterious place. According to one woman, this seventh cake looked special: “It was a beautiful little yellow cake… it was taller than the other ones and also much heavier.”³⁴

• (5) Abnormal Eating Response by Victim: One of the women observed that Anna behaved strangely. It was as if she could not stop eating the cake. Anna also ate a cake baked by one of her guardians, subsequently claiming that that cake tasted much better than Eva’s.³⁵

• (6) Social Order Reversal: Eva offering cake to Anna subverted the normal order of the village. Eva was from a miller family, so it was unusual for a person of her position to offer a gift of food to a cottage household, such as that of Anna Fessler. The fact that Eva’s mother had a bad reputation—for being stingy and mean—made it doubly unusual.

• (7) Abnormal Pacing Back and Forth: That night before Anna died, around seven in the evening, her husband saw Anna do something he never had seen before: “she paced around the room several times.”³⁶ Having seen her apparent restlessness, he was puzzled as to whether she was psychologically excited or disturbed.

• (8) Extraordinary Illness Symptoms: Anna behaved strangely after she went to bed. She sat up abruptly and said that she was hurting. According to Robisheaux, “She got up and went into the other room, Michel [her husband] followed her. Anna felt that ‘her body would explode on her.’ Losing some of her personal decorum, she ‘acted like she does when she is dressed and needs to pass some gas but she had on nothing but a shirt.’ She then threw back her head in pain, bending over as if she needed to vomit. Michel thought that something was forcing its way up her throat, but nothing came out… Her torso began to swell. Waves of heat broke over her. An unquenchable thirst seized her. She got up, and her bowels suddenly loosened, as she passed ‘brick red blood.’”³⁷ She struggled more that night, and then slipped into unconsciousness and died.

• (9) Abnormal Internal Organs Seen in Autopsy: Her autopsy report from the physician identified that poisoning was likely, but the exact source was unknown. But he said that some parts of her internal organs were torn, “as if someone had forcibly ripped them apart,” and the condition of the intestines was “as if someone had exploded them with great violence.” The liver was enlarged. Near the spleen was a “large black mass.” In the stomach, “there was still an undigested piece of Shrove Cake.”³⁸

Many empirical materials, in the form of highly visual imageries, were being presented in a highly compact and organized manner. Each imagery carries a direct sensorial appeal (a set of pre-coded information) such as the appearance of the abnormal-looking cake, the scene of the foam dripping from Fessler’s mouth, Fessler’s unusual decorum and loosened bowel movements, and the shocking description of the internal organs with a leftover piece of the Shrove Cake. Individually, these pieces of pre-coded information qualify to become evidence because of their fit to build the idea the illness was abnormal, possibly caused by witchcraft. By providing the information the way they did, the social actors—deliberately or unintentionally—help to organize these pieces of data into a cohesive data set, making the suspicion of diabolic involvement much more “justified.” That collective organization is based on the repeated copresence of different classes of things and events, which suggests a higher pattern of correlative associations.

Causal Linkage from Repeated Associations, Copresences, and Correlations

Humans learn from the relations between two classes of things, two concepts, or between classes of things and concepts through repeated associations.³⁹ Therefore, by listening to and taking part in discourses, the structural relations between things and concepts are learned.

Embedded in the pre-coded information compiled against Eva Küstner was a series of correlative relationships. This form of correlation is highly qualitative in nature—that is to say, it retains its distinct richness in qualitative details. Accounts that build correlations thrive on such richness rather than on highly reified numbers (which are simpler, standardized, filtered codes).

One form of qualitative correlation is correlative copresence.

A single correlative event of the copresence of two unusual signs (or one normal sign coupled with an unusual sign) would render a theoretical connection thin. But if they appear side by side repeatedly, a pattern could be observed—a pattern in which two classes of pre-coded things that readily fit into witchcraft-related codes have a copresent relation. The two classes of pre-coded things should appear at the proximal time and space to create a “copresent” appearance.

It is possible for correlative relations to be established with only a few occurrences. In those cases, some of the information must be extraordinary—imagine that, as a suspect appears, a mysterious-looking creature also materializes. The quantity is limited, but the co-relation, even suggestive, lives out in the details of the moment. Cognitively, because of this co-relation, it becomes justified to become suspicious that some “pattern” is emerging.

