Essays · by Frater Custos · 12-minute read

The knocks against the wall

In a dozen entries of this atlas the creature is not seen: it is heard. The repertoire of sounds is almost always the same, from Pliny’s Athens to the Val Masino, and it has not changed in eighteen centuries. What changes is who is supposed to be making them — and in 1762, in London, somebody tried to make them answer.

Around the year 100, Pliny the Younger writes to his friend Licinius Sura the letter that opens the documented history of haunted houses in Europe. At Athens there is a house that nobody wants. The philosopher Athenodorus takes it because it is cheap, and on the first night he hears this: «per silentium noctis sonus ferri, et si attenderes acrius, strepitus vinculorum longius primo, deinde e proximo reddebatur» — in the silence of the night a sound of iron, and if one listened more closely, a clanking of chains, at first from far off, then from close by.

It is worth noting the order, because it is the order of almost everything else. First the noise, then the thing. The spectre in chains arrives at the end of the paragraph, when the sound has already entered the room; and the sound is not just any sound, it is iron, which is to say that it is already information — it says that the one arriving is in chains, and therefore that he was badly buried. The account is built upon hearing, and sight serves only to confirm.

Rereading the entries of this atlas one by one, this is not an ancient exception: it is the rule. In a notable number of entries the trait that identifies the creature is not visual. It is a knock, a footstep, a scraping in the wood, a weeping. This article sets those entries in a row, looks at what they really have in common, and follows what happened when somebody decided that that noise could be questioned.

Where the sound comes first

I begin with the cases in which it is the entries themselves that say so, with their own sources.

Thirteen entries, three continents, and a surprisingly narrow repertoire of sounds: knocks, footsteps, scrapings, weeping, chains or metal. One never hears, in the sources, complicated noises. What is heard is what really is heard in a house at night.

Three ways of standing in relation to a noise

Having got this far, the temptation is to treat the thirteen entries as a single thing. They are not, and it is worth separating them at once, because the relation between the creature and the sound is of three different kinds, and confusing them is the quickest way to write something foolish.

First: the sound is the creature. There is no body to see, and nobody claims that there is one. The Klabautermann is explicitly invisible, and its entire activity is knocking on wood; the zashiki-warashi is attested by the noise of a spinning wheel in an empty room; the mazzamurello is defined by the knocks against the wall. Here hearing is not a channel of access to something else: it is all there is.

Second: the noise was already there, and the creature explains it. It is a reversed relation, and the entries say so without mincing words. Of the gigiàt the documented function is the «explanation of anomalous noises at altitude»; of the munaciello, to cover the «nocturnal noises in Neapolitan buildings» — and in Naples, in overcrowded tenements built above an aqueduct dug out of the tuff, nocturnal noises were not in short supply. In these cases the sound is a pre-existing physical fact and the creature is its explanation, not its cause.

Third: the sound is a message. The bean sídhe does nothing: it announces, and it does so to one particular family. The domovoj knocks, and the knock is an omen of death or of epidemic. The Cŵn Annwn bay and somebody in the village will die. Here the noise is neither the creature nor a phenomenon to be explained: it is a signal with a content, and it will be this third kind that produces everything that happens from the eighteenth century onwards.

The name is the noise

There is a check that can be made on the words, and it gives a clear result: in several languages the name of the creature does not describe an appearance, it describes a sound.

The German Poltergeist is compounded of poltern, to make a racket, and Geist: literally «spirit that makes a noise». The Grimms’ Deutsches Wörterbuch defines it as «larmendes, polterndes, spukendes nachtgespenst», a noisy nocturnal ghost, and glosses it with the Latin lemures. In the same entry the headword is attested in Luther — «von den rumpelgeistern oder poltergeistern» — and in Fischart, who in 1591, translating Bodin, lines up three synonyms: «lemures, hauszrumorer oder poltergeister». Hauszrumorer: «house-rumbler». The language, in the sixteenth century, classifies this entity by what it makes one hear, not by what it shows.

The same holds for the Klabautermann. Of the two etymologies in play, the one the Grimms record is klabastern — which the Deutsches Wörterbuch glosses in Low German as «polternd laufen, reiten, fahren», to go about making a racket — and the entry refers explicitly to the kobold des schiffes. The alternative, preferred by modern etymological dictionaries, is Kalfater, the caulker: that is, the man who goes round the ship knocking on the planking to hear where the wood is giving way. Of the two etymologies proposed, one is the noise and the other is the trade that makes that noise for a living.

In Italian the most discussed case is mazzamurello, where the transparent reading — mazza plus murello, that is «little-wall-beater» — is exactly the description of the phenomenon. But a reservation must be stated at once, because the entry states it: that etymology is suspect precisely because it is transparent, and the variants with initial sc- suggest that the first syllable may not be mazza- at all. The Vocabolario Treccani, for its part, records a third etymology again — from «ammazza Mori», on the model of the Spanish matamoros — which the entry considers a folk etymology and implausible. Three proposals, none of them demonstrated. The solid linguistic fact is not the origin of the Italian word: it is that in German, where the documentation exists and is of the sixteenth century, the category is named for the noise.

The noise that cannot be a noise

There is a detail that recurs and that seems to me the most interesting in the whole dossier. The sound of the sources almost always has a property that a real sound does not have.

The limiting case is Welsh. The baying of the Cŵn Annwn, tradition says, is louder when the pack is far off and fades as it draws nearer. It is a declared acoustic impossibility: intensity falls with distance, it does not rise. And it is not an embellishment, it is the function: that property is what allows whoever is listening to know that that baying is not a dog. It is the marker that separates the signal from the background noise of the world.

The entry on the Cŵn Annwn records, alongside the belief, its naturalistic rationalisation as well: the call of wild geese on nocturnal migration. It is a valuable attestation, because in the sources the phenomenon and the explanation of the phenomenon stand together — and because it shows how the impossible trait is formed. A flock passing high up in the dark is heard for a long time and is never seen; the direction it comes from is hard to fix; and what remains in the memory of whoever listened is not a measured curve of intensity, it is the impression that the sound was behaving badly.

The same mechanism can be seen in the most material case of all. The Nachzehrer is a dead man who chews inside the coffin, and the sound is described by the sources with a precise comparison, that of a pig eating. In 1725 Michael Ranft writes a dissertation on those sounds at Leipzig — and it must be said, because it is almost always reported the other way round, that Ranft did not believe in them: his book is a work of dismantling, and it explains the noises by underground gases, the collapse of the coffin and animals shifting the shroud. The noise in the tomb was really there. What was not there was whoever was making it.

❉

1762: the noise that answers

So far the sound is a sign: it is interpreted, as an omen is interpreted. In London, in January 1762, something else happens, and it changes the genre.

In a house in Cock Lane, in the parish of St Sepulchre, scratchings and knocks are heard around the bed of Elizabeth Parsons, who is eleven years old. At the time the sound is compared to that of a cat scratching a chair, and the ghost acquires the nickname of Scratching Fanny. A Methodist preacher, John Moore, then establishes a convention that is the turning point of this whole story: one knock means yes, two knocks mean no.

It is not a detail of local colour. It is the moment at which the noise ceases to be a sign to be interpreted and becomes an answer to a question. From a sign one gets what one manages to get out of it; of a binary code one can ask questions indefinitely, and what comes out of it has the form of a testimony. And indeed, at Cock Lane, what comes out is an accusation: the knocks declare that Fanny Lynes did not die of smallpox but was poisoned with arsenic, and they point to her brother-in-law William Kent.

The reaction is the most modern thing in the eighteenth century. On 1 February 1762 a group of gentlemen, among them Samuel Johnson, meets at the house of the Reverend Aldrich in order to verify. Johnson’s minute is a model of formulation: «it is, therefore, the opinion of the whole assembly, that the child has some art of making or counterfeiting a particular noise, and that there is no agency of any higher cause».

Note that Johnson does not deny the noise. He denies the agent. On 10 July 1762 the trial for conspiracy closes with the sentencing of Richard Parsons, the girl’s father, to two years in prison and three exposures in the pillory; his wife to one year, the servant Mary Frazer to six months’ hard labour. Moore and an associate pay Kent five hundred and eighty-eight pounds in damages.

1848: the noise that speaks

Moore’s code does not die with him. In March 1848, at Hydesville, in the State of New York, knocks are heard at night in the Fox house. The youngest daughter, Maggie, does something simple and decisive: she snaps her fingers and says «do as I do». The knocks imitate her.

It is the step of 1762 again, but this time it holds. From yes and no one moves on to spelling out the alphabet, and from the alphabet to texts of any length. In November 1849 the Fox sisters appear at the Corinthian Hall in Rochester before committees that search them, make them stand on cushions and tie the hems of their dresses to their ankles. Within a few years modern spiritualism is born out of this, with its periodicals: in New York, from 1852 to 1860, Charles Partridge publishes a weekly entitled The Spiritual Telegraph.

That title is a fact, and it is worth holding it still before interpreting it: the first commercial telegraph line in the United States, between Washington and Baltimore, had gone into service on 24 May 1844, four years before Hydesville. My reading — and I declare it mine — is that the code of knocks was able to become a language because in those years there already existed a public and prestigious model of a language made of knocks. It is not an original thesis of mine: it is the core of a historical literature on the media whose most cited title is Jeffrey Sconce, Haunted Media, 2000. I have not read that book, and I attribute nothing precise to it.

The end of the affair is well known and must be reported in full, because it lies entirely in a dated document. On 21 October 1888, at the Academy of Music in New York, Margaret «Maggie» Fox, the daughter, publicly declares — in a statement that had come out that morning in the New York World — that she made the knocks herself, with the joints of her toes, thanks to a control of the muscles of the leg below the knee. A little more than a year later, in November 1889, she retracts, and says that she confessed under pressure.

The atlas is not in a position to establish which of the two declarations is true, and will not try. The fact is that both exist, thirteen months apart, made by the same person.

1979: the noise counted

A century later somebody tried to count. Alan Gauld and Tony Cornell, in Poltergeists (Routledge & Kegan Paul, 1979), subjected to statistical analysis five hundred cases gathered from the literature of four centuries, coding their traits. The most often cited frequencies of that classification are these: movement of small objects in 64 per cent of cases, knocks in 48, movement of large objects in 36, apparitions between 26 and 29, voices and groans in 26.

It must be said precisely what these numbers are and what they are not. They are not a statistic of events: they are a statistic of accounts, gathered from a literature that has its own criteria of selection, and the work comes from a field — psychical research — whose premises this atlas does not endorse. Taken for what it is, though, the result is useful and points in the same direction as the entries: in the reports the acoustic channel and the movement of small objects come well before the figure that is seen. Those who write about hauntings, over four centuries, describe a noise almost twice as often as they describe a ghost.

On the folklore side the classification has existed for longer. In Stith Thompson’s Motif-Index of Folk-Literature the number is F473, Poltergeist, defined as «invisible spirit (sometimes identified as ghost or witch) responsible for all sorts of mischief in or about a house». And it is instructive to see where the noise ends up: in F473.5, «poltergeist makes noises», one of the six subtypes, after the throwing of objects, objects behaving contrary to their nature, and the ill-treatment of people and animals. The German language puts the noise in the name; the motif index puts it in fifth place. They are two different classifications because they answer two different questions — what is heard, and what happens — and the second, cataloguing by action, loses exactly the trait that the first considered to be the defining one.

My reading, declared as such

What follows is not in the sources: it is mine, and let it be taken for what it is worth.

It seems to me that the acoustic channel is the most stable of the channels of the domestic supernatural, for a reason that has nothing mysterious about it. A house at night is an environment that produces sounds and does not produce images. Wood contracting when the temperature drops, mice in the loft, wind in a chimney, the settling of a beam, water in a pipe: these are real, recurrent events, and almost always invisible. In the dark the material the ear receives is abundant, and the material the eye receives is nil.

And a knock has a property that an image does not have: it is unequivocal as an event and completely ambiguous as a content. Nobody doubts having heard a knock; nobody can tell by himself what produced it. It is this combination — certainty of the occurrence, total emptiness of the attribution — that makes nocturnal noise the most readily available raw material there is for a story. The munaciello, the gigiàt and Ranft all three perform the same operation on that emptiness, and fill it with three different things: a little monk, a mountain beast, underground gases.

The step of 1762, then, seems to me the real watershed of the whole affair, and not for the reason for which it is usually cited. Moore’s code does not serve to prove that the noise comes from somebody: it serves to make it say something. Once it has been established that one knock means yes and two mean no, whatever the physical cause of the knocks may be, the apparatus produces sentences — and the sentences, at Cock Lane, accused of murder a man who had killed nobody. It is the reason why this atlas looks more attentively at the devices than at the creatures.

What this does not prove

It does not prove that there exists a single universal belief in the knocking spirit. The thirteen entries above have separate histories, documentation of very different quality — it ranges from the A of the domovoj and the bean sídhe to the C of the gigiàt — and at least three different relations with sound. To put them into a single story would be the error that these pages exist in order not to make.

Nor does it prove the opposite, that is, that it is always a matter of beams and mice. Of the five hundred reports of Gauld and Cornell I know nothing at first hand, and of Cock Lane we know that a committee of 1762 attributed the knocks to an eleven-year-old girl, not that we know how she made them.

What the material supports is narrower, and it seems to me enough. In a substantial part of domestic folklore the creature is a hypothesis about the origin of a noise; the noise, in the sources, has a narrow and recurrent repertoire that closely resembles that of an empty house in the dark; and the trait that makes it recognisable as a sign — the baying that fades as it comes closer — is usually a trait that a real sound cannot have.

One thing remains that can be said without reservation. From Athens to Hydesville, for eighteen centuries, when somebody heard at night a knock that he could not explain, he almost always had a name ready to hand to give to that knock — and from 1762 onwards, a way of asking it questions as well. The second was a progress only in appearance: the name left the noise a noise, the code made it speak.