Magic in the Ancient Texts

Introduction

What counts as magic, and what this library can tell us

The ancient words for magic were mostly names for other people's rites, so this report follows each culture's own vocabulary through 1,343 verified passages in one digital library, and is plain about what that library cannot show.

The words were accusations

The English word "magic" comes from a people. Herodotus, writing around 430 BCE, knows the Magi as one of the tribes of the Medes, some of whom read dreams at the Median court. King Astyages takes a frightening dream to them, and when their later judgement that its danger had passed proves wrong he punishes them as experts who got it wrong (Hdt. 1.107.1). A dialogue transmitted under Plato's name, of disputed authorship, glosses μαγεία (mageia), the teaching of Zoroaster, as "the worship of the gods", a subject for Persian princes ([Pl.] Alc. 1, 122a). Around 100 CE Dio Chrysostom complains that Greek ignorance of the word has turned the Persians' word for men skilled in serving the divine into a word for γόητες (goētes), sorcerers (Dio Chrys. Or. 36.41).

That shift happened early. In Sophocles' Oedipus the King (probably c. 429 BCE) Oedipus abuses the prophet Teiresias as a scheming μάγος and a begging priest with eyes only for profit, and the irony of the scene is that the "quack" is telling the truth (Soph. OT 387–389). A Hippocratic doctor of the same period blames the notion of a "sacred disease" on "magoi, purifiers, begging priests and charlatans" who pretend to special piety (Hippoc. Morb. Sacr. 1). The sophist Gorgias sets the two Greek words side by side as "two arts of sorcery and magic", which he calls "errors of the soul and deceptions of opinion" (Gorg. Hel. 10; see chapter IV).

The native Greek word, γοητεία (goēteia), belongs to the γόης, a figure many modern scholars connect with γόος, ritual lament for the dead. That etymology is an inference. In the texts the goēs can be a foreign charmer, as when Pentheus in the Bacchae calls the disguised Dionysus a "goēs and enchanter" who offers secret rites to women (Eur. Bacch. 233–238). Plato's Laws separates poison, which harms bodies by bodily means, from a second kind of φαρμακεία worked by trickery, incantations and "so-called bindings". He describes this second kind as working on the belief of attacker and victim alike, and declines to say whether it really works (Pl. Leg. 933a–b).

Latin took its words from law. Pliny the Elder quotes two clauses of the Twelve Tables, Rome's earliest code, against "whoever has charmed away crops" and "whoever has chanted an evil charm" (Pliny NH 28.17–18). Magic would later be prosecuted under Sulla's law of 81 BCE on assassins and venefici, which as Cicero quotes it orders an inquiry into poison (Cic. Clu. 148). Veneficium, from venenum, "drug" or "poison", could cover both, and Quintilian uses the question whether the spells of magi count as veneficium as a textbook dispute over definition (Quint. Inst. 7.3.7). The Hebrew law of Exodus 22.18 condemns a feminine mĕkaššēpâ, which the library's English renders "You shall not allow a sorceress to live" (Exod. 22.18). The library's Greek translation reads Φαρμακοὺς οὐ περιποιήσετε, "you shall not keep drug-makers alive", turning a feminine singular into a masculine plural and the crime into a matter of drugs (Exod. 22.18 LXX).

The clearest witness to how the label worked is a man tried under it. Apuleius, philosopher and novelist, was accused around 158 CE of winning a rich widow by spells, and his defence speech survives. He offers the court two definitions (see chapter VI).

Nam si, quod ego apud plurimos lego, Persarum lingua magus est qui nostra sacerdos, quod tandem est crimen, sacerdotem esse et rite nosse atque scire atque callere leges cerimoniarum, fas sacrorum, ius religionum

For if, as I read in a great many authors, a magus in the Persian tongue is what we call a priest, what crime is it, I ask, to be a priest, and to know properly, to understand and to be expert in the rules of ceremonies, the proper forms of rites and the law of religion?

The second, "vulgar" definition makes the magus a man who talks with the gods and can do anything he likes by the force of his chants. Apuleius turns it on his accusers, who would never dare to prosecute such a man if they believed it (Apul. Apol. 26.6–9). Four and a half centuries later Isidore of Seville closes the gap. His magi "are commonly called malefici because of the enormity of their crimes", and they kill by the force of a spell without any poison (Isid. Etym. 8.9.9). The Persian priest has become the evil-doer.

Words that were not accusations

Other cultures in the library had words for powerful ritual that carried no blame. In the Pyramid Texts, carved inside royal tombs from about 2350 BCE, Egyptian ḥkꜣ (heka) is a force that gods possess and that the dead king can eat.

Wnꞽś pꞽ wnm ḥkꜣ(.w) ⸗śn ꞽ:ꜥm ꜣḫ(.w) ⸗śn … ꞽw ḥkꜣ(.w) ⸗śn m ẖ.t ⸗f

Unas is the one who eats their heka and swallows their spirit-power … their heka is in his belly.

A spell for Pepi I simply says "Pepi is heka", or "is the god Heka" (PT 472 §924). In Mesopotamia the line ran between two figures. The āšipu, or exorcist, was a temple scholar whose incantations have him declare "I am Ea's man … the hand of Ea is in my hand" (GAAL 2 pl. 24–25, 15–22). Against him stood the witch, in Akkadian usually kaššāptu and here elēnītu, who in a commentary on the anti-witchcraft series Maqlû binds her victim while her kind pluck his hair and gather the dust of his footprints (KAR 94 7'–11'). Snorri Sturluson, writing in Iceland around 1230, calls seiðr the art "which carries the greatest power" and says Óðinn practised it, though it brought such ergi, unmanliness, that men shunned it and it passed to priestesses (Yng. 7). A German worm charm in a twelfth-century manuscript commands the worm by the life of Christ, and a Latin note calls it a medicina that loses its power for people once it is used on a beast of burden (Contra vermem edentem). An Old French epic of the late thirteenth century puts nigromance last among the arts a clerk learns, the most hated and the devil's own, and then lets the barons wish the clerk were on their side (Anseïs de Metz U 33r).

The report uses these words wherever it can and explains each when it first appears. "Magic" here is a finding-term, a net for collecting passages, and makes no claim that everything in the net is one thing.

How scholars have framed it

Modern theories began with the Victorian anthropologists. Edward Tylor and then James Frazer, in The Golden Bough (1890, much enlarged in later editions), treated magic as mistaken science, built on two laws Frazer called similarity (like produces like) and contagion (things once in contact go on acting on each other), and placed it at the foot of a ladder that climbed through religion to science. Marcel Mauss and Henri Hubert (1902–3) moved the question from logic to society. Religion for them was public and approved, magic private, secret and unofficial; Émile Durkheim put it as "there is no church of magic". Bronisław Malinowski stressed use, with Trobriand gardeners and sailors turning to spells where skill ran out. E. E. Evans-Pritchard showed in 1937 that Azande witchcraft belief formed a coherent system, answering the question "why me, why now?" that physical causes leave open.

Each frame catches something real in the ancient texts, and each has been criticised. Frazer's ladder does not fit cultures where the most "magical" acts belong to kings and gods, and the line between public and private fails for Egyptian heka and the Assyrian court's exorcists. Since the 1990s many historians of antiquity, among them Fritz Graf, Matthew Dickie and Kimberly Stratton, have preferred to ask who called whom a magician, and why. J. Z. Smith proposed dropping the term from scholarship. Bernd-Christian Otto and Michael Stausberg keep it as a loose heuristic category while tracing the history of the word itself. This report takes a middle course. It records each culture's own categories first (the "emic" view) and uses the modern word only to set them side by side (the "etic" view).

Why the line keeps moving

The passages draw the boundary against medicine, religion and science, and draw it differently each time. Pliny thought magic began inside medicine, "creeping in under the guise of healing as a loftier and holier medicine", then absorbed religion and astrology to become "the most fraudulent of the arts" (Pliny NH 30.1–2). Augustine, in the 390s, offered a test that sounds almost experimental.

Aliud est enim dicere, Tritam istam herbam si biheris, venter non dolebit; et aliud est dicere, Istam herbam collo si suspenderis, venter non dolebit.

It is one thing to say, "If you crush this herb and drink it, your stomach will not hurt", and another to say, "If you hang this herb round your neck, your stomach will not hurt."

The first acts by nature and is medicine. The second acts through a sign agreed with demons and is superstition, like the amulets and tied-on charms that, Augustine notes, "the discipline of physicians also condemns" (Aug. Doctr. chr. 2.20.30). Plotinus drew the line in the opposite place. For him "the true magic" is the love and strife that run through the cosmos, itself "the first sorcerer and drug-maker", so a human practitioner only borrows a power already in nature (Plot. Enn. 4.4.40).

Science moves the line as well. Pliny praises the astronomers whose predictions freed people from blaming eclipses on witches' drugs (Pliny NH 2.53–55), and complains that many of the common people still think eclipses are forced by drugs and herbs, "this one science of women" (Pliny NH 25.9–11). Hippolytus, in the early third century, explains a sorcerer's oracle as chemistry and stagecraft, down to the oak-gall fumes that make invisible ink appear (Hippol. Ref. 4.28). Philostratus explains why belief survives failure. When a love affair succeeds the art is praised as all-powerful, and when it fails the blame falls on some step left out (Philostr. VA 7.39). Chapters XII and XIII take up these questions.

What the library holds

Classica Codex is a reading library of texts in their original languages, with translations where they exist and a catalogue of Greek and Roman objects. It was not built to study magic, so its coverage is uneven. The table lists the fifteen collections surveyed, the rough span of the passages found, and the number of verified entries each contributed.

The evidence is lopsided in one important way. Most of these texts were written about magic by people who did not practise it, usually educated men. In the collections surveyed, the practitioners' own words survive in only a few places, chiefly the Pyramid Texts, the Mesopotamian exorcists' commentaries, the German charms and a handful of recipes in Cato and the medical writers; the library has since gained many more (see below). In tone the survey is more balanced. Of all the entries, 303 tell of magic as something that happens, 151 condemn it, 142 use it in fiction and 82 expose a trick; the rest instruct, mock or simply mention it.

Some texts are stored in ways a reader should know about. The Septuagint and some patristic and Greek editions are raw scans with recognition errors, and quotations keep those errors as stored. Several editions number passages their own way, so the library's reference can differ from the conventional one. Some passages carry a flag marking them as spurious that the researchers found to be an artefact of the encoding, and the report does not treat those passages as doubtful on that ground. A few volumes filed under ancient names hold early-modern Latin, and these were set aside.

How the research was done

The work had four stages. AI agents did the research and the writing, and scripts did the checking (How this report was made sets out who did what, and what the checks can and cannot show). First, an AI agent surveyed each collection, sweeping it for that culture's own vocabulary (heka, kišpū, φάρμακον, carmen, seiðr, segen and many more), checking leads from earlier scholarship and opening each passage before recording it, with its edition, reference, original excerpt, translation, claimed effect, practitioner and the author's stance. Second, a script checked every entry against the library. The edition and reference had to exist and the quoted excerpt had to be found there. The scripts found 1,277 excerpts at their references (21 of them only after a second script undid the first run's wrong corrections) and 3 beginning at a neighbouring passage, whose references were corrected; an AI agent confirmed 6 more where the transcription defeated the script, and for the 57 museum objects the script confirmed only that the catalogue entry exists. Third, the surveyors flagged 351 claims that could be tested against modern knowledge. Seven panels of AI agents took the 339 from fourteen of the surveys (the Renaissance English survey's 12 were never sent to them), merged them into 260 and returned 283 verdicts. In each panel one agent researched the evidence and drafted the verdicts, two others challenged the drafts they thought too generous or too dismissive (95 of the 260 were challenged), and a fourth answered the challenges and gave each verdict, from "real physical effect", "real effect by a different mechanism" and "knowledge presented as magic" through "psychological" and "trick or fraud" to "implausible", with "insufficient evidence" where the panel could not decide. Fourth, AI writing agents drafted each chapter from the verified entries and from passages they opened in the library, and AI fact-checkers then read each chapter.

Dates are approximate throughout and many are disputed, among them saga composition dates, the age of a charm copied into a later manuscript and the authorship of works such as the Alcibiades. The text says so where it matters. Translations are the AI authors' own and have not been checked by a specialist.

How to read the citations

Every reference printed in red names a passage in the library. Its label gives the conventional citation, such as Hdt. 1.107.1 or PT 274 §403. Clicking it copies the library's own identifier for the passage, which may be numbered differently, so that it can be opened in Classica Codex. Original-language quotations are copied exactly as the library stores them, scribal spellings and scanning errors included, and are kept short. Objects are cited by museum number. Where a statement rests on modern scholarship and not on a passage, the text says so.

The chapters that follow go culture by culture, from Egypt to Renaissance England. Chapter XI compares them, chapter XII asks which claims could have worked, chapter XIII turns to the tricks and the people who exposed them, chapter XIV counts the vocabulary across time, and the Conclusion draws the threads together.

Chapter I

Egypt: the power that gods and kings could eat

In the Pyramid Texts heka is a force that gods carry in their bellies, kings swallow and priests speak into being, and it was only in Greek, Jewish, Roman and Christian eyes that Egypt became the homeland of magicians.

A king at a feast of gods

Around 2350 BCE the burial chamber of King Unas at Saqqara was covered from floor to ceiling with columns of hieroglyphs. They are the oldest large body of religious writing to survive from Egypt, and modern scholars call them the Pyramid Texts. Most of the spells feed, purify, protect and raise the dead king. One of them, now numbered 273–274 and known as the Cannibal Hymn, does something stranger. It shows Unas hunting the gods across the sky, cooking them and eating them.

Wnꞽś pꞽ kꜣ p.t nhṭ m ꞽb ⸗f ꜥnḫ m ḫpr(.w) n(.ꞽ) nčr nb wnm(.w) wsm.w ⸗śn ꞽw.w mḥ ẖ.t ⸗śn m ḥkꜣ.w m Ꞽw-nśꞽśꞽ

Unas is the bull of the sky, raging in his heart, who lives on the being of every god and eats their entrails, when they come with their bellies full of heka from the Island of Flame.

The hymn goes on to say plainly why the king eats. A few lines later he is "the one who eats their heka and swallows their akh-power", and "their heka is in his belly" (PT 274 §403, §411). What Greeks and Romans would later call Egyptian magic appears here, at the start of the written record, as a substance inside divine bodies that a king can take by force. Nothing in the text treats this as wrong.

This chapter follows heka through the Pyramid Texts and the later Egyptian books in the library, and then follows the much longer story, from Homer to a medieval Swedish poet, in which outsiders turned the land of heka into the land of sorcerers.

The words for it

The Egyptian word is ḥkꜣ, conventionally pronounced heka. (The library's transliteration writes č̣ where Egyptologists usually write ḏ, and ꞽ for j.) It occurs in 23 lines across the pyramids in the library, and every one describes a power belonging to someone with a right to it.

Gods have heka in measurable amounts. The sky goddess Nut is told that she has "taken Horus, his great heka" and "taken Seth, his great heka" (PT 443 §823), though the two are rivals in myth. The red crown and the cobra on the king's brow are addressed as a goddess called Wr.t-ḥkꜣ.w, "Great of Magic", and asked to give Unas her power to strike terror (PT 221 §196). As the dead king rises, "his ba has brought him, his heka has equipped him" (PT 245 §250), and another sets "his heka at his feet" (PT 306 §477). In the pyramid of Pepi I the identification is complete.