Without any unusual things present in the correlation, suspicion would be unjustified. An object such as a house might always appear alongside many suspected or later-verified witches, but a house also appears with many non-witches. Therefore, for a house to be considered unusual, some unusual information accompanying the house must be noted: such as a foul smell or a passing toad. Accordingly, a lightning strike—an extremely rare event—would be a great candidate.

Applying this to the Langenburg case, why would such an extraordinary event—Fessler losing her decorum in the most dramatic manner—be copresent with two, three, and even more extraordinary events, like an abnormal-looking cake delivered during twilight hours, that even a dog would not eat? To conceive it mathematically, if one extraordinary event is unlikely (say a 2% chance), then several unlikely events happening together would be highly unlikely (0.02 × 0.02 × 0.02 = 0.000008). A select set of pre-coded information was elicited to suggest this circumstance—this pretext of the extraordinary copresence of unlikely events. In this pretext, the explanation by “random chance” is easily and justifiably challenged by the explanation by extraneous cause—that is to say, an external cause that led to the extraordinary copresence.

Another form of qualitative correlation is sequential correlations.

If one “class” of things or events seems to consistently, sequentially follow another, then another correlative relationship is established. This is a specific type of correlative copresence, which can be visualized in Table 2.2.

Table 2.2 A simple theorization model based on sequential correlations

Readers can vividly construct a causation between the events happening in Time A and those in Time B. The diagram helps to distill the phenomena into a few select events. Irrelevant pre-coded information, such as the size of the room, color of the flames, and arrangement of furniture, is filtered out of the idea finally being generated. With this informational filtering, readers could process correlative relations much more easily.

There are two ways to look at these correlations.

One is to view, on aggregate, that all qualitative details in column A correlate to all those of column B. That is to say, the two have some kinds of vague correlative relationships. While this approach is efficient, the correlation samples are actually fewer, since it groups together all events in column A in one entity and in column B in another. Consequently, these correlative samples are bound to support ideas in only a generic manner. Essentially, there was only one correlative sample, and upon questioning, the boundary revolving around the grouping of information would appear to be quite vague and arbitrary, thus much more prone to error.

An alternative way, which takes much more effort on the part of a thinker (whether by rumor-makers, legal authorities, or by historians), is to restructure the information so that specific temporal correlations (along with spatial correlations) could more easily be seen. Given enough detailed information to begin with, a well-triangulated account can group together information strategically, while creating separations when needed. With these specific sequences, it is possible to discern more specific sequentially correlative samples; in other words, there would be more correlative samples in the pool of information that could be turned into data to support the idea of witchcraft.

To arrange specific correlations to be seen, each of the correlative samples should beg for a specific explanation for the extraordinary occurrence—and the ultimate explanation for all mysterious correlations should be a unifying, explanatory idea. For example, the sequential correlations between the abnormality of the cake (#1a) and the abnormality of the dog’s behavior (#1b)—all described with a compact amount of information and occurring sequentially—toward it allow a specific extraordinary sequential correlation to be observed, begging for very specific explanations. Moving to #2a and #2b, the sequential correlation between Eva’s unusual delivery context and Anna’s abnormal enthusiasm—equally rich in descriptive details—would serve the same purpose.

Ideally, what could explain the specific extraordinary copresence between #1a and #1b can also be used to explain the correlation between #2a and #2b. Doing so would mean that one explanation simultaneously explains three correlative samples: #1a and #1b, #2a and #2b, and [#1a and #1b as a grouped event] and [#2a and #2b as a grouped event]. Multiplying this process forward with more information, data, and organization, a unifying pretext could explain many correlations specifically. To make this possible, the vivid details in witchcraft accounts are not ornamental, but pivotal, in their cognitive purpose.

Data Set: How Much Correlative Evidence Is Enough?

A larger idea system may be supported by hundreds, or even thousands, of these samples. The database being developed in the later period of the Great Hunt was indeed enormous. But for local cases, a few correlative samples, preferably of high quality (i.e., demonstrated extraordinariness of correlations), are what could be reasonably demanded from accusers. This situation does not indicate a lackluster judicial or cognitive standard. Consider the hypothetical case of murder, one also cannot reasonably expect to find thousands of extraordinary correlative forms of evidence. What matters here, as it was in the case of witchcraft cases, is the quality of a few remarkable “coincidences.” It was up to court officials to strategically connect local cases to the other correlative samples in the larger database, a topic we will discuss in the following chapter.

Using Language to Organize Copresent and Sequential Correlations

Language use was important in organizing compact information so that different dramatically amplified details seemed to be appearing “together.” One specific role it played to make correlation seeable was compressing time and space dynamically.

Historian Robin Briggs narrated an instance where a witchcraft suspect (Mother Staunton of Wimbish, England) had, upon being refused milk, suspiciously “made a circle in the ground with a knife” by the door of the refuser. Following this, it was observed that “[t]he next day the wife [of Robert Cornell] coming out at the same door was taken sick, and began to swell from time to time as if she had been with child; by which swelling she came so great in body as she feared she should burst, and to this day is not restored to health.”⁴⁰

Not only did all these extraordinary events and symptoms appear in the approximate time frame, the compressed time sequences also conveyed subtler sequential correlations. The extraordinary symptoms were immediate, cast as “the next day”—not two or three days. A common link was established—the extraordinary events taking place at the door and the wife who came out at “the same door,” and the events “coming out of the same door” and “taken sick” were mentioned in a compressed time and space. Swelling “from time to time” might be normal in illness and infection, but it was extraordinary when compared to pregnancy. The description of the symptoms certainly matched, if not exceeded, those of normal childbirth (“she feared that she should burst”). The gap between that day of the unusual symptoms and the present day was articulated as if the cause and effect was direct—that the aftermath of whatever happened had lingered in some form “to this day.” To summarize, one extraordinary event immediately occurred after a theorized effect; the time frame was so compressed that it was abnormal for the events to occur without a causal relationship with one another.

How much did these events actually occur in the said time sequence, and how much was because of linguistic tricks? This brief quote offers us a clue. Suppose the description and metaphors that anchor time and space are changed.

First, if the initial act was described with fewer details or in a diluted manner—say including many events that also occurred with Robert Cornell’s wife. Consequently, many things would be copresent in the same time and space, obfuscating the correlative relationships.

Second, if the speaker includes the many possible things that happened before and after the wife coming out of the door, the causal relationship could be also be diluted and weakened: many other things may also interject themselves “right before” the person was taken sick.

Third, the symptoms of illness would be less abnormal if they were not implicitly compared with pregnancy to begin with. The speaker could have compared the dramatic physical symptoms with ordinary illnesses like food poisoning or inflammation, as long as the cultural members have enough prior preexisting knowledge about them. If such mental examples do not exist, they would nevertheless be extraordinary, but they might not be linked with a predicated cause of witchcraft by Mother Staunton so directly—by her verbal articulation of suggestive, pre-coded information.

Finally, the description of the effect from an unnamed, original cause suggests an abnormally enduring legacy. The image of the long-lasting harmful consequences would be diluted if many other details are included—perhaps how the subject may recover most of their capability, or that there are fluctuations in health, partly varied by other factors (e.g., food intake, habit of rest, amount of physical activity).

On reflection, the skillful use of language was crucial in constructing correlative relationships by selectively harvesting from a domain of pre-coded information, and in so doing establishing pertinent data to support the idea of witchcraft.

Causal Linkage from Experiments and Miscellaneous Testing Mechanisms

Aside from causation drawn from repeated copresence and sequential correlations, various “tests” or “experiments” are used to validate a certain theory of causal relation, involving witchcraft as a predicated cause.

These “tests” may look more or less rigorous. The basic idea is to first posit a theory (or a set of ideas), and then use empirical observations to demonstrate a match; or it first uses a set of empirical observations and then demonstrates a good fit with some existing, hypothetical ideas. A more rigorous exercise to create causal linkage is, by comparison, counterfactual thinking, or even an experiment-like procedure. In essence, such procedures go a step beyond demonstrating a fit, but they see to demonstrate such a fit after controlling the most relevant conditions except the one to be tested in the hypothesis.

These “tests,” however, may primarily consist of verbal creations as well as actual creations by some physical arrangements. The verbally created kind—also the less rigorous kind—first and foremost has to establish that the observed relations are beyond a threshold of normally expected probability. To use a quick analogy, the approach is akin to demonstrating flipping a coin and ending up with tails nine out of ten times. Secondarily, then, one could verbally emphasize the qualitative fitness between empirical observations and a hypothetical idea—mainly by selectively describing the details of what happens during a test (or perhaps a natural event). Ideally, such empirical details (as pre-coded information) would only fit well into one theorized idea, and not easily into other ones.

In the aforementioned case of Anna Fessler, feeding the cake to the dog was a test. A rough logic of experimental control was implicated: dogs eat most food, and its rejection could thus justifiably be considered extraordinary. The preexisting knowledge that “dogs could sense spirits” qualified as a theoretically justified hypothesis as to why the cake was rejected; a hypothesis that would be subject to validation from further empirical observations or testing. Other pieces of suitable information to add justification for bringing Eva to court should likewise meet two conditions: (1) departure from normal patterns (e.g., a pattern exceeding a threshold of normally expected probability), and (2) a fit with an existing, theoretically grounded hypothesis (even if such a fit might be explicitly articulated).

Many seemingly minute details reported about these tests therefore mattered. Let us consider a few more of the many details offered by the author of The Last Witch of Langenburg. For example, Anna Fessler had also eaten a piece of cake offered by her sister, after eating Eva’s. Then she commented that the cake tasted much better than Eva’s, even though, according to Robisheaux, she “seemed strangely drawn to Eva’s cake. She seemed unable to stop herself from eating the delicacy.”⁴¹ The details here highlight some unusual contradictions (with normally expected probability). Usually, we are drawn to something by either its attractive appearance or smell, but the cake looked strange to the subject’s eyes. The cake chosen by Anna also had a distinct origin. It was not among the six little cakes that were initially visible, but rather appeared somewhat mysteriously—“no one at the time seemed to notice whether she drew it from a secret fold in her apron or not”—and it even looked different; it was yellow and taller and heavier than the others, and Eva claimed it was “baked with butter.” Implied in these descriptions is that not only did Anna fail to notice the strangeness of it all, but also that she was unusually attracted to it (Contradiction #1). And if the cake looked good but the taste was subpar, it would be normal to just have a small bite of the cake, or perhaps a “normal” receptive response. Instead, Anna finished it all with unusual receptiveness and then complained, albeit indirectly, of its quality (Contradiction #2).

These details were not yet fitted with a witchcraft explanation—they were not yet coded with witchcraft-related categories—but would fit very well with it. In the larger idea system, the Devil was predicatively theorized to be a master of illusion; a special characteristic being that he never delivered real benefits through his power but merely offered tricks, harm, and seduction. Without explicit employment of codes, the microscopic details were compiled and organized to fit these beliefs about the Devil. With a well-selected and neatly organized group of pre-coded information, it is akin to a putting a fire next to a haystack. Who actually ignites the fire may not be as important as the fact that the materials are laid out in front of them.

Beyond the Fessler case, we can examine other similar tests that are largely verbal constructions. In the case of the Warboys witches of 1593, a suspect (Mother Samuel) was subjected to a scratching test by a family. In England, it was believed that a person afflicted by witchcraft could find relief by scratching. It was reported that an afflicted girl “lying down on her belly, her face and turned down and her eyes shut” would not touch or scratch several people’s hands when they were put into her hands, but that as soon as Mother Samuel’s hand was offered she scratched it vehemently. Such details would have provided weak potential evidence if more empirical details from more tests had not been incorporated. When the uncle (Mr. Pickering) tried inserting his hand in between Mother Samuel’s and the child’s, the girl avoided scratching his hand, and used one of her fingers to reach out to Mother Samuel’s hand and scratched hers alone. After Mother Samuel’s arrest, a jailor who suspected Mother Samuel’s role in causing his servant’s death and son’s illness decided to scratch her, and his son was reportedly cured shortly after the scratching.⁴²

Contrasting with the verbally constructed tests, some tests are constructed with more rigorous control procedures—at least ostensibly so. If only the accounts are true, then they indeed deserve powerful evidence. Consider a case which occurred in Hungary around 1615. A vineyard owner’s daughter of ten to twelve years old “said that she could bring rain and even hail, and at his request she immediately brought an abundant shower on his vines, without wetting those of his neighbors [italics mine].” There had already been some folklore beliefs that witches could affect the weather methodically. The girl was “tested” to perform an extraordinary act that only could be brought about by a witch, and she indeed did so—and doing so without wetting the neighbors’ vines served as an experimental control.⁴³

In another scandal of weather-changing, which took place in Szeged [Hungary], an eyewitness allegedly reported that thirteen people had ultimately been convicted and then burned after legal judgment. But before they were burned:

they were put to the water ordeal, when they all swam. Then the weight ordeal was tried, when a large and fat woman weighed only 1½ drachms, and her husband, who was not small, 5 drachms, and of the remainder none weighed over 1 pennyweight 3 drachms.⁴⁴

In this case, two tests have been compiled together. There was a long-circulated notion that water was a medium of baptism and, therefore, would reject witches. The suspect who appeared to be “floating” during the swimming test would be a sign of a witch’s identity; the report that “they all swam” meant that they all showed signs of floating.⁴⁵ The weight ordeal was linked to the preexisting idea that witches had little weight. In the Netherlands and Germany, people used “weigh houses” with supposedly standardized scales for weighing suspects (along with non-supernaturally corrupted people and goods). In this particular Hungarian case reported by historian Henry Charles Lea, “the control group” was a husband of normal size—“not small”—who weighed several times more than the suspects, including a “large and fat woman.” Although these tests did not provide definitive proof, they provided empirical signs that fit with—as well as substantiated—the larger idea system, while setting a very high bar of entry for other explanations—mainly explanation by nature or by science.

Backstage Manipulation

The possibility of backstage manipulation was not entirely unknown. In the theater, the audience only sees the front stage, where the performance happens, but to produce the performance also requires elaborate procedures in the backstage.⁴⁶ Contemporary scholars—as well as some observant, skeptical voices back then—have discovered hints of these backstage manipulations. For example, weighing scales may be rigged; there are ways by which the rope could be held—or the addition of a buoyant plank—while a witch was dunked that could affect the outcome of floating or sinking. In scratching tests, people could secretly collaborate with one another; a family member could be sending subtle hints to the supposedly afflicted daughter, even if the daughter was “blindfolded.”

Aside from these occasional discoveries, some tests seemed harder, though not impossible, to fake. The girl who, upon testing, made rain pour on a particular vine without soaking other vines, was one example. Another example was the “bier test” (or cruentation, or “the ordeal of the bleeding corpse”), positing that a person who was recently deceased, with some sensation of sight and hearing remaining, would bleed if the murderer was present and touched him or her. A common procedure was to have a suspect walk up and touch the corpse’s wounds while it was lying on a bier, and then observe any unusual signs that might have appeared. A nursemaid in Nördlingen, Germany around 1590 was coerced by family members to participate in such an ordeal. The dead child’s body reportedly began bleeding as she underwent the test.⁴⁷

Interpretive and Design Manipulations

Suppose that not all such test results were faked. And they probably were not. But several more processes mediate the final appearance of outcomes. The first concerns the interpretation of the test results. The responsibility for interpretation and judgment fell on particular designated epistemic authorities—or “epistemic experts”—to interpret the meaning of the signs. Not all cases of sinking or swimming were objectively clear-cut; neither were what counted as failure to recite a prayer and what counted as an unusual mark on the body. In these cases, the chosen authorities could simply be making judgments arbitrarily. If the error rate was 50%, then there could simply be a 50% chance of “wrongful” interpretation.

Secondly, the whole design of these tests could be manipulated merely to seek to induce—and harvest from—confirmational bias. Because the conclusions of these tests are not cognitively (or legally) binding, the negative result of one test does not necessarily disprove a suspicion. One could run various tests to look for affirmative signs until the desired information is gathered. If each test “naturally” has a 50% chance of affirmative result by mere coincidence, subjecting a person to multiple types of tests, and repeating each type more than once, would inevitably yield some kind of positive sign of witchcraft. For a “prayer test,” people could simply have a person recite a prayer until they found a point of failure, and perhaps certain phrases are simply more easily confused in the mind—such as the famous phrase, “And lead us not into Temptation.” ⁴⁸ Or, in another case, the English witchfinder John Stearne tried to locate the “Devil’s marks” on a suspect, and after failing to do so, found that suspect to be a witch only when applying a swimming test.⁴⁹ Without taking into account disaffirming signs and inconvenient information, tests that produce affirmative signs 50% of the time could be deadly.

Types of Definitive and Uncertain Signs

The evidentiary data set—generated from correlative samples or testing mechanisms—inevitably contains signs with a range of certainty.

Near the outer edge of the range are uncertain signs. Such signs might fit the coding requirement of an idea system, but they retain an ambiguous status—perhaps because these signs also fit an alternative idea, demonstrate only a minimal level of fit, exist as a source of support, or establish themselves based on a marginal departure from normally expected probability. Because the threshold to produce uncertain signs is low, they is always a reserve pool from which social actors could draw. Even though they are not the strongest evidence or proof, these signs excel in their superabundance—and most will not, ever, lose their status of uncertainty.

Definitive signs, in contrast, contain a much more respectable level of certainty. Their fit with cognitive codes is less ambiguous. They have more sources of support, derived from more rigorous testing procedures, or demonstrate a clear-cut departure from normally expected probability.

Although the information itself, by its inherent characteristics, construes the definitiveness of the sign, external sources could help. One notable external source is the amount of credibility lent by epistemic peers who have sincerely looked into the phenomenon. The more epistemic peers affirming an idea means having more mental powers and interpersonal credibility put into double- and triple-checking the evidence and its interpretations. Even though everyone could be wrong in absolute terms, just the interpersonal aspect of cross-checking increases the definitiveness of the sign within an epistemic community. Another notable external source is the introduction of special, qualifying authorities. These people stand several notches above the peer level, often seen as the trustees of various epistemologies developed in various social institutions. Their voices make the meanings of signs more definitive.

Table 2.3 is a rough typology of different kinds of signs (or evidence) that combinatorically uphold an idea system in public discourse.

Table 2.3 Types of definitive and uncertain signs

For the axis of Definitiveness, we see a definitive-to-uncertain continuum. Objectively definitive signs occur when a group of people are in agreement that the signs lend proper evidentiary support to an idea. In contrast, objectively uncertain signs occur when a group of people are in agreement that the signs lend weak, ambiguous, or partial evidentiary support—but nevertheless some support—to an idea.⁵⁰

For the axis of Authority Sources, we may see a public-to-authority continuum. Some signs were meant to only be interpreted by specific people with special epistemic superiority, for instance a church authority, a witchfinder, or one of the cunning folk and people with special abilities (e.g., former witches). If the signs definitively support an idea, only they are in the position to conduct the interpretation and validation. Public signs, in contrast, are ones that “everyone could see” with their own eyes and validate with their own senses and cognition.

These four quadrants of signs have notable differences in how they help to develop compact symbolic structures. Ostensibly, peasants’ “superstitious” methods and beliefs compared unfavorably to the more rational, evidence-based methods and knowledge of the judges. But once these signs enter the mind, and away from major social institutions, the separation would be highly artificial at best. Peasants are “free” to use the preexisting ideas and codes they heard from the elite to construct eclectic rumors. A rumor is justified in being just a rumor, with a low threshold of evidentiary requirement, and a private group of people is entitled to act, feel, and think differently because of rumors.

Elites have the methods, codes, and requirements that they prefer. The demonology corpus shows a rich record of “public” signs being affirmed after they move toward the “authoritative” quadrants—buttressed with more suitable codes and evidence, resulting in compact symbolic structures so artfully formed that they could not be created in the public sphere. Even if elites disaffirm particular judgments, they might only clear away the criminal status on the grounds of insufficient proof—and not necessarily the subjects’ innocence. The rumors and their preliminary evidence might never be provable or disprovable to begin with. Unless the accusers were overwhelmingly proven to have lied, the public could well believe that the difference simply depended upon the threshold being used, and they might create their own interpretations of the contradictions that differed from those made by the authorities. The clearing procedures might even add a few more uncertain signs—such as from witness testimonies—on top of the existing ones. In viewing the quadrants of signs this way, their combinative potential far surpasses their ostensible tension. In combination, they afford people to eclectically create compact symbolic structures, using variegated information and multiple methods.