č̣(ṭ) mṭw ꜣwꞽ p.t śṭꜣ tꜣ ṭp-ꜥw(.wꞽ) Ppy pn / Ppy pw ḥkꜣ.w / Ppy pw ẖr(.ꞽ) ḥkꜣ

Words to be spoken. The sky stretches, the earth trembles before this Pepi. Pepi is heka. Pepi is the one who holds heka.

"Pepi is heka" can also be read "Pepi is Heka", the god of that name known from other Egyptian sources. The library's Egyptian word list now also has Heka as a god's name, and a New Kingdom healing book calls on him beside Horus, Re, Shu, Geb, Osiris and Nun (London Medical Papyrus, TLA 1.393), but the line itself cannot settle the question. Either way the king is not asking for magic. He is it.

Several other words do the work that "magic" does in English. ꜣḫ (akh) means effective, radiant, a spirit transfigured into its full power; its causative ś:ꜣḫ means "to make someone akh". An ointment is told to "make him akh" (PT 77 §52). The rubric č̣(ṭ) mṭw (djed medu), "words to be spoken", opens most spells and appears in 412 lines. A spell is rʾ, "mouth" or "utterance", and a name, rn, carries much of the weight. Egyptologists debate whether "magic" translates heka at all, since nothing here sets it against religion, medicine or kingship.

Who wielded heka

In the Pyramid Texts the practitioners are gods and kings. The king is the one whose heka matters, and he acquires it by eating, by inheritance and by identity. Gods hold it too, and so do the goddesses who protect him: the crown goddess, Nut, and Mafdet, a cat-like or mongoose-like huntress whose claws kill snakes.

Someone had to speak the words, and the texts never name him. Many spells have a first-person speaker: "Osiris Unas, I have given you the Eye of Horus" (PT 25 §18), and in the Opening of the Mouth, "I am Horus … I am your beloved son" (PT 20 §11, Pepi II). Egyptologists identify this voice, on evidence from outside the pyramids, with the lector priest who read the sacred book in temple and funeral rites. The title is not in the library's Pyramid Texts, but it is in temple papyri of the same century (see 'Lector priests and magicians' below). He spoke as a god (Horus) to a god (the dead king as Osiris).

The library adds one detail about women. Queen Neith, a wife of Pepi II, received the same offering spells as the kings. In her copies she is addressed as "Osiris Neith", but in the lines I opened the word for "you" keeps its masculine form: "Osiris Neith, take for yourself the Eye of Horus" (PT 58, Neith). The spells were made for kings and passed on with their grammar unchanged.

The rite could pass to others in another way too. A spell in Pepi I's pyramid, repeated in Pepi II's, promises its reward to anyone who knows it.

rḫ św rr(ꞽ) rʾ pn n(.ꞽ) Rꜥw ꞽr ⸗f śn ḥkꜣ(.w) ꞽpn n.(ꞽ)w Ḥr.w-ꜣḫ.tꞽ wnn ⸗f rr(ꞽ) m rḫ.ꞽ n(.ꞽ) Rꜥw wnn ⸗f m śmr n(.ꞽ) Ḥr.w-ꜣḫ.tꞽ

Whoever knows this spell of Ra and performs these heka of Horakhty will be one whom Ra knows, will be a companion of Horakhty.

The next line restates the promise in the king's own voice (PT 456 §856). Here heka is plural and countable, "these heka", a set of acts that can be learned and done. The clause anticipates the "whoever knows this spell" rubrics of the Book of the Dead (see 'Rubrics, finds and foreign tongues' below). In the Old Kingdom the knower in practice was still the king.

What they did with it

Snakes

No danger fills the Pyramid Texts like snakes. A block of spells (PT 226–244) runs around Unas's sarcophagus chamber, and another series guards the antechamber. They are short, direct and unpleasant for the snake.

č̣(ṭ) mṭw šn(.w) nꜥw ꞽn nꜥw šn(.w) bḥs ḫꜣb.w pr m ḥsp … ḥfꜣ.w pnꜥ mꜣ čw Rꜥw

Words to be spoken. The serpent is coiled round by the serpent; the toothless calf that came out of the garden is coiled round … Snake, turn on your back! Ra sees you.

There is no prayer and no offering. The command does the work, with the sun god as witness. Other spells chain powers together. "Your mouth is closed by the šmś.t; the mouth of the šmś.t is closed by Mafdet" (PT 230 §230); the meaning of šmś.t is unknown. In another, the sun's cobra rears against a snake that rises from the earth under the king's fingers, Mafdet's claw takes its head, and the spell ends with a stock cry, "Hiu-snake, lie down!" (PT 298 §442). In Teti's pyramid Horus shuts the snake's mouth with his sandal (PT 388 §681), and in Pepi I's Mafdet's claw blinds it (PT 501C, Pepi I).

Some of Unas's snake spells cannot be read as Egyptian at all.

č̣(ṭ) mṭw kbb hꞽ tꞽtꞽ bꞽtꞽ šś sꜣ ḥꞽfg.t rn ⸗k pw

Words to be spoken. Kbb hi titi biti šś sꜣ ḥifget: that is your name.

Another begins "m(ꞽ)tꞽ m(ꞽ)tꞽ mꞽtꞽ mꞽtꞽ" (PT 232 §236). The Semitist Richard Steiner has argued that several of these lines are early Canaanite snake charms written down in hieroglyphs, which would make them among the oldest Semitic texts known. Others read them as sound-play meant to seem foreign. The reading is disputed. On either view the oldest Egyptian spells already use words that are opaque on purpose, and the "barbarous names" Greek philosophers defended two and a half thousand years later had a long past.

Threatening the gods

The most surprising acts in the Pyramid Texts, to a modern reader, are the threats. The dead king needs a place among the gods, and if he is refused he promises cosmic damage.

ꞽr tm ⸗k ꞽr(.w) ś.t n Wnꞽś ꞽr.kꜣ Wnꞽś fꜣ.t m ꞽt(ꞽ) ⸗f Gbb / tꜣ nꞽ mṭw.n ⸗f / Gbb nꞽ wꜣ.n ⸗f

If you do not make a place for Unas, Unas will lay a curse on his father Geb. The earth will not speak. Geb will not be able to …

The word rendered "curse", fꜣ.t, is uncertain; it may mean a lifting or a carrying-off. The threat is addressed to the Lord of the Horizon, and Geb is the earth itself. The next spell goes further: if the god does not leave his seat and lay his dignity on the ground, Unas will come with the face of the Great One, set the fire of his eye round the gods, and break the arms of Shu that hold up the sky (PT 255 §297–299). Teti's pyramid carries the same threats. For a dead king, forcing a god was a proper act, carved beside hymns of praise. It would scandalise a Greek philosopher in the third century CE.

Names, known and hidden

Names give power over their bearers, so the king's must be safe. Unas is "the lord of …, whose name his mother does not know", and he "judges together with him whose name is hidden" (PT 273 §394, §399). The other side of this is knowledge. "Unas knows him, knows his name: Neh is his name, Neh-lord-of-the-year is his name" (PT 301 §449). A recurring formula, "in this your name of X", ties a god's name by a pun to the act in hand. The library's Unas edition opens against "all who hate Unas, who speak evil against his name", whom Osiris is told to seize and "not let go", the words to be said four times (PT 23 §16). To slander the name was to attack the person.

Substances, gestures and tools

Much of the Unas text (PT 23–171) is an offering ritual in which speech and matter work together. Each offering is declared to be the Eye of Horus, the god's wounded and restored eye, with the stage direction beside the words: "Osiris Unas, I have given you the Eye of Horus; your face is equipped with it … Words to be spoken four times. Incense, fire" (PT 25 §18). Purification with natron, the salt of embalming, rests on a pun between two words for natron and specifies the dose: "southern natron, five pellets, from Nekheb" (PT 34 §26). Teti's copy of an offering line that in Unas has no rubric adds "words to be spoken four times … lifting, four times" (PT 155 §93, Teti). These are scripts, with gestures, counts and materials.

The rite of Opening the Mouth, which restored the senses of the dead, used a real tool.

[hꜣ] [Nfr]-[kꜣ]-[Rꜥw] [wp.n] [(⸗ꞽ)] [<n>] [⸗k] [rʾ] [⸗k] m nwꜣ Wp-wꜣ.wt mśḫt.w bꞽꜣ wp rʾ n(.ꞽ) nčr(.w)

O Neferkare, I have opened your mouth for you with the adze of Wepwawet, the "foreleg" adze of bia-metal that opens the mouth of the gods.

Most of this line is a modern restoration, marked by the brackets, and only the end is read from the wall. bꞽꜣ is the metal word Egyptologists associate with meteoric iron, so a rare substance from the sky sat at the centre of the rite. In these texts natron, iron and incense work because they are named, counted and spoken over.

How it was meant to work

The Pyramid Texts contain no theory of why a spell works, only spells. Read as a whole, though, they rely on a few consistent means, which modern scholars have named.

• Speech that acts. "Words to be spoken" introduces commands in the imperative. Saying "Snake, turn on your back" is the protective act.
• Repetition. "Four times" follows many lines, often explained by modern scholars as one for each quarter of the world.
• Identity. The officiant speaks as Horus; the king is a bull of the sky, Osiris, heka itself. Becoming a god is the means of acting as one.
• Mythic equation. Bread, water and incense are the Eye of Horus because the speaker says so.
• Puns and names. A god's name is bound by wordplay to the act in hand, and knowing a name gives access while hiding one gives protection.
• Taking in power. Heka can be eaten and kept in the belly.
• Coercion. A god who refuses can be threatened with the collapse of earth and sky.
• Chains of agents. One power shuts a mouth, a stronger one shuts that power's mouth.
• Materials. Natron in counted pellets, ointment addressed as a person, an adze of sky-metal.

Later Egyptian books state fragments of a theory (see 'Rubrics, finds and foreign tongues' above). The fuller theory came much later, in Greek, from Egyptians and admirers of Egypt under Roman rule, and it reached for the features the old spells show. A treatise of the Hermetic corpus, presented as advice to a king, insists that its discourse must stay untranslated.

καὶ ἡ τῶν Αἰγυπτίων ὀνομάτων ⎡δύναμις⎤ ἐν ἑαυτῇ ἔχει τὴν ἐνέργειαν τῶν λεγομένων. … ἡμεῖς δὲ οὐ λόγοις χρώμεθα, ἀλλὰ φωναῖς με[γι]σταῖς [τῶν] ἔργων.

And the power of Egyptian names holds within itself the working of the things said. … We do not use mere words. We use sounds full of deeds.

Iamblichus, a Syrian Neoplatonist who around 300 CE wrote in the person of an Egyptian priest, made the same argument: names lose their force in translation, because "there are idioms in each nation that cannot be signified in another nation's speech" (Iambl. Myst. 7.4–5). The Christian Origen agreed, against Aristotelians and Epicureans, that magic is coherent, since names spoken "in the Egyptian language" bind particular daimones (Orig. C. Cels. 1.24–25). The Hermetic Asclepius added that the Egyptians "discovered the art of making gods" by installing daimones in sacred statues (Asclepius 37). None of them cites the Pyramid Texts, yet the principle they defended was already at work in Unas's chamber.

Where the line fell, and who was punished

Inside the Pyramid Texts there is no line between religion and magic. The boundary that matters runs between the king and his enemies. Dangerous power belongs to snakes that bite, to those who "speak evil against his name", to gods who will not make room. The punishment the texts imagine is cosmic. Osiris seizes the slanderers and does not let go (PT 23 §16), and Mafdet's claw takes the snake's head. The library holds no Egyptian trial concerning heka (an Old Kingdom court judgement in it decides a dispute over an estate: pBerlin 9010, TLA 1.3–8), so it cannot show how a pharaonic court treated a private person who turned heka against another. Later Egyptian books draw a line of their own, between one's own heka and other people's (see 'Heka turned against people' below). The foreigners drew it in different places.

Herodotus, who travelled in Egypt in the mid-fifth century BCE, calls nothing Egyptian μαγεία or γοητεία. Divination there, he says, "belongs to no human being, but to certain of the gods" (Hdt. 2.83); portents are written down with their outcomes so that the same sign can be read the same way again (Hdt. 2.82); and the only sorcerers in his Egyptian book live far beyond the Libyan desert (Hdt. 2.33). For Herodotus magic lay further out than Egypt.

The Hebrew Bible draws the line by contest. The library has Exodus in the Greek of the Septuagint, translated in Ptolemaic Egypt around 250 BCE, and in English from the Hebrew. When Aaron's staff becomes a serpent, Pharaoh calls his σοφισταί (experts), φαρμακοί (drug-workers, sorcerers) and ἐπαοιδοί (singers of spells) (Exod. 7.11 (LXX)). The story concedes their power: they match Aaron's wonders of blood and frogs, and first fail at the gnats, when the enchanters themselves tell Pharaoh, "this is the finger of God" (Exod. 8.19 (LXX)). Chapter III follows the contest in full. The boundary is one of source and degree. At home the same Greek word becomes a legal category, "you shall not keep φαρμακοί alive" (Exod. 22.18). Pharaoh's dream experts in Genesis, "magicians" in the English from the Hebrew, are simply ἐξηγηταί, "interpreters", in the Greek (Gen. 41.8). The Hebrew word, ḥarṭummîm (Gen 41.8), which the Hebrew of Exodus sets beside mᵊkaššᵊpîm, sorcerers (Exod 7.11), is generally derived from the Egyptian title of the chief lector priest, the office that spoke the Pyramid Texts and that Westcar gives to its court wonder-workers (see 'Lector priests and magicians' above).

Josephus, retelling Exodus in Rome around 94 CE, sharpened the line. His Moses works "not by sorcery and the deception of true opinion, but by the providence and power of God" (Joseph. AJ 2.286), and the Egyptians' staffs only "seemed to be serpents" (AJ 2.287).

The clearest statement of the boundary in the whole library is put in the mouth of an Egyptian priest, Calasiris, in Heliodorus' novel Aethiopica (3rd or 4th century CE, date disputed). He divides Egyptian wisdom in two. The common kind "crawls along the ground", waits on phantoms and the bodies of the dead, and relies on herbs and incantations; the true kind looks to heaven and the motions of the stars (Heliod. Aeth. 3.16; see chapter V).

Later in the novel a woman of Bessa raises her dead son with a pit, fire and libations, and soon dies on a spear-shaft (Heliod. Aeth. 6.14–15). The low kind is female, local and fatal; the high kind is male, priestly and astral. Galen drew a similar line for medicine, accepting plants but not the "silly Egyptian sorceries" recited while gathering them (Gal. SMT 6 prooem.). At Rome the line was a matter for the courts. Apuleius, tried for magic around 158 CE, mocked accusers who took Greek zoological terms for spells recited "in the Egyptian or Babylonian manner" (Apul. Apol. 38).

The land of magicians

The outside view of Egypt began with admiration. In the Odyssey, Helen calms the grieving men at Sparta with a drug she was given in Egypt.

τοῖα Διὸς θυγάτηρ ἔχε φάρμακα μητιόεντα, / ἐσθλά, τά οἱ Πολύδαμνα πόρεν, Θῶνος παράκοιτις / Αἰγυπτίη, τῇ πλεῖστα φέρει ζείδωρος ἄρουρα / φάρμακα, πολλὰ μὲν ἐσθλὰ μεμιγμένα πολλὰ δὲ λυγρά·

Such clever drugs the daughter of Zeus possessed, good ones, given her by Polydamna, wife of Thon, an Egyptian woman. There the grain-giving earth bears the most drugs of all, many good in a mixture and many deadly.

Homer's Egypt is a land of drugs and physicians, where the power lies in the substance and the expert is a woman. Six centuries later Diodorus heard that the women of Thebes still used a drug against anger and grief (Diod. 1.97.7), and that Isis heals in dreams the sick whom doctors have given up (Diod. 1.25.5). Plato could already call Proteus "the Egyptian sophist" who bewitches, though only as a joke (Pl. Euthd. 288b).

The change came in the Hellenistic and Roman centuries. Strabo's Decaeneus learned to read signs of the future in Egypt and became the power behind a king of the Getae, regarded in the end as a god (Strabo 7.3.11). A stay in Egypt had become a holy man's credential. The New Testament named Pharaoh's magicians Jannes and Jambres (2 Tim. 3.8). Pliny the Elder made Moses and Jannes, enemies in Exodus, co-founders of a single school of magic (Pliny NH 30.11), and the pagan Numenius took the Egyptian side, calling the pair sacred scribes "judged second to none in magic" who could undo the worst of Moses' plagues (Numenius fr. 9).

Some of what Greek writers reported matches the Pyramid Texts closely, though none of them could have read the pyramids. Plutarch describes what Egyptian priests did in drought or plague.

ἔνια τῶν τιμωμένων οἱ ἱερεῖς ἀπάγοντες ὑπὸ σκότῳ μετὰ σιωπῆς καὶ ἡσυχίας ἀπειλοῦσι καὶ δεδίττονται τὸ πρῶτον ἂν δʼ ἐπιμένῃ, καθιερεύουσι καὶ σφάττουσιν,

The priests lead away some of the sacred animals in darkness, silently and quietly, and at first they threaten and frighten them. If the trouble goes on, they consecrate and slaughter them.

Around 300 CE Porphyry objected to Egyptian practitioners' violent threats against the gods. Iamblichus did not deny them. He listed them.

ἢ γὰρ τὸν οὐρανὸν προσαράξειν ἢ τὰ κρυπτὰ τῆς Ἴσιδος ἐκφανεῖν, ἢ τὸ ἐν ἀβύσσῳ ἀπόρρητον δείξειν ἢ στήσειν τὴν βάριν, ἢ τὰ μέλη τοῦ Ὀσίριδος διασκεδάσειν τῷ Τυφῶνι

… that he will smash the heaven, or bring the secrets of Isis to light, or show the forbidden thing in the abyss, or stop the sacred boat, or scatter the limbs of Osiris to Typhon …

His defence was psychological: the threats are aimed at lower, irrational powers that cannot tell true from false (Iambl. Myst. 6.5–7). From Unas threatening to silence the earth to Iamblichus' operator threatening to stop the sun's boat, the same kind of act runs for some 2,600 years, and none of the three sources cites the others. Plutarch also describes the incense κῦφι, mixed "while sacred writings are read aloud to the perfumers" (Plut. De Is. et Os. 80), much as the Pyramid Texts pair natron with words. Ptolemy made the Egyptian gift a fact of nature: under Pisces and Mercury the people of Lower Egypt are "given to magic" (Ptol. Tetr. 2.3).

By the second century CE the image was fixed, and it fills the fiction of the age. Lucian's Pancrates, a sacred scribe from Memphis, "lived twenty-three years underground in the sanctuaries, being taught magic by Isis", and could make a pestle walk and carry water (Lucian Philops. 34–36). Apuleius' Zatchlas, "an Egyptian, a prophet of the first rank", is hired to make a corpse speak (Apul. Met. 2.28), and at the end of the same novel the priest of Isis brings out books "inscribed in unknowable letters" (Met. 11.22). Egyptian writing was valued because outsiders could not read it.

The reputation could be turned against anyone. Celsus claimed that Jesus had worked for hire in Egypt and there "tried his hand at certain powers on which the Egyptians pride themselves" (Orig. C. Cels. 1.28), and pagans told Arnobius that Christ had stolen angel-names from Egyptian shrines (Arnob. Adv. nat. 1.43). Christians replied by making Egypt the home of demonic magic. In Jerome's life of Hilarion a young man of Gaza studies for a year at Memphis and then buries a copper love-tablet under a virgin's threshold (Jer. V. Hilar. 21). Athanasius, bishop of Alexandria, conceded the old power and declared it ended.

ἴσχυε καὶ ἐνήργει παρ’ Αἰγυπτίοις καὶ Χαλδαίοις καὶ Ἰνδοῖς αὕτη καὶ ἐξέπληττε τοὺς ὁρῶντας· τῇ δὲ παρουσία τῆς ἀληθείας, καὶ τῇ ἐπιφανεία τοῦ Λόγου διηλέγχθη καὶ αὕτη, καὶ κατηργήθη παντελῶς.

Magic was strong and worked among the Egyptians, Chaldaeans and Indians, and amazed those who saw it. But at the presence of the truth and the appearing of the Word it too was refuted and wholly abolished.

The medieval vernaculars kept the Egyptian magicians under new names. A German poet of about 1120 calls Pharaoh's two experts goukelære, conjurers or jugglers, who made dragons "by the cunning of conjuring" (Altdeutsche Exodus 1242–1245). The word survives as modern German Gaukler, a street entertainer. In the Old Swedish Konung Alexander of about 1380, Nectanebus still watches wax ships sink in a bronze basin of water to foresee his enemies' fleet, and the poet calls it "his devil's devices" (Konung Alexander, Holm D 4 fol. 111v).

Voices of doubt and mockery

Nothing in the Pyramid Texts doubts heka, and none of the later Egyptian texts surveyed for this chapter does, though two set limits on it. The Ebers papyrus gives the physician's power only for the patient whom the god wishes to keep alive (pEbers, TLA 1.11), and the Teaching for Merikare says that no heka can turn back the ba from its path (Merikare, TLA 1.124). Ipuwer's lament fears exposure, not failure: among the disasters of a world turned upside down, the temple's writings have been removed and "magic is stripped bare" (Ipuwer, TLA 1.208–211). Doubt came from outside. Diodorus explained Proteus' shape-changing as royal head-dresses, "the foreparts of lions, bulls and serpents" worn as signs of rule (Diod. 1.62.4). Juvenal mocked an Isiac priest at Rome who wins a sinner pardon from Osiris, bribed, he says, with a goose and a thin cake (Juv. 6.526–541). Pliny noted that Apion, who claimed to have called up the dead to question them about Homer, never dared say what answer he got (Pliny NH 30.18). Tertullian, who believed in demons, called magic "what almost everyone calls it: deception" (Tert. De an. 57). Chapters XII and XIII return to these. Every sceptic here was a foreigner, and the Egypt they doubted was largely one their own literature had made.

Change over time

Even inside the Pyramid Texts the rites changed from tomb to tomb. Unas's snake spells (PT 226–244) have no copies in the library's Teti or Pepi editions, but those kings have snake spells of their own, built from the same ideas with new wording. The Cannibal Hymn runs to 60 lines for Unas and survives in only 4 for Teti, where its syntax is recast: "it is Teti who eats their heka" (PT 274 §403, Teti). Gesture rubrics were added. Within about 150 years the royal liturgy was revised king by king.

The outside view changed far more, and the library traces it in stages.

The sharpest turn falls between Herodotus and the Septuagint, when translators working in Egypt itself made Pharaoh's priests φαρμακοί and ἐπαοιδοί. Josephus still granted Egyptian sacred scribes real foresight, "for they are clever at telling the truth about the future" (Joseph. AJ 2.205), but from then on the Greek and Latin words for illicit ritual, μάγος, γόης, magus, were applied to Egyptians as readily as to Persians.

What is distinctive about Egypt

Set beside the other cultures in this report, the Egyptian evidence in the library has five features of its own.

First, it is mostly royal and funerary. Elsewhere in this report ritual power often serves private clients; the Pyramid Texts serve one dead king at a time, and no private magician or client appears in them. The later Egyptian texts stay mostly funerary or royal: 179 of the 216 lines that contain heka are in funerary books, royal tombs, temple rites for Pharaoh, a court tale, royal teachings and inscriptions or a royal letter. But 98 of those lines, the largest group, are in Book of the Dead papyri made for private owners, and the medical and protective books serve private patients, among them women in labour (pBrooklyn 47.218.2, TLA 1.103–105), and one household names its magician (Tübingen 460, TLA 1.6).

Second, heka is a substance and a property. Gods are "great of heka", a crown is "Great of Magic", and a king can eat it. Magic elsewhere in the library is mostly something people do; in Egypt it is also something beings have.

Third, it coerces gods openly. Threats stand beside hymns in the Pyramid Texts, and the same kind of threat is reported by Plutarch and defended by Iamblichus two and a half thousand years later. Porphyry found such coercion scandalous; the oldest Egyptian texts treat it as a prerogative of kingship.

Fourth, its power lies in words as sounds: names known and hidden, puns, counted repetitions, and possibly foreign charms written in hieroglyphs. The later Greek idea that Egyptian names lose their force in translation, held by Hermetists, Iamblichus and Origen alike, is a theory of something the pyramids already practised.

Fifth, the outsiders' picture had an Egyptian counterpart. In the outsiders' languages a people's sacred power became the very type of foreign magic; but Egyptian tales already told of court magicians who worked wonders for kings, and an Egyptian king already warned his viceroy against the magicians of Nubia (see 'Heka turned against people' above). Unas ate the gods' heka as a king's right; in Celsus' day Egyptian-trained performers were said to sell their sacred lore in the market-place for a few obols (Orig. C. Cels. 1.68).

Chapter II

Mesopotamia: the exorcist, the witch and the written sky

In the cuneiform texts of Assyria and Babylonia, what we call magic was a salaried, hereditary and highly literate branch of temple scholarship, organised around reading the signs the gods wrote into livers, births and the sky and around undoing the hidden work of the witch.

A witch in the commentary

Sometime in the seventh century BCE, in a house at Assur that modern excavators call the House of the Exorcist, a scholar sat down with the great anti-witchcraft series Maqlû ("Burning") and explained its hard words one by one. The tablet he produced, known as KAR 94, survives in this library. It opens a window on the patient's complaint as the incantation framed it, in the first person, and on the scholar's dry glosses beneath it.

e-le-ni-tu₂ ub-bir₂-⸢an⸣-[ni] / ub-bu-ru ka-[su-u] / SIK₂.UZ₃ im-li-šu ma-la-šu ⸢ba⸣-[qa-mu] / i-ti-iq SAHAR GIR₃-MIN-ia₂ ⸢iš⸣-[bu-šu₂] / ša₂-ba-šu₂ na-⸢sa⸣-[hu]

"The witch has bound me." To bind means to tie up. "They have torn out my hair": tearing means plucking. "They have scooped up the dust of my footsteps": scooping means removing.

The passage holds both halves of Mesopotamian magic as the library preserves it. One is the theory: a woman who obtains a man's hair or the earth his feet have pressed can "bind" him, tying up his body, speech and luck from a distance. The other is the setting. A professional exorcist, trained in a temple school, glosses the witch's methods as he would gloss an obscure verb in an omen. It is a world full of demons, ghosts, curses and portents, described by men who wrote commentaries on their own manuals and dated their tablets by the reigning king.

The culture's own categories

Akkadian has no word that corresponds to "magic". The broad category was ummânūtu, "scholarship", the expertise of the ummânu or master. When Esarhaddon (680–669 BCE) restored the cult of Babylon after his father had destroyed the city, he reinstated its personnel as a single body: cult priests, "guardians of the secret", purification priests, exorcists, lamentation priests and singers, "masters of all scholarship" (RINAP 4 105 vi 21–27).

The exorcist's craft was āšipūtu or mašmaššūtu, the word an Uruk scholar wrote as a heading over a recipe for pounding aromatic resin (SpTU 1 47 o 2–5, with o 11). The diviner's lore was bārûtu. An incantation was a šiptu, closed by a rubric beginning ka-inim-ma, "wording of", that names its ritual; the accompanying actions were the kikiṭṭû, the "procedure". Against these licensed crafts stood kišpū, witchcraft, worked by the kaššāpu (sorcerer) and far more often, in the incantations, by the kaššāptu (witch). The verbs carry the moral geography. The witch "binds" (kasû, ubburu); the exorcist "loosens" or "releases" (pašāru), and the ritual that cancels the evil announced by an omen is called namburbû, from a Sumerian phrase meaning "its loosening".

Between these poles lies the most distinctive concept in the library's material, māmītu. The word means an oath. It also means the curse that falls on whoever breaks an oath, and by extension the curse that clings to anyone who has broken a taboo or committed a sin, even unknowingly. In the incantations māmītu becomes a demon in its own right, adjured to go away: "Oath, oath, be loosed! Withdraw, be far off! Oath, I adjure you by heaven!" (GAAL 2 pl. 17–18, 1–4). The same word names the treaty-oath that bound Assyria's vassals, so a single concept joins what we would split into religion, law, ethics and magic.

Who practised it

The practitioners in the library are almost all men, literate, salaried and attached to temples or the court, often from scholarly families who handed their tablets down for generations.

An administrative letter from the reign of Sargon II (721–705 BCE) shows how ordinary this was. It allots two measures of bread and two of beer to the exorcist and one of each to the temple steward, out of the regular offerings of the temple of Nabû (SAA 1 128 r 16–19).

The exorcist did not claim his power as his own. A bilingual incantation against the oath (māmītu) gives him a formula of delegated authority.

ša₂ {⸢d}e₂⸣-a a-na-ku ... ša₂ {d}DAM.KI.NA a-na-⸢ku⸣ ... DUMU šip-ri ša₂ {d}AMAR.UTU a-na-ku ... ša₂ qa-ti {d}e₂-a ina qa-ti-ia i-ba-aš₂-ši

I belong to Ea; I belong to Damkina; I am the messenger of Marduk. Ea's hand is in my hand.

Ea (Sumerian Enki), god of the fresh-water deep and of wisdom, was the source of every incantation; his son Marduk, identified with the older exorcist-god Asalluhi, was the divine practitioner. A list of gods inscribed for Sennacherib calls Marduk "sage of the gods, exorcist of the Igigi and Anunnaki" (RINAP 3 153 o 15), and a bilingual hymn hails its god as "pure diviner" (CT 58 28 r 8'). The human expert imitated a divine office, and in the culture's own terms that made him legitimate.

Colophons, the scribal notes at the end of tablets, name the experts. At seventh-century Assur the young exorcist Kiṣir-Nabû, son of Šamaš-ibni, copied commentaries on house-protection and on Šurpu (AfO 12 240), one of them labelled his maš'altu, his "questioning", an oral lesson put in writing (JRL 1053). Three centuries later the Uruk exorcists Iqīšâ of the Ekur-zākir clan and Anu-ikṣur of the Šangi-Ninurta clan wrote on eclipses, malformed births, potency, lice and a demon of the washroom. Iqīšâ signed his commentary on birth omens as "exorcist of Uruk" and dated it to the fourteenth of Duʾuzu in the sixth year of "Philip, king of the lands" (SpTU 2 38, 35), that is, Alexander's half-brother Philip Arrhidaeus, in 318 BCE.

The only female specialists the Mesopotamian texts name are hostile. They are the witch of the incantations and the "witches" whom one Assur commentator, in a broken gloss, places under the goddess Bēlet-Akkade as their "mistress" (KAR 94 33'–35'), which would give even illicit witchcraft a place in the divine order; but the line he glosses sends that goddess, Ishtar of Agade, against the witch (Maqlû II 195–197; see 'Maqlû in its own words' below), so the witches' mistress is a power turned against them. Midwives are implied in a birth incantation and ecstatics could be of either sex, but no woman in these texts writes, signs or teaches. In three of the anti-witchcraft rites for pregnancy, though, the woman herself speaks and performs parts of the rite (CMAwR 5.5.1:26–32; see 'Stars, ghosts and births' below), and Maqlû gives the witch the titles of women's religious offices, among them exorcist and ecstatic (Maqlû III 41–44).

What they did

Undoing the witch

The witch worked in secret, through contact with her victim's traces, and through food. The exorcist answered in the open, on her image. KAR 94 explains what lay behind a line of Maqlû III, in the incantation "O you who have performed sorcery against me", where the speaker threatens the witch with residues from the kiln and the cooking-pot.

a-laq-qa-kim-ma ha-ha-a ša₂ UDUN dih₂-me-e ša₂ UTUL₂ / ma-a NU ša₂ IM DU₃{uš} / kur-ban-nu ša₂ UDUN dih₂-me-e ša₂ UTUL₂ / a-mah-ha-ah ina SAG.DU NU ša₂ IM a-tab-bak

"I take slag of the kiln and ash of the pot against you." That means: he makes a figurine of clay. "A clod of the kiln and ash of the pot I soak, and I pour it over the head of the clay figurine."

The figurine (ṣalmu, written NU) stood for the witch. What was done to it was done to her. In Maqlû the climax was burning, and a Late Babylonian schoolboy copied the cry to the fire-god on an exercise tablet: "I shall burn them in a blazing kiln. Girra, burn my sorcerer and my sorceress!" (LB school tablet, 15–16). The same tablet carries lexical lists and the creation epic, so anti-witchcraft liturgy was part of an ordinary scribal education.

Witchcraft could also explain ailments a modern doctor would see as poisoning or neurological disease. The Diagnostic Handbook (Sakikkû) describes a patient who cannot speak properly, whose hands and feet shake, who salivates and keeps vomiting, and a Babylonian commentator explains its key phrase.

[ne₂-per-ku-u₂] ba-ṭa-lu ŠU-MIN-šu₂ u GIR₃-MIN-šu₂ i-ra-ʾu-bu ... [at-mu-u₂] da-ba-bi il-la-ti ru-ʾu-tu lu-ʾa-a-ti ⸢šu-kul⸣ / [lu-ʾa-(a)]-⸢ti⸣ kiš-pi ŠA₃-šu₂ ana BURU₈{a-re-e}e i-te-ne₂-el-la-⸢a⸣-[ma]

"Halting" means ceasing. "His hands and feet shake." ... "Talk" means speaking; "slaver" means spittle. "He has been fed debility": debility means witchcraft. "His stomach keeps rising to vomit."

Witchcraft was eaten. A commentary by Iqīšâ on the potency texts records a man "made to moan through witchcraft" in a series concerned with his desire for his wife (SpTU 2 39 r 4'–8'). Sexual failure, like the loss of speech, was a classic sign that someone had worked against you.

Expelling demons and drawing boundaries

Demons came in classes. The bilingual incantation against the oath lists the utukku, the alû, the gallû, the ghost (eṭemmu, the spirit of a dead person), the child-killing demoness Lamaštu and the night-demons lilû. Against all of them the exorcist drew a line on the ground.

lu-u u₂-tuk-ku lem-nu lu-u a-lu-u lem-nu lu-u e-ṭem₂-mu lem-nu ... ša₂ a-na zi₃-sur-re-e ša₂ {d}ni-sa-ba i-ṣar-ru-ru / ... sa-pa-ru ša₂ {d}ni-sa-ba lik-su-šu₂

Be it an evil utukku, an evil alû or an evil ghost ... whoever casts his eye on the flour circle of Nisaba, may the net of Nisaba bind him.

The zisurrû was a circle of flour poured around the patient or the bed; Nisaba was the goddess of grain. Here the exorcist turns the language of binding against the demons. A rubric from Assur names the ritual "for the flour circle and the drawn design" and declares that "the incantation of the Apsû and of Eridu cannot be undone" (VS 24 54 r 2'–3'); the Apsû was Ea's subterranean ocean and Eridu his city. The house itself could be sealed by oath. Kiṣir-Nabû's commentary on the incantation "The house is conjured" identifies, opening by opening, the windows the demon is forbidden to use: "the side window" is the bathroom window, "the outer side window" the window of the animal pen (AfO 12 240, 1–6).

Entering a patient's house, the exorcist recited that Šamaš the sun-god walked before him, Sîn the moon-god behind, Adad on his right and Ninurta on his left; the Assur commentator read this as an instruction to face the rising sun at dawn (JRL 1053, 11–14). Another copy puts Nergal, god of the underworld, on the right, and glosses the demon "who has set foot in the netherworld" as someone who was killed (LKA 82 o 6–11).

Healing

No sharp line divided the exorcist from the physician. The Uruk exorcist Anu-ikṣur explained a demon of the lavatory by taking its name apart into cuneiform signs.

a-na E₂ mu-sa-a-tu₂ la₃ KU₄{ub} {d}ŠU.LAK SIG₃-su / {d}ŠU.LAK ša₂ E{u₂} ŠU qa-tu₄ la la-a KU₃ el-lu / ana E₂ mu-sa-a-tu₂ KU₄{ub} ŠU₂-šu₂ NU KU₃ ana UGU qa-bi

"He must not go into the washroom, or Šulak will strike him." It says "Šulak" because ŠU means hand, LA means not, KU means pure: it is said of one who goes into the washroom with unclean hands.

The learned wordplay delivers a hygienic rule. The same man's commentary on a remedy for scalp disease explains upul as "one whose head the lice gnaw", identifies "black river-stone" as sulphur, and promises that "the old hair falls out and new hair comes out" (SpTU 1 51 r 2–8).

Seizures were described with special care. A commentary on the fumigation series Qutāru defines "Fallen-from-Heaven" (antašubbû) as a condition in which the patient chokes and his spittle runs, "Lord-of-the-Roof" as one in which his eyes roll back, and "Hand-of-a-god" as one in which he curses the gods (BRM 4 32 o 1–3). Each disease was the "hand", the touch, of some agent. Diagnosis meant identifying who had touched you; therapy meant driving that agent off. The same tablet then turns to the ingredients.

A.RI.A NAM.LU₂.U₁₈.LU {u₂}maš-ta-kal ... {u₂}IGI-lim ki-ma GIR₃.PAD.DU NAM.LU₂.U₁₈.LU ... MUD₂ ka-mi-i MUD₂ {lu₂}ga-ar-ba-nu

"Human semen" is the maštakal plant ... the "It-faced-a-thousand" plant is like human bone ... "blood of a captive" is "blood of a leper".

Modern editors call such entries Decknamen, cover names. Recipes that sound like a pharmacy of filth turn out, at least in part, to be coded herbal prescriptions, though some such ingredients may have been meant literally. A Nineveh commentary similarly glosses the "plant of Šamaš" as šakirû, an identification often rendered "henbane" but disputed (K.4320 o i 1–3, ii 1).

Childbirth had its incantation too. A commentary on "The woman giving birth is a boat loaded with lapis lazuli" explains that the reed is mentioned "because the reed is what cuts the umbilical cord" (OIC 22 131 no. 20, 1, 12–13). Its author, Enlil-kāṣir, a lamentation priest of Enlil at Nippur, passes from the myth of the moon-god's cow in labour to the instruction to reach in with the hand, so that mythic analogy and midwifery share a single text.

Reading the signs

Divination was the largest and most prestigious craft, and the one most bound up with the king. Its theory was that the gods wrote: on the liver of a sacrificed lamb, in the moon and planets, in a malformed birth, in thunder, in a snake on the threshold, in a remark overheard in the street. Trained readers decoded the signs by rules of analogy, collected in vast omen series of the form "If X, then Y".

The liver came first. An extispicy commentary from Ashurbanipal's library marks some configurations as "omens that not every diviner knows" and speaks of "lines that are strange in the diviner's lore" (CT 30 43 o 14'–15', r 6'–7'). Malformed births (izbu) were read by resemblance. In Iqīšâ's commentary a newborn "covered with scales like a goat-fish or a snake" means the anger of the moon-god, and a man will be covered with a scaly skin disease (SpTU 2 38, 21–22); in another, a birth with a lion's head means "a prince will seize kingship of the world" (CM 31 139, 1–2). Everyday omens of the series Šumma ālu ("If a city") made the same act good or bad according to the day: a man who kills a snake on the moonless twenty-ninth will be free of his fears, and the commentary refers him to the matching namburbû (Commentary A on Šumma ālu 22–23, 29–31).

The sky was read with the greatest system. Venus rising in the morning watch means killings in the land (ACh Ištar 5, 20–21); Mercury flickering at its rising in early summer means a good harvest and falling prices (ACh Ištar 21 o 13'–14'). Weather was part of the sign. "Thunder out of its time" is defined as thunder in high summer, when Mesopotamian thunder is in fact rare (ACh Adad 25 o 1–3), and an eclipse with cloud, thunder, rain or lightning each changes the forecast (BM 47447, 1–3).

The richest single text is Iqīšâ's commentary on the lunar-eclipse tablet of Enūma Anu Enlil, copied at Uruk in the second year of Philip, 322 BCE (SpTU 4 162 r 21'–23'). An eclipse in the constellation of the Hired Man (roughly our Aries) means "the king of Akkad will die", and the scholar must weigh the wind he has recorded before he "gives the decision to the king of Akkad" (SpTU 4 162 o 6–9). A few lines later comes a rule of a different kind.

KU₄ ina U₄.NA₂.A dele-bat ana ŠA₃ 30 KU₄-ma 1 me AN.GE₆ DUG₄.GA / [U₄].NA₂.A AN.GE₆ u₂-kal-lam-ka ul-tu U₄.NA₂ ša₂ {iti}GAN / [ana] U₄ AN.GE₆ ša₂ {iti}BARA₂ ta-qab-bi 1 me U₄{mi}

"If on the day the moon disappears Venus enters the moon, you announce an eclipse for a hundred (days)." The day of disappearance shows you the eclipse: from the disappearance day of Kislīmu to the eclipse day of Nisannu you count a hundred days.

This forecasting rule sits in an omen commentary by a man whose title was exorcist. Babylonian scholars of this period could compute lunar eclipses from long-observed periodicities, but the Venus condition is the kind of correlation an omen series collects, and the plausibility panel found that it would usually fail however "enters the moon" is read (chapter XII). Here it shows that exorcism, astronomy and political advice were one profession.

The king's divination

The Assyrian royal inscriptions publish divination as public proof of divine favour. Sennacherib (705–681 BCE), choosing his younger son Esarhaddon as heir, "asked Šamaš and Adad by extispicy, and they answered him a firm yes: he is your replacement"; the people of Assyria then swore "by water and oil" to protect the new heir (RINAP 4 1 i 12–14, 50–51). Before new statues of the gods were made, Esarhaddon "lined up the diviners", and the extispicies "agreed as with one mouth" (RINAP 4 48 r 73–75). The statues were then brought to life by "mouth-washing" (mīs pî), performed at night in the temple courtyard "before the stars of heaven ... through the craft of the sage" (RINAP 4 1015 r v 10–13), the closest the library comes to the later Greek idea of animating images. Sennacherib laid the foundation of a festival house "in a good month, on a favourable day, through the craft of purification-priesthood and the wisdom of exorcism" (RINAP 3 168, 29–31).

Unsolicited signs counted too. Esarhaddon reports "messages of ecstatics" and good signs "in dreams and in chance utterances" (egerrû, a remark overheard and taken as an oracle) (RINAP 4 57 ii 12–19; compare RINAP 4 1 i 73, ii 5–6). Jupiter, Marduk's star, "reached its place of exaltation" as a sign that Babylon might be rebuilt (RINAP 4 104 ii 33–41), and in the diviner's bowl the flesh gave "reliable omens" for the same work (RINAP 4 104 iii 12–14).

Any general account would add the substitute-king ritual, in which, when an eclipse portended the king's death, a stand-in was set on the throne to absorb the evil. It is known from the scholars' letters in SAA 10, which this library does not hold, and no text here attests it. The library does hold its logic: the eclipse omen naming the king of Akkad as victim, and the instruction to give the king a decision (SpTU 4 162 o 6–9). The doctrine that signs warn rather than doom appears in a commentary on a namburbû for the case "if evil signs are repeatedly seen against a man and his house", whose equipment includes an unfired clay vessel and "a figurine of the sorcerer and the sorceress" (BM 38271, 10–11, 20–21). The evil an omen announced could be traced to witchcraft and discharged onto effigies before it arrived.

Oaths and curses

The state used the witch's powers openly. Royal inscriptions close by cursing whoever alters them: may Aššur and the great gods "curse him with a grievous curse that cannot be loosened, overthrow his kingship and take his life" (RINAP 3 10, 25–29). The phrase arrat lā napšuri, "a curse without release", is the exact opposite of the exorcist's loosening. Treaty-oaths bound vassals. Samsi, queen of the Arabs, "had transgressed the oath of Šamaš" (RINAP 1 TP III 20, 18'); the people of Ekron put their king Padî, "bound by treaty and oath to Assyria", in iron fetters (RINAP 3 4, 42). When a rebel governor fled to Elam and was killed there, Esarhaddon presented the death as the oath doing its work: the gods "imposed a heavy punishment on him" because of the oath he had broken (RINAP 4 1 ii 55–57).

The dead could be made to enforce a curse. The tomb of Yabâ, queen of Tiglath-pileser III, found at Nimrud, carries this warning.

ša₂ {na₄}KIŠIB ša₂ KI.MAḪ šu-a-tu₂ BAD-u₂ / e-le-nu ina šu-ru-ru {d}šam-ši / e-ṭe₂-ma-šu₂ ina ṣu-me-e ka-ma-te / li-ir-pu-du / šap-la-nu ina KI-tim ina na-qa me-e ... la i-ma-ḫar

Whoever breaks the seal of this tomb: above, in the rays of the sun, may his ghost wander outside in thirst; below, in the underworld, when water is poured, may it receive nothing.

The robber is to become the kind of hungry, unquiet ghost the exorcist's rituals laid to rest. Scholarly tablets were protected the same way. Iqīšâ's birth-omen commentary ends: "whoever carries it off, may Adad carry him off" (SpTU 2 38, 34).

The craftsman's ritual

The glass-making texts from Ashurbanipal's library wrap exact recipes in ritual.

e-nu-ma uš-ši ku-u₂-ri ša NA₄ ŠUB{⸢u₂⸣} / i-na ITI šal-me U₄ ŠE.⸢GA KIN⸣.KIN-ma uš-ši ku-u₂-ri ta-nam-di ... {d}KU₃.BU-MEŠ tu-⸢še⸣-[šib]-⸢ma⸣ ša₂-nu-u a-hu-u NU KU₄{ub} NU KU₃ a-na pa-ni-šu₂-nu la GIB

When you lay the foundations of a kiln for "stone": you look for a favourable month and a propitious day, and you lay the foundations ... You seat the kūbu gods; no outsider or stranger may come in, no one impure may pass before them.

The kūbu were deified stillborn foetuses, dangerous beings who received at the kiln the kispu, the offering normally made to the dead (Glass text K 4266+ r i 33'–36'). On the same tablet as the kiln ritual stands a quantitative recipe for blue glass: ten minas of sand, fifteen of plant ash, a little "white plant", ground separately, mixed, and melted over "a good, smokeless fire" (K 2520+ o 13–16). The texts treat the ritual and the chemistry as equally necessary parts of the procedure.

How it was supposed to work

The commentaries rarely theorise, but their glosses imply consistent assumptions.

• The world is written. Gods inscribe their decisions in livers, stars, births and events, and trained readers decode them. What is written can also be read again.
• Signs warn. An omen announces what will happen unless something is done, and the namburbû does it.
• Illness has an author. A god, ghost, demon or witch has laid a "hand" on the patient, and naming the agent is half the cure.
• Contact and substitution. The witch acts through hair, dust and food; the exorcist acts on her image, which suffers in her place.
• Sin binds. Moral offences generate a māmītu that clings to the offender. Kiṣir-Nabû's commentary on Šurpu II glosses the sin of one who "said no for yes, yes for no" as denying what was said and asserting what was not (AfO 12 240, 39–42). Confession and ritual burning released sin and curse together. Šurpu III adds a long litany of curses incurred unknowingly (see 'Šurpu at Uruk and Nimrud' below), such as "sitting facing the Sun", which the divine exorcist Asalluhi releases (KAR 94 46'–49').
• Authority is delegated. The exorcist's words are Ea's words, which is why "the incantation of Eridu cannot be undone".

The theology of history used the same terms. Esarhaddon explained his father's destruction of Babylon by saying that under an earlier king "evil omens appeared" and the people "answered one another yes for no and spoke lies" (RINAP 4 104 i 19–24), the very sin of Šurpu II. Then he described how the sentence was shortened.

⸢mar-qi₂-ti⸣ 70 MU.AN.NA.MEŠ / ⸢mi⸣-nu-ut ni-du-ti-šu / ⸢iš⸣-ṭur-ma re-me-nu-u₂ / {d}AMAR.UTU sur-riš lib₃-ba-šu / i-nu-uḫ-ma e-liš a-na / šap-liš uš-bal-kit-ma / a-na 11 MU.AN.NA.MEŠ / a-šab-šu₂ iq-bi

He had written seventy years as the length of its desolation. But merciful Marduk's heart was quickly calmed; he turned it upside down and ordered its resettlement in the eleventh year.

In cuneiform 70 is written with the sign for 60 followed by the sign for 10; in reverse order the same two signs give 10 and 1, eleven. The inscription presents this as Marduk's mercy. Modern historians see the king's scholars finding in the god's written decree a reading that let rebuilding begin at once. Both accounts share a premise, that a divine decision is a text and experts in texts may interpret it.

Where the line was drawn

The boundary between licensed and illicit practice was clear in principle and invisible in technique. Witch and exorcist both made figurines, spoke binding formulae and worked through material substances; the exorcist bound demons with Nisaba's net as surely as the witch bound her victim. What separated them was authorisation, openness and aim. The exorcist acted as Ea's man, openly, for a patient, from tablets his profession could read; the witch acted on her own account, in secret, to harm. The state's curses were legitimate binding because they served the gods' justice.

How the law punished the illicit kind cannot be documented here. Scholars know from the laws of Hammurabi, which the library lacks, that an unproven accusation of witchcraft could go to the river ordeal. The library preserves the ordeal itself. A commentary on the poem Ludlul bēl nēmeqi explains the healed sufferer's passage along "the bank of the River, where people's cases are decided" as huršān, the ordeal, in which the divine River judged guilt (K.3291 r 30–31). In the ritual texts the witch is punished through her image, burned in effigy and handed to the fire-god, so that in the library's sources the punishment of witchcraft is itself a rite performed by the exorcist.

The licensed crafts kept secrets of their own. Babylon's cult experts were "guardians of the secret" (pirištu) (RINAP 4 105 vi 21–27); some liver omens were known "not to every diviner"; recipes hid their plants behind cover names; glassmakers barred strangers from the kiln. Secrecy went with expertise and did not mark a practice as illicit.

Voices of doubt

No text in this corpus mocks magic or calls its practitioners frauds. Witchcraft is feared and fought; nobody says it is impossible. Doubt appears in two subtler forms.

The first is theological. Ludlul bēl nēmeqi ("I will praise the lord of wisdom"), a Babylonian poem composed in the late second millennium and preserved here through a seventh-century commentary and in Assyrian copies of its first tablet from Sultantepe and Nimrud (STT 1 32; CTN 4 201), is the complaint of a righteous man whom every expert failed.

i-na maš-šak-ki {lu₂}⸢ENSI⸣ ul u₂(I-)ša₂-pi ⸢di-ni⸣ / maš-šak-ku sur-qe₂-nu ša₂ {lu₂}⸢ENSI⸣ / {lu₂}MAŠ.MAŠ ina KID₃.KID₃{ṭe₃-e} ki-mil-ti ul ip-ṭur / KID₃.KID₃{⸢ṭu-u₂⸣} ne₂-pe-ši

"The dream-interpreter did not clear up my case with his incense": maššakku is the dream-interpreter's incense offering. "The exorcist did not undo the anger against me with his ritual": kikiṭṭû means procedure.

The sufferer is healed only when Marduk relents. The poem admits that rites can fail and locates the failure in the god's will, leaving the craft intact; the commentator glosses it without remark.

The second form is procedural. The Assyrian court knew that diviners might tell the king what he wished to hear. One inscription, probably describing Sennacherib's choice of heir (the attribution is uncertain), records a precaution that reads like a modern blind trial.

⸢ul⸣ u₂-pat-ti ša DUMU.MEŠ {lu₂}⸢ḪAL pal⸣-ka-a ḫa-si-sun₂ / i-na ṭup-pi iš-ṭur e-ri-im ib-⸢ri⸣-im id-din-šu₂-nu-ti / ⸢ul⸣ iṣ-lul kal mu-ši a-di ip-pu-ḫa ... {d}UTU

He did not open the matter to the diviners, broad as their understanding was. He wrote it on a tablet, wrapped it, sealed it and gave it to them. He did not sleep all night until Šamaš rose.

The question was hidden so that the answer could not be steered, as the extispicies about the gods' statues had to agree "as with one mouth". Such checks guarded against human bias and assumed that the method works when honestly applied. Nobody in the library questions the gods' writing on the liver itself.

Change over time

The library's Mesopotamian texts span well over a millennium, though most copies are late. The Old Babylonian period (early second millennium) appears through a scribal examination, copied later, in which the student must know "the language of the purification priest" among other professional jargons (Examination Text A, 9–10), and through the Sumerian roots of the bilingual incantations. The Kassite period (later second millennium) gave Ludlul and the canonical shape of the great series.

The evidence is densest and most public at the Neo-Assyrian court of the eighth and seventh centuries. Kings announced extispicies, eclipses and prophecies in stone, bound vassals by oath, and gathered the scholarly corpus at Nineveh. Esarhaddon even carried off Egyptian ritual experts after taking Memphis: "exorcists, ḫarṭibu, ... veterinarians, Egyptian scribes ... snake charmers" (RINAP 4 9 i' 9'–11'). Ḫarṭibu renders the Egyptian title of the chief lector-priest and comes from the same Egyptian title as the Hebrew ḥarṭummîm, Pharaoh's magicians in Genesis and Exodus (see chapters I and III). Foreign expertise was booty worth taking.

The texts did not fade when Assyria fell. Under Persian, Macedonian and Seleucid rule the temple scholars of Uruk, Babylon and Nippur kept copying and explaining. The same incantation against the oath survives in copies from Nineveh, Assur, Sultantepe, Nippur, Ur, Uruk, Sippar and Babylon (GAAL 2 pl. 17–18; VS 24 54). What changed was the direction of interpretation, as commentators increasingly read the incantations against the sky. The Assur commentator had already said that "Šamaš" in Šurpu means Jupiter (KAR 94 46'–49'). A Hellenistic-period commentary from Babylon assigns each of Asalluhi's boasts ("I am Asalluhi, the diviner who decides the verdict") to a region of the zodiac, here Sagittarius, and glosses the sign HAL as both "divination" and "secret" (BM 47529+, 14–15). The system grew more astronomical and more exegetical, and the old demons kept their places within it.

What is distinctive here

Almost everything we can see of Mesopotamian magic is written scholarship produced by the state's own experts, with commentaries on commentaries and a curriculum that set anti-witchcraft liturgy beside lexical lists. Divination dominates, as a vast rule-based science of signs tied more closely to kingship than anywhere else in the library. The archetypal evil is the witch, female in the incantations, working in secret through bodily traces and food, and the archetypal remedy is to burn her image. Sin, oath, curse and disease are fused in māmītu, so that a lie, a broken treaty and an unlucky seat facing the sun are all removed by the same release. The texts also hold things found nowhere else in the library: deified stillborn children guarding a glass furnace, drug cover names decoded in writing, a demon's name etymologised from the signs that spell it, and a god's seventy years reread as eleven. Later chapters meet the Chaldaean astrologer and the Babylonian magus as figures of Greek and Roman imagination. Here they appear in their own records, as temple scholars drawing a double ration of bread and beer.

Chapter III

Israel: forbidden arts and sanctioned signs

The Hebrew Bible bans the diviner, the sorceress and the medium in some of the harshest laws of the ancient world, yet it keeps priestly lots, a curse dissolved in water and a bronze serpent that heals, because it judged a rite by the power it served and by the person who performed it.

A night at Endor

Saul, the first king of Israel, had driven the mediums out of his land (1 Sam 28.3). On the eve of a battle with the Philistines he asks his god for guidance and hears nothing, "neither by dreams, nor by Urim, nor by prophets" (1 Sam 28.6), the three channels the Bible approves. So he disguises himself and goes by night to a woman at Endor whom the Hebrew calls "mistress of a ghost" (baʿălat ʾôb, 1 Sam 28.7 WLC), and asks her to bring up the man he will name (1 Sam 28.8). She suspects a trap: "why then lay you a snare for my life, to cause me to die?" (1 Sam 28.9). Saul swears by Yahweh that she will come to no harm, and asks for Samuel, the dead prophet who had made him king.

The king said to her, Don’t be afraid: for what do you see? The woman said to Saul, I see a god coming up out of the earth.

"Do not be afraid. What do you see?" The woman said to Saul, "I see ʾĕlōhîm, a divine being, rising out of the earth."

Nothing in the Bible's fullest consultation of the dead suggests that the rite failed. Only the woman sees. Saul recognises Samuel from her description of an old man in a robe, and Samuel foretells Saul's death on the following day, a prophecy the story then fulfils (1 Sam 28.12–25). Her words are ʾĕlōhîm rāʾîtî ʿōlîm min-hāʾāreṣ (1 Sam 28.13 WLC). ʾĕlōhîm is the ordinary word for God or gods; the participle with it, ʿōlîm, "coming up", is plural, though Saul's next question treats the figure as one, "What form is he of?" (1 Sam 28.14 WLC), and the Greek translators kept the plural: Θέους ἑόρακα ἀναβαίνοντας ἐκ τῆς γῆς, "I have seen gods coming up from the earth" (1 Sam 28.13 LXX). Then the king collapses, and the medium, whose trade could cost her her life, kills her fattened calf and bakes bread for him and his men before they leave in the dark (1 Sam 28.24; 28.25).

The scene holds most of the tensions of this chapter. In the story a forbidden practice works, the king who banned it uses it in secret, and the forbidden woman is the kindest person in the story. The Greek version of Chronicles, condemning Saul for the consultation, adds "and Samuel the prophet answered him" (1 Chr 10.13 LXX). The Greek Bible also calls the woman an ἐγγαστρίμυθος, a "belly-talker", the everyday Greek word for a ventriloquist medium (1 Sam 28.7 LXX), a word that fed later Christian argument over whether the woman threw her voice.

The words for it

The Hebrew texts have no single term matching the modern word "magic". They name specialists, and most of the names survive in lists of what is forbidden. The table sets the main terms beside the English of the World English Bible (WEB), one of the forms in which the library stores the Old Testament, and the Septuagint (LXX), the Greek translation made by Jews in Egypt from the third century BCE. The Hebrew forms are those of the library's Hebrew text, the Westminster Leningrad Codex (WLC); the senses come from standard lexical scholarship, and several meanings are uncertain. What the Hebrew adds is set out after the list in Deuteronomy below.

Most of these words name ways of finding things out. The master list in Deuteronomy, the law code associated with King Josiah's reform in the late seventh century BCE, opens with child sacrifice and then names seven or eight specialists, as the Hebrew is counted (see below).

There shall not be found with you anyone who makes his son or his daughter to pass through the fire, one who uses divination, one who practices sorcery, or an enchanter, or a sorcerer,

Let there be found among you no one who passes son or daughter through the fire, no diviner of divinations (qōsēm), no soothsayer (mĕʿônēn), no omen-reader (mĕnaḥēš), no sorcerer (mĕkaššēp).

The next verse adds the spell-binder, the one who asks a ghost or a "knowing one", and the one who seeks the dead (Deut 18.9–14). All but two of them seek knowledge from signs or from the dead. Only the sorcerer and the binder act directly on the world. The Greek turned the list into professions Alexandrian readers knew: μαντευόμενος μαντείαν, κληδονιζόμενος καὶ οἰωνιζόμενος, φαρμακοῖς, "one who prophesies prophecies, reads chance words, takes bird-omens, drug-men" (Deut 18.10 LXX). Calling sorcerers φαρμακοί tied sorcery to φάρμακον, a drug, remedy or poison, and the φαρμακεία of Paul and Revelation (chapter VII) descends from that choice.

Who practised

The foreigners are mostly men at royal courts: Pharaoh's ḥarṭummîm (Gen 41.8) and Nebuchadnezzar's "magicians, and the enchanters, and the sorcerers, and the Chaldeans" (Dan 2.2–13). The Philistines consult "the priests and the diviners" (1 Sam 6.2–5), and the elders of Moab bring the seer Balaam "the rewards of divination" (Num 22.7).

At home the picture is more mixed than the laws suggest. In the eighth century Isaiah lists "the diviner" beside the judge, the prophet and the elder among the supports God will take from Jerusalem, and in the next verse the "clever enchanter", without a word of blame (Isa 3.1–3). Among the finery of the city's wealthy women he names lĕḥāšîm, "whisperings", usually taken to be inscribed amulets, and condemns them as luxury (Isa 3.20). Kings appear as patrons: Manasseh "practiced sorcery, and used enchantments, and dealt with those who had familiar spirits" (2 Kgs 21.6), and the foreign queen Jezebel is accused of "witchcraft" in the same breath as "prostitution" (2 Kgs 9.22).

Women appear where the practice is domestic, cheap or necromantic, and the only detailed picture of Israelite specialists at work is a picture of women.

and say, Thus says the Lord Yahweh: Woe to the women who sew pillows on all elbows, and make kerchiefs for the head of [persons of] every stature to hunt souls! Will you hunt the souls of my people, and save souls alive for yourselves?

Say, Thus says the Lord Yahweh: Woe to the women who sew bands on every wrist and make veils for heads of every height, to hunt lives. Will you hunt down the lives of my people and keep lives alive for yourselves?

Ezekiel, a priest among the exiles in Babylonia, describes "daughters of your people" who kill some souls and keep others alive, paid "for handfuls of barley and for pieces of bread" (Ezek 13.19). The Hebrew words for the bands and veils are rare and disputed, and the WEB's "pillows" follows an old Greek guess (Ezek 13.17–23).

Households kept their own ritual objects. Rachel steals her father's teraphim and hides them in a camel saddle (Gen 31.34). David's wife Michal lays one in his bed with goat hair at its head to fool the king's men (1 Sam 19.13), so such an object stood in the house of Israel's greatest king and was large enough to pass for a man in a bed. Snake-charmers were a familiar trade: the wicked are like a cobra "which doesn’t listen to the voice of charmers, No matter how skillful the charmer may be" (Ps 58.5). On the sanctioned side stand Moses and Aaron, the high priest, the prophets Elijah and Elisha, Daniel and, late, the angel Raphael, whose deeds often resemble what the law condemns.

What they did

Asking the god

Israel kept a divination of its own at the centre of its cult. The high priest carried the Urim and Thummim in his breastpiece "on Aaron’s heart, when he goes in before Yahweh" (Exod 28.30). The Greek of one passage keeps a fuller prayer of Saul that sets out a two-way choice, δὸς δήλους, "give the Clear Ones", for one outcome and δὸς δὴ ὁσιότητα, "give Holiness", for the other (1 Sam 14.41 LXX). Lots decide which goat on the Day of Atonement goes "for the scapegoat", Hebrew la-ʿăzāʾzēl, often read as the name of a wilderness demon (Lev 16.8).

The early stories also let their heroes use techniques the laws would ban. Joseph's silver cup is the one in which, the Greek says, οἰωνισμῷ οἰωνίζεται, "he takes omens" (Gen 44.5 LXX), a method well known in Mesopotamia; the narrator never comments, though the claim may be part of Joseph's ruse. The prophets record forbidden forms: "My people consult with their wooded idol, And answer to a stick of wood" (Hos 4.12). Ezekiel watches the king of Babylon at a fork in the road: "he shook the arrows back and forth, he consulted the teraphim, he looked in the liver" (Ezek 21.21). Ezekiel does not say the omen lies. The lot falls on Jerusalem because Yahweh is using the Babylonian's sheep-liver to carry out his sentence (Ezek 21.21–23). At the Persian court Haman casts pûr, the lot, "from day to day, and from month to month" to choose a date for killing the Jews (Esth 3.7).

Calling up the dead

Necromancy is the art the Hebrew Bible attacks most often and most concretely. Isaiah records the sound the spirits made: "Consult with those who have familiar spirits and with the wizards, who chirp and who mutter" (Isa 8.19). His objection concerns allegiance: "shouldn’t a people consult with their God?" He does not say the dead are silent. Leviticus calls turning to the mediums a way "to play the prostitute after them" (Lev 20.6), the language the prophets used for worshipping other gods.

Cursing and binding

The Balaam story is a whole narrative about a hired curse. King Balak of Moab believes "he whom you curse is cursed" and pays Balaam, who builds seven altars and goes off to seek omens (Num 22.6–7; 23.1–3). Yahweh turns each curse into a blessing, and Balaam declares "Surely there is no enchantment with Jacob; Neither is there any divination with Israel" (Num 23.23). The Hebrew preposition can mean "in" or "against", so the line either defines Israel as a people without diviners or declares that no spell can touch it.

Yahweh's prophets curse too, and in the stories their curses kill. When boys jeer at Elisha's baldness he "cursed them in the name of Yahweh. There came forth two she-bears out of the wood, and mauled forty-two lads of them" (2 Kgs 2.24). Dying, he has King Joash strike the ground with arrows and rebukes him for stopping at three: "You should have struck five or six times" (2 Kgs 13.19). The number of blows fixes the number of victories. In the narrative even enemies' rites could work. Besieged, King Mesha of Moab burns his heir on the city wall, "There was great wrath against Israel", and the siege is lifted (2 Kgs 3.27). The historian does not say whose wrath it was.

Healing and protection

The best-known healing image in the Bible was commanded by God. When snakes bite the people, "Moses made a serpent of brass, and set it on the standard", and whoever looked at it lived (Num 21.9). An image of the harm cures the harm. The Philistine diviners prescribe the same logic, golden "images of your tumors, and images of your mice" sent to Israel's god (1 Sam 6.5), and in the story their advice works. Priestly rites protect in ways that would be called magic on anyone else's ground. Lamb's blood on the doorposts keeps "the destroyer" out on Passover night (Exod 12.23). After skin disease the priest dips a live bird in another bird's blood, sprinkles the patient seven times "and shall let the living bird go into the open field" (Lev 14.7). In Israel's own terms this is purification. The priest declares the person clean once the disease has gone and claims no cure.

The prophets' cures are physical and often numbered. Elisha revives a dead boy by lying on him "his mouth on his mouth, and his eyes on his eyes" (2 Kgs 4.34), and, after his death, revives a corpse that touches his bones (2 Kgs 13.21). David's lyre drives off the "evil spirit" that troubles Saul, a spirit sent "from God" (1 Sam 16.23).

The fullest recipe comes late and in Greek. In Tobit, a Jewish tale of the third or second century BCE set in Assyria, the angel Raphael tells young Tobias that a fish's heart and liver, burned as smoke, will drive off any demon, and that its gall cures white films on the eyes (Tob 6.8). Tobias's bride has lost seven husbands to the demon Asmodeus. On the wedding night he lays the organs on the incense ashes.

ὅτε δὲ ὠσφράνθη τὸ δαιμόνιον τῆς ὀσμῆς, ἔφυγεν εἰς τὰ ἀνώτατα Αἰγύπτου, καὶ ἔδησεν αὐτὸ ὁ ἄγγελος.

When the demon smelled the odour, it fled to the farthest parts of Egypt, and the angel bound it.

A substance moves the demon, the angel binds it with the standard verb of exorcism, and because an angel taught the method the text counts it pious. After a battle in 163 BCE, Judas Maccabeus's men found ἱερώματα τῶν ἀπὸ Ἰαμνίας εἰδώλων, "sacred tokens of the idols of Jamnia", under the tunics of their dead (2 Macc 12.40). The author concludes that the charms meant to protect them had killed them.

The bitter water

The clearest sanctioned rite built from the methods of binding magic is the ordeal of the suspected wife, in a priestly law usually dated to the sixth or fifth century BCE that may preserve an older practice. A husband who suspects adultery but has no witness brings his wife to the priest, who mixes holy water with dust from the sanctuary floor. The Greek calls it τὸ ὕδωρ τοῦ ἐλεγμοῦ τοῦ ἐπικαταρωμένου, "the water of proof that brings the curse" (Num 5.18 LXX). The priest pronounces the curse and the woman answers "Amen, Amen" (Num 5.22).

The priest shall write these curses in a book, and he shall blot them out into the water of bitterness.

The priest shall write these curses on a scroll and wash the writing off into the bitter water.

She drinks the words. If guilty, "her body will swell, and her thigh will fall away" (Num 5.27); if innocent, "she shall be free, and shall conceive seed" (Num 5.28). Writing a curse and dissolving it into a drink so that it enters the body is among the commonest operations of ancient magic, and here a priest performs it inside the Torah. In the text's own terms the oath before Yahweh, the sanctuary dust and the priest make it lawful. Only women were tested this way. Chapter XII asks whether water and dust could have done anything at all.

How it was supposed to work

The narratives rarely explain, but a few mechanisms recur. Likeness: a bronze snake against snakebite, golden mice against mice, arrows struck on the ground as victories. Transfer: the goat carries sins away, the released bird carries impurity. Ingestion: the curse drunk in water. Contact: Elisha's body on the child's, a corpse touching bones. Smell: the demon fleeing the smoke. Number: seven sprinklings, seven bathings for the Aramean general Naaman, three stretchings by Elijah over a dead boy (1 Kgs 17.21).

Naaman also supplies a rare outsider's description of how a healer was expected to behave. He had thought the prophet "will surely come out to me, and stand, and call on the name of Yahweh his God, and wave his hand over the place, and recover the leper" (2 Kgs 5.11). Elisha refuses the performance and sends a message telling him to bathe.

Explicit theory is scarce. A proverb gives the rationale for sacred lots: "The lot is cast into the lap, But its every decision is from Yahweh" (Prov 16.33). Chance is God's channel, so the lot counts as no technique at all. The Greek-speaking writers of the last two centuries BCE add natural causes. Ben Sira, a Jerusalem scribe of about 180 BCE whose book survives here in his grandson's Greek, writes Κύριος ἔκτισεν ἐκ γῆς φάρμακα, καὶ ἀνὴρ φρόνιμος οὐ προσοχθίσει αὐτοῖς, "The Lord created medicines out of the earth, and a sensible man will not despise them" (Sir 38.4). He rereads Moses sweetening bitter water with wood at Marah (Exod 15.25) as a lesson in the "strength" of natural substances (Sir 38.5). The Greek word-family that names sorcery in the Exodus translation names, for him, a gift of God.

Where the line ran, and how it was punished

The oldest law is a single line in the Covenant Code of Exodus, a collection of disputed date often placed in the eighth or seventh century BCE: "You shall not allow a sorceress to live" (Exod 22.18). The Hebrew is three words, mĕkaššēpâ lōʾ tĕḥayyeh (Exod 22.17 WLC; the Hebrew numbers the verse one lower), and the first is a feminine participle: this is the only biblical law that names a practitioner in the feminine alone, and this verse later became a favourite proof-text of Europe's witch trials. The library's Greek reads Φαρμακοὺς οὐ περιποιήσετε, "you shall not preserve drug-makers" (Exod 22.18 LXX), turning one woman into a masculine plural. The Holiness Code of Leviticus, usually dated to the sixth century, names both sexes and sets the penalty.

Καὶ ἀνὴρ ἢ γυνὴ ἃς ἂν γένηται αὐτῶν ἐνγαστρίμυθος ἢ ἐπαοιδός, θανάτῳ θανατούσθωσαν ἀμφότεροι· λίθοις λιθοβολήσατε αὐτούς, ἔνοχοί εἰσιν.

A man or a woman among them who becomes a belly-talker or a chanter of spells: let both be put to death. Stone them with stones; they are guilty.

The client fares differently. Whoever turns to the mediums is "cut off from among his people" (Lev 20.6), which leaves the punishment to God. No biblical narrative shows a sorcerer stoned, though the rabbis later fixed stoning as the sorceress's penalty (see 'Witches and amulets in the Talmud' below). The stories show expulsion. Saul "put away" the mediums (1 Sam 28.3), and Josiah removed "those who had familiar spirits, and the wizards, and the teraphim" to fulfil "the words of the law which were written in the book that Hilkiah the priest found" (2 Kgs 23.24).

Deuteronomy gives the reason. The forbidden arts are the "abominations" of the nations Israel is to displace, and in their place God "will raise up to you a prophet from the midst of you, of your brothers, like me; to him you shall listen" (Deut 18.15). The prophet replaces the technician. The law even tests the prophet by results: "if the thing doesn’t follow, nor happen, that is the thing which Yahweh has not spoken" (Deut 18.22). It never applies that test to the diviners, whose offence lies in whom they consult.

So the same act can be holy in one place and abominable in another. Divination through the high priest's Urim is commanded; divination through a diviner is a capital crime. A curse drunk in water is law when a priest administers it. Objects could cross the line over time. Moses' bronze serpent survived in the temple until Hezekiah, in the late eighth century, "broke in pieces the brazen serpent that Moses had made; for to those days the children of Israel did burn incense to it" (2 Kgs 18.4). The labels were weapons among Israelites too. Ezekiel calls rival prophets of Yahweh men of "lying divination, who say, Yahweh says; but Yahweh has not sent them" (Ezek 13.6), and Samuel tells Saul that "rebellion is as the sin of witchcraft" (1 Sam 15.23), where the Hebrew word is qesem, divination (1 Sam 15.23 WLC).

The magicians of Pharaoh and Babylon

The Bible's favourite way of handling foreign magic is the court contest. Joseph interprets Pharaoh's dreams after the experts fail, Moses and Aaron face the magicians of Egypt, and Daniel outdoes the wise men of Babylon. In each story the foreign specialists have real powers, and each time they lose.

συνεκάλεσεν δὲ Φαραὼ τούς σοφιστὰς Αἰγύπτου καὶ τοὺς φαρμακούς· καὶ ἐποίησαν καὶ οἱ ἐπαοιδοὶ τῶν Αἰγυπτίων ταῖς φαρμακίοις αὐτῶν ὡσαύτως,

Pharaoh called together the sophists of Egypt and the drug-men, and the chanters of the Egyptians did the same with their sorceries.

The Greek, made in Egypt, gives Pharaoh's court the vocabulary of drugs and incantation that its readers knew. In the Hebrew the magicians do likewise bĕlahăṭêhem, and in the later plagues bĕlāṭêhem, which JPS renders each time "with their secret arts" (Exod 7.11 WLC; 7.22 WLC); as the WEB renders it, they "did in like manner with their enchantments" (Exod 7.11). Their staffs become snakes, though Aaron's swallows them, and they match the water turned to blood and the frogs (Exod 7.8–13, 7.22, 8.7). Nobody calls them frauds. At the third plague "The magicians tried with their enchantments to bring forth lice, but they couldn’t" (Exod 8.18), and they are the ones who name what they face.

Then the magicians said to Pharaoh, This is the finger of God: and Pharaoh’s heart was hardened, and he didn’t listen to them; as Yahweh had spoken.

The ḥarṭummîm said to Pharaoh, "This is the finger of a god." But Pharaoh's heart was hard and he did not listen to them, as Yahweh had said.

Their art copies wonders until a god takes over. The Babylonian tales of Daniel, stories of the third century BCE gathered into a book around 165 BCE, add a test designed to catch fraud. Nebuchadnezzar refuses to tell his experts the dream he wants explained.

But if you don't make known to me the dream, there is but one law for you; for you have prepared lying and corrupt words to speak before me, until the time be changed: therefore tell me the dream, and I shall know that you can show me the interpretation of it.

If you do not tell me the dream, there is one sentence for you all. You have agreed to tell me lies and deceit until the times change. So tell me the dream, and I will know that you can give me its meaning.

Anyone can invent a meaning; only real insight can recover the dream. The experts admit no one on earth can do it, and Daniel agrees, in Theodotion's Greek, that the mystery οὐκ ἔστιν σοφῶν, μάγων, ἐπαοιδῶν, γαζαρηνῶν ἀναγγεῖλαι, "is not for sages, magi, chanters or fate-reckoners to tell" (Dan 2.27 Th). Nebuchadnezzar puts Daniel over the wise men, and the queen later remembers him as "master of the magicians, enchanters, Chaldeans, and soothsayers" (Dan 5.11), head of the professions Deuteronomy bans. The book judges him by the source of his knowledge and ignores his title. The Old Greek translator went further and rendered the "magicians and enchanters" whom Daniel surpasses as τοὺς σοφιστὰς καὶ τοὺς φιλοσόφους, "the sophists and the philosophers" (Dan 1.20 OG). Second Isaiah, a prophet writing in Babylonia about 540 BCE, mocks its omen science to its face: "let now the astrologers, the star-gazers, the monthly prognosticators, stand up, and save you" (Isa 47.13). They will burn like stubble in a fire that "shall not be a coal to warm at" (Isa 47.14).

Doubt and mockery

Most of the Hebrew Bible's hostility says the forbidden arts are disloyal. Mockery and doubt form a smaller stream. On Mount Carmel the prophets of Baal "cut themselves after their manner with knives and lances" to bring down fire (1 Kgs 18.28), and Elijah taunts them at noon: "either he is musing, or he is gone aside, or he is on a journey, or peradventure he sleeps and must be awakened" (1 Kgs 18.27). Elijah only prays, and "the fire of Yahweh fell" (1 Kgs 18.38). After the exile a prophet says flatly that "the teraphim have spoken vanity, and the diviners have seen a lie" (Zech 10.2). Even the Philistine diviners build a test with chance as its alternative, yoking cows whose calves are shut up at home: if the cows walk away toward Israel the plague was God's, "but if not, then we shall know that it is not his hand that struck us; it was a chance that happened to us" (1 Sam 6.9).

Real scepticism arrives with Greek learning.

μαντεία καὶ οἰωνισμοὶ καὶ ἐνύπνια μάταιά ἐστι, καὶ ὡς ὠδῖν’ οὔσης φαντάζεταί σου ἡ καρδία.

Divinations and omen-takings and dreams are empty things, and your heart has fancies like a woman in labour.

Ben Sira allows one exception, a dream "sent from the Most High" (Sir 34.6 (stored as 31.6)). The Wisdom of Solomon, retelling the plague of darkness, says μαγικῆς δέ ἐμπαίγματα κατέκειτο τέχνης, "the tricks of the magic art lay prostrate" (Wis 17.7), and those who promised "to drive away terrors and disturbances" from the soul were sick with fear themselves (Wis 17.8). The Greek Isaiah writes an explanation of necromancy into the prophet's own words.

Ζητήσατε τοὺς ἐνγαστριμύθους καὶ τοὺς ἀπὸ τῆς γῆς φωνοῦντας, τοὺς κενολογοῦντας οἳ ἐκ τῆς κοιλίας φωνοῦσιν,

Seek out the belly-talkers and those who make voices come from the earth, the empty-talkers who speak out of their bellies.

Where the Hebrew had spirits chirping, the Greek has the medium throwing her own voice, and necromancy moves from forbidden to fake. The one fully worked exposure is the tale of Bel, an addition to Daniel that survives only in Greek. Bel's priests claim their idol eats the food set before it each night. Daniel has ashes sifted over the temple floor in the king's sight, and the doors are sealed. In the morning the seals are whole.

καὶ ἐγέλασεν Δανιὴλ, καὶ ἐκράτησεν τὸν βασιλέα τοῦ μὴ εἰσελθεῖν αὐτὸν ἔσω, καὶ εἶπ εν Ἴδε δὴ τὸ ἔδαφος καὶ γνῶθι τίνος τὰ ἴχνη ταῦτα.

Daniel laughed, held the king back from going in, and said, "Look at the floor, and see whose footprints these are."

The prints lead to the priests and their families, who show the king τὰς κρυπτὰς θύρας, "the hidden doors", by which they came in to eat (Bel 21). Daniel then kills Babylon's living serpent-god with cakes of pitch, fat and hair, with no prayer or spell (Bel 27). Chapter XIII follows these exposures further.

Change over time

The books are hard to date and were revised over centuries, so any sequence is a reconstruction. The patriarchal and early royal stories are relaxed: their heroes keep teraphim and divine with cups, and the narrators shrug. Eighth-century Isaiah counts the diviner among Jerusalem's leading men while attacking necromancy. In the seventh century the Deuteronomic reform turns those complaints into a system, defining Israel against "the nations" and blaming royal magic for national ruin.

The exile brought Judahites face to face with Babylonian divination in its homeland. After the return the priestly oracle was gone. Priests who could not prove their descent were barred from the most holy food "until there stood up a priest with Urim and with Thummim" (Ezra 2.63), and no later text in the library shows the Urim in use again; lots, dreams and angels filled the gap. The Hellenistic centuries brought Greek terms, doubt and natural explanation, while popular practice flourished. The prefatory letter of 2 Maccabees tells how "thick water", found where the temple fire had been hidden, was poured on the sacrifice and blazed up when "the sun, which had been clouded over, shone out" (2 Macc 1.22). Nehemiah and his companions called it Νεφθάρ and glossed it as "Purification" (2 Macc 1.36). A modern reader recognises naphtha, a petroleum that can ignite; the text presents a temple miracle.

Demons, almost absent from the Hebrew Bible, arrive with Tobit's Asmodeus and with the spirits of the giants in the Book of the Watchers (see 'The angels who taught sorcery' below), and fill the gospels. Baal-zebub, the god of Ekron whom King Ahaziah consulted about his injury (2 Kgs 1.2), returns in Mark as Beelzebul, "the ruler of the demons" (Mark 3.22). Chapter VII takes the story on from there.

What is distinctive here

Set beside the Egyptian and Mesopotamian material of chapters I and II, the biblical corpus is a literature about magic. Apart from Tobit's fumigation and the bitter water it preserves no recipes, and nothing like its neighbours' incantation series or spell papyri. It records a contest over authority. In the recurring court contest, from Joseph to Daniel, the sanctioned hero beats the foreign specialist by greater power, and exposes fraud only in the late Greek additions. Divination is banned while the high priest keeps his lots and Daniel heads the magicians, because the line follows allegiance and source and largely ignores technique. The practitioner the law most fears is a woman, and so is the most humane practitioner in the narratives. Demons and exorcism are nearly absent outside Tobit; in the Hebrew Bible the one spirit driven out is driven out by a lyre.

The Bible in this library is also layered. The Greek translators changed the witch's gender, made the medium a ventriloquist, turned Babylonian magi into philosophers and added that Samuel really answered Saul. Read side by side, the WEB and the Septuagint show one tradition arguing with itself over whether these arts were dangerous, foreign, fraudulent or simply real.

Chapter IV

Archaic and Classical Greece: drugs, songs and a borrowed word

Early Greek poetry treats drugs and charms as part of an ordinary world of gods and skills; between about 430 and 350 BCE Athenian doctors, orators and philosophers took a Persian priestly title, magos, made it an insult, and with it drew the first lines between medicine, religion and "magic".

Circe's cup

The first magician in Greek literature is a hostess. Odysseus' scouts reach a stone house on the island of Aiaia, where mountain wolves and lions, enchanted by evil drugs, fawn on them like dogs round a master. Inside, a woman is singing in a lovely voice (Hom. Od. 10.212–221). She seats the men, mixes cheese, barley meal and honey into wine, and adds something else.

οἴνῳ Πραμνείῳ ἐκύκα· ἀνέμισγε δὲ σίτῳ φάρμακα λύγρʼ, ἵνα πάγχυ λαθοίατο πατρίδος αἴης. αὐτὰρ ἐπεὶ δῶκέν τε καὶ ἔκπιον, αὐτίκʼ ἔπειτα ῥάβδῳ πεπληγυῖα κατὰ συφεοῖσιν ἐέργνυ. οἱ δὲ συῶν μὲν ἔχον κεφαλὰς φωνήν τε τρίχας τε καὶ δέμας, αὐτὰρ νοῦς ἦν ἔμπεδος

She stirred it up in Pramnian wine, and into the food she mixed baneful drugs, so that they would forget their homeland altogether. When she had served it and they had drunk, she struck them at once with her wand and shut them in the pigsties. They had the heads of pigs, the voice, the bristles and the shape, but their minds stayed as they had been.

Several things in this scene set the pattern for what follows. Circe uses a pharmakon, a substance, and a wand (rhabdos). Homer gives her no incantation. She is a goddess, the daughter of the Sun, so the scene is no picture of an ordinary human witch. And the defence against her is another substance. Hermes pulls a plant from the ground for Odysseus.

ὣς ἄρα φωνήσας πόρε φάρμακον ἀργεϊφόντης ἐκ γαίης ἐρύσας, καί μοι φύσιν αὐτοῦ ἔδειξε. ῥίζῃ μὲν μέλαν ἔσκε, γάλακτι δὲ εἴκελον ἄνθος· μῶλυ δέ μιν καλέουσι θεοί· χαλεπὸν δέ τʼ ὀρύσσειν ἀνδράσι γε θνητοῖσι

With these words the Slayer of Argos handed me the drug, drawing it out of the earth, and showed me what it was like. Its root was black and its flower like milk. The gods call it moly. It is hard for mortal men to dig up.

In the story moly works. Circe's drink fails to enchant Odysseus, he draws his sword as Hermes told him, and she swears not to harm him. She then turns the men back by rubbing a second drug on each of them, and they come out of the sties younger and handsomer than before (Hom. Od. 10.388–396). The same art that degrades also restores. A black-figure cup in Boston, painted about 550–525 BCE, shows Circe stirring a cup while the men around her sprout the heads of a boar, a ram, a wolf and a lion, and Odysseus comes at her with his sword (Boston, MFA 99.518).

This chapter runs from Homer (late eighth or early seventh century BCE) to the generation after Plato (about 300 BCE), with a glance at the early Hellenistic poets. Over those four centuries the Greeks took a foreign priest's title, made it the word from which "magic" descends, and turned it into an accusation.

The Greeks' own words

Homer has no word for magic. He has a word for the substance, pharmakon, which means drug, remedy, poison, dye or charm according to context, and he says plainly that it can go either way. Helen's grief-killing drug came from Egypt, where the earth bears "many good drugs mixed, and many baneful" (Hom. Od. 4.219–232). He has a verb for the effect on the mind, thelgein, "to enchant" or "to spellbind", which is used of drugs, of Hermes' wand closing men's eyes in sleep (Hom. Od. 24.1–5), of love and of song. And he has, once, a word for a spoken charm, epaoidē, literally "a song sung over" something. The table gives the words that matter most in this chapter.

None of these words divides "religion" from "magic" the way modern English does. A purification (katharmos) could be a civic duty or a racket. An incantation could stop a wound bleeding or bind an enemy. The Greeks drew their lines with other questions. Was the act open or hidden? Did it help or harm? Was it done by a citizen or a foreigner, for the community or for a fee?

A Persian word turned against Greeks

The word that eventually gave Europe "magic" arrives in Greek as an ethnic and professional label. Aeschylus' Persians (472 BCE) can still list "Magos the Arab" among the dead as a personal name (Aesch. Pers. 318). Herodotus, writing around 430 BCE, describes the Magi as a Median tribe and as the priests of the Persians, without any hint of disapproval. No Persian sacrifice could be performed without one.

Μάγος ἀνὴρ παρεστεὼς ἐπαείδει θεογονίην, οἵην δὴ ἐκεῖνοι λέγουσι εἶναι τὴν ἐπαοιδήν· ἄνευ γὰρ δὴ Μάγου οὔ σφι νόμος ἐστὶ θυσίας ποιέεσθαι.

A Magus stands beside it and chants a birth of the gods over it, which is what they say the chant consists of. It is not their custom to sacrifice without a Magus.

Herodotus' Magi read dreams for kings (Hdt. 1.107.1), and "worked charms" on the river Strymon by sacrificing white horses, where he uses the verb pharmakeuein for a ritual with no drug in it (Hdt. 7.113.2–114.2). When a storm wrecks Xerxes' fleet off Magnesia they chant to the wind and sacrifice to Thetis, and it stops on the fourth day, "or else it somehow dropped of its own accord" (Hdt. 7.191.2). Xenophon's Magi are court priests (Xen. Cyr. 8.1.23), and the Platonic Alcibiades I, whose authenticity is disputed, defines the "mageia of Zoroaster" as "the service of the gods" ([Pl.] Alc. 1, 121e–122a).

Inside Greece the same word was already a weapon. In Sophocles' Oedipus the King, probably staged about 429 BCE, Oedipus turns on the blind seer Teiresias.

ὑφεὶς μάγον τοιόνδε μηχανορράφον, δόλιον ἀγύρτην, ὅστις ἐν τοῖς κέρδεσιν μόνον δέδορκε, τὴν τέχνην δʼ ἔφυ τυφλός.

…having slipped in this plot-stitching magos, this crafty begging-priest, who can see only where there is profit and was born blind in his own art.

The irony is the point of the scene. The man called a mercenary magos is the only one telling the truth. Soon the label had a theory attached to it. Gorgias of Leontini, in his showpiece Encomium of Helen (undated, but usually put in the last decades of the fifth century), set the two words side by side as names of arts.

συγγιγνομένη γὰρ τῇ δόξῃ τῆς ψυχῆς ἡ δύναμις τῆς ἐπῳδῆς ἔθελξε καὶ ἔπεισε καὶ μετέστησεν αὐτὴν γοητείᾳ. γοητείας δὲ καὶ μαγείας δισσαὶ τέχναι ηὕρηνται, αἵ εἰσι ψυχῆς ἁμαρτήματα καὶ δόξης ἀπατήματα.

The power of the incantation, meeting the soul's opinion, charms it, persuades it and changes it by sorcery. Two arts of sorcery and magic have been found, and they are errors of the soul and deceptions of opinion.

A few paragraphs later Gorgias makes speech itself a pharmakon of the soul, able to heal or to kill (Gorg. Hel. 14). By the 340s goēs, sometimes paired with magos, was ordinary courtroom abuse for a slippery speaker, and Aeschines threw it at Demosthenes, while Demosthenes used it of Philip (Aeschin. 3.137; Dem. 19.109).

Who practised it

The texts name a wide cast, and they divide roughly into five groups.

• Goddesses and heroines. Circe; Aphrodite, whose embroidered band holds "all enchantments" (Hom. Il. 14.214–221); Agamede, "who knew all the drugs the broad earth nourishes" (Hom. Il. 11.740–741); Helen, whose drug was a gift from an Egyptian woman, Polydamna; and Medea.
• Men with recognised skill. The sons of Autolycus with their bleeding charm, Asclepius and his sons, and seers such as Teiresias.
• Freelance specialists. Begging priests and seers who worked at rich men's doors for fees (Pl. Resp. 364b–c), some performing their rites from books of Orpheus and Musaeus (Pl. Resp. 364e–365a), along with purifiers and the "Orpheus-initiators" whom Theophrastus' superstitious man visits every month (Theophr. Char. 16).
• Ordinary women. Midwives, nurses, "old women who chant", the Thessalian moon-pullers, and the courtesans and concubines of the love-charm stories.
• Foreigners. Persian Magi, Thracian healers, the Neuri of the far north, the Lydian "sorcerer-enchanter" that Pentheus takes the disguised Dionysus to be (Eur. Bacch. 233–238), and the people beyond the Libyan desert whom Herodotus reports were "all goētes" (Hdt. 2.33).

The word field leans female. Euripides' Medea, about to kill, says that poison is "the straight road, in which we women are by nature most skilled" (Eur. Med. 384–398), and Aristotle can mention "the pharmakides" who collect a substance from horses as a known type of woman (Arist. HA 6.22, 577a). Practice was more mixed than the stereotype. The men in the record are healers, seers and the purifiers whom doctors attacked, and the necromancers in Aeschylus are a chorus of old Persian men.

The vases add the person the literature hardly mentions, the mother or nurse who protects a child. An Athenian scene of about 440–430 BCE shows a mother handing her baby to a Thracian slave nurse. The museum's catalogue reads the string of charms across the infant's chest as protection against sickness (Harvard 1960.342). On another vase a woman at a washbasin wears a thread round her thigh that the catalogue reads as an amulet (Munich 2411). Painters showed such things without comment, as part of ordinary life.

What they did

Healing with songs, drugs and amulets

The oldest Greek spoken charm is a first-aid measure. When a boar gashes the young Odysseus on Parnassus, his uncles treat him.

ὠτειλὴν δʼ Ὀδυσῆος ἀμύμονος ἀντιθέοιο δῆσαν ἐπισταμένως, ἐπαοιδῇ δʼ αἷμα κελαινὸν ἔσχεθον

They bound up the wound of blameless, godlike Odysseus with skill, and with an incantation they held back the dark blood.

Bandage and song stand in one sentence as parts of one competent treatment. In the Iliad Machaon treats Menelaus with no charm at all, sucking out the blood and sprinkling on soothing drugs (Hom. Il. 4.217–219). Pindar, in an ode whose date is disputed (about 474 BCE), lists the hero Asclepius' whole repertoire.

τοὺς μὲν μαλακαῖς ἐπαοιδαῖς ἀμφέπων, τοὺς δὲ προσανέα πίνοντας, ἢ γυίοις περάπτων πάντοθεν φάρμακα, τοὺς δὲ τομαῖς ἔστασεν ὀρθούς.

Some he tended with gentle incantations, others by giving soothing drinks, or he fastened remedies all round their limbs, and others he set on their feet by the knife.

Charm, drink, amulet and surgery form a single toolkit here. A century later Plato could still list drugs, cautery, surgery, incantations and amulets together as treatments that will do an intemperate invalid no good until he changes his life (Pl. Resp. 426a–b), and he describes midwives who, "by giving little drugs and singing charms", bring on or ease labour and procure miscarriage (Pl. Tht. 149c–d). The toolkit had its doubters long before Plato. Sophocles' Ajax, in a play probably of the mid-fifth century, says that a wise doctor does not "wail incantations over a sore that needs the knife" (Soph. Aj. 581–582).

Healing by the gods themselves was public cult. A relief of about 430–400 BCE in Athens shows a priest leading a patient to bed in the god's sanctuary, and Asclepius appearing twice, laying his hand on the sleeper's head (Athens, NM 1841). To the Greeks this temple sleep (enkoimēsis) was religion and medicine together, and none of the classical texts surveyed here calls it mageia.

Love: philtra, wheels and poison

Greek love-magic is told mostly from a woman's side, and its story keeps ending in a death. Pindar's version of the Argonaut myth reverses that. Aphrodite brings the iynx down from Olympus, "the maddening bird" bound to a four-spoked wheel, and teaches Jason "suppliant incantations" so that Medea will lose her reverence for her parents (Pind. Pyth. 4.213–223). The future witch is first the victim. South Italian vases of about 350–330 BCE put the wheel in the hands of Aphrodite and Eros, once at the triumph and marriage of Andromeda (Malibu 86.AE.680; Malibu 87.AE.23).

Tragedy shows the danger. In Sophocles' Women of Trachis Deianira, displaced by a younger captive, remembers the dying centaur Nessus' gift. His blood, he said, would be "a charm for the heart of Heracles" (Soph. Trach. 555–581). She draws a line before she uses it.

κακὰς δὲ τόλμας μήτʼ ἐπισταίμην ἐγὼ μήτʼ ἐκμάθοιμι, τάς τε τολμώσας στυγῶ· φίλτροις δʼ ἐάν πως τήνδʼ ὑπερβαλώμεθα τὴν παῖδα καὶ θέλκτροισι τοῖς ἐφʼ Ἡρακλεῖ,

Wicked boldness I hope I may never know or learn, and I loathe the women who are bold that way. But if somehow, with love-charms and enchantments worked on Heracles, I can get the better of this girl…

The robe she smears with the blood eats into Heracles' flesh. The tuft of wool she used dissolves in sunlight into a boiling froth (Soph. Trach. 672–706). The distinction she trusts, between harmful drugs and a wife's love-charm, is real to her and impossible to apply. Phaedra's nurse in Euripides' Hippolytus has "love-charms that enchant" in the house and needs only a token from the young man, a word or something taken from his clothes (Eur. Hipp. 478–517).

The courts saw the same plot. In Antiphon's speech Against the Stepmother (420s BCE) a concubine poured a "love-drug" into a libation to win back her lover. He died at once and the speaker's father within twenty days, and she was tortured and executed. The speaker adds that his father had once caught the stepmother herself preparing drugs for him, and that she "claimed she gave it not to kill but as a love-potion" (Antiph. 1.9). A Peripatetic treatise, perhaps post-Aristotelian, remembers a woman acquitted on the Areopagus because her fatal love-potion was given without intent to kill ([Arist.] MM 1.16).

Binding and harming

Greek archaeology knows hundreds of lead curse tablets, and the library now holds those of Attica and Sicily (see 'What the binders wrote' below). The literary texts name the practice too. Aeschylus' Furies dance round Orestes and sing a "binding hymn", "binder of wits, lyreless, withering for mortals" (Aesch. Eum. 306–396). A pseudo-Homeric poem of uncertain date, the Kiln, is a working threat. If the potters cheat the singer, he will call up the kiln-wreckers Smasher, Crasher, Unquenchable, Shatterer and Raw-Breaker, and "Circe of many drugs" to throw wild drugs on the firing ([Hom.] Epigr. 14). Plato supplies the technical names and the market.

ἐάν τέ τινα ἐχθρὸν πημῆναι ἐθέλῃ, μετὰ σμικρῶν δαπανῶν ὁμοίως δίκαιον ἀδίκῳ βλάψει ἐπαγωγαῖς τισιν καὶ καταδέσμοις, τοὺς θεούς, ὥς φασιν, πείθοντές σφισιν ὑπηρετεῖν.

And if he wants to hurt an enemy, for a small outlay he can harm just and unjust alike with certain attraction spells and bindings, since they persuade the gods, so they say, to work for them.

In the Laws he adds the frightening detail of "wax images" set at doorways, at crossroads or on parents' tombs (Pl. Leg. 932e–933c).

Raising the dead

Circe gives Odysseus the most detailed ritual recipe in early Greek poetry: dig a pit a cubit square, pour honey-milk, then wine, then water for all the dead, sprinkle white barley, and sacrifice a ram and a black ewe (Hom. Od. 10.508–540). In the poem it works. The souls gather out of Erebus, and Teiresias explains the rule that whichever of the dead drinks the blood "will tell you the truth" (Hom. Od. 11.23–50, 95–99, 145–149). Homer presents this as a heroic consultation and does not condemn it. In Aeschylus' Persians the queen pours libations at the tomb of the dead king Darius while the elders sing "soul-leading laments", and the ghost rises to complain that the gods below "are better at taking than letting go" (Aesch. Pers. 681–693). By the late fifth century the "soul-leader" (psychagōgos) was a suspect trade. Euripides' Heracles denies being one (Eur. Alc. 1127–1128), Aristophanes casts Socrates as one (Ar. Av. 1553–1564), and in the fourth century Plato makes it a crime (Pl. Leg. 909a–d).

The moon, the weather and the sky

The signature feat of the Greek witch was to draw the moon down from the sky, and it was attached to one region. In Aristophanes' Clouds (423 BCE) the debtor Strepsiades, whose interest falls due at the new moon, has an idea. "What if I bought a Thessalian witch and drew the moon down by night, and shut it up in a round box like a mirror?" (Ar. Nub. 749–757). It is among the earliest explicit mentions of the feat, and the joke works only if the audience already knew the belief. Socrates in the Gorgias alludes to "the Thessalian women who draw down the moon" and the price they pay in what is dearest to them (Pl. Grg. 513a). A later Peripatetic collection of marvels has a Thessalian woman draw a circle, set down her drugs, step inside with her son, and imitate a deadly snake's voice until it crawls into the circle and shrivels ([Arist.] Mir. 151).

Against the evil eye

Later Greek writers explain that envy could harm through the eyes, and modern scholars read the eyes that Archaic painters put on cups and ships as an answer in kind. Athenian "eye-cups" of the 530s and 520s BCE become a staring mask over the drinker's face when he raises them (Munich 2044). On one cup the catalogue reads the teardrop corner of each eye as a copy of the Egyptian Eye of Horus, a combination it says does not recur on later Greek vases (Champaign 70.8.1). On a vase in Boston a male siren whose whole body is a single eye shares the vase with masturbating revellers and defecating dogs (Boston 10.651), so the protective eye and obscene joking could sit side by side. The earliest text about protecting a baby from such harm is in the Homeric Hymn to Demeter (seventh or sixth century BCE), where the goddess, disguised as a nurse, promises a "counter-cut" and a "safeguard" against the "attack" that might harm the baby. The words are riddling, and their exact sense is disputed (Hymn. Hom. Dem. 227–230).

How it was supposed to work

No Greek text before Plato sets out a full theory of magic, but the stories assume several, and they overlap.

1. Substances have powers. Moly, Helen's Egyptian drug and Nessus' blood act by what they are. Some need a trigger. Deianira was told to keep the blood from fire and the sun, and heat set it off.
2. Like acts on like. A wheel spins so the lover will circle the door, and wax melts so he will melt. Something of the target, such as a word or a piece of his clothing, makes the link.
3. Gods and daimones do the work. The practitioner calls on Hecate, the Moon or Aphrodite, or claims to persuade the gods to serve, and they act. The Classical binders' own tablets call above all on Hermes, and none names the Moon or Aphrodite (see 'What the binders wrote' above).
4. Song is a force. The same verb, thelgein, covers a drug, a wand, a poem and a persuasive speech. Lament (goos) draws up the dead.
5. Belief does it. This was the critics' theory, discussed below.

Plato gave the first explicit account of the third idea. In the Symposium the priestess Diotima explains that "god does not mingle with man", so all traffic between them passes through the daimonic realm, "all divination and the art of priests concerned with sacrifices, initiations, incantations, and all prophecy and sorcery" (Pl. Symp. 202e–203a). Sorcery sits in that list without blame, and in the same passage Eros is "a clever goēs and pharmakeus and sophist".

Where the line was drawn

The doctors' attack

The first sustained attack on magic in Greek comes from medicine. The Hippocratic treatise On the Sacred Disease, written about 420–400 BCE, argues that epilepsy is no more divine than any other illness, and names the people who said otherwise.

ἐμοὶ δὲ δοκέουσιν οἱ πρῶτοι τοῦτο τὸ νόσημα ἀφιερώσαντες τοιοῦτοι εἶναι ἄνθρωποι οἷοι καὶ νῦν εἰσι μάγοι τε καὶ καθάρται καὶ ἀγύρται καὶ ἀλαζόνες, ὁκόσοι δὴ προσποιέονται σφόδρα θεοσεβέες εἶναι καὶ πλέον τι εἰδέναι.

In my view the first people to make this disease sacred were the sort of men we still have today, magoi, purifiers, begging priests and impostors, who pretend to be very god-fearing and to know more than other people.

The author gives a detailed picture of what his rivals did. They diagnosed by symptom. A patient who bleated had been struck by the Mother of the Gods, one who neighed by Poseidon, and night terrors were "attacks of Hecate and assaults of heroes". They treated with "purifications and incantations", washed the patient with blood as if he were polluted, and buried or threw into the sea what they had wiped off him (Hippoc. Morb. Sacr. 4 (Jones)).

His central objection is religious. If a man, "by magic and sacrifice, will bring down the moon, make the sun vanish, and produce storm and fair weather", then those things are human and not divine, and the men who claim such powers must believe that the gods either do not exist or have no strength (Hippoc. Morb. Sacr. 3–4 (Jones)). The physician presents himself as the truly pious party. His closing chapter keeps the divine and spreads it through nature. Cold, sun and winds are divine, so "all are divine and all are human" (Hippoc. Morb. Sacr. 21 (Jones)). Medicine was no purely secular party in this quarrel. The Hippocratic Regimen tells patients that prayer is "fitting and very good" so long as they also help themselves, and its dream-cures include prayers to the averting gods (Hippoc. Vict. 4.87).

Plato's laws

Plato's last work, the Laws (about 350 BCE), turns the objection into statute. Its harshest treatment of the impious falls on the men who claim "to lead up the souls of the dead" and promise "to persuade the gods" with sacrifices, prayers and incantations, ruining individuals, households and cities for money. They are to be imprisoned for life in a remote prison and denied burial (Pl. Leg. 909a–d). In the law on harmful drugs Plato divides pharmakeia in two. One kind harms bodies by bodies, as poison does. The other is described in this passage.

ἄλλη δὲ ἣ μαγγανείαις τέ τισιν καὶ ἐπῳδαῖς καὶ καταδέσεσι λεγομέναις πείθει τοὺς μὲν τολμῶντας βλάπτειν αὐτούς, ὡς δύνανται τὸ τοιοῦτον, τοὺς δʼ ὡς παντὸς μᾶλλον ὑπὸ τούτων δυναμένων γοητεύειν βλάπτονται.

The other, by trickeries, incantations and what are called bindings, persuades those who dare to do harm that they are able to, and persuades their victims that they really are being harmed by people with the power to bewitch.

This is a psychological theory, close to what we would call a nocebo effect, and Plato does not try to settle whether the spells also work in some other way. He advises the lawgiver not to argue people out of their fear of wax figures but simply to forbid the practice. The penalty depends on expertise. A doctor who poisons dies, and so does a seer or reader of portents who "seems like one who harms by bindings or attraction-spells or incantations"; a layman's penalty is left to the court (Pl. Leg. 933d–e). Just before this law Plato affirms that parents' curses "have been fulfilled for many people, many times" (Pl. Leg. 931b–932a). In Plato's scheme the power to curse is legitimate inside the family and criminal when it is hired.

Real trials

No surviving Athenian law names "magic" as such, and the charges in the surviving cases are not always stated. Still, women were put to death. A speech preserved under Demosthenes' name, probably not his, reminds the jury that they executed "the foul Theoris, the Lemnian, the pharmakis, both her and all her family", and accuses the defendant's twin brother of taking her drugs and incantations from the slave who informed against her and now claiming to cure epileptics ([Dem.] 25.79–80). Demosthenes himself alludes to a priestess executed for running the private cult groups that Aeschines' mother led (Dem. 19.281). In Plato's Meno the charge even touches philosophy. Meno tells Socrates that he "bewitches, drugs and chants spells" over him, and that in another city he might be arrested as a goēs (Pl. Meno 80a–b).

The limit set by the gods

Before there were laws there were limits. In the Odyssey a man refuses to give Odysseus arrow poison, "for he feared the indignation of the gods" (Hom. Od. 1.259–264). The hard limit was death. Pindar's Asclepius, bribed with gold, raised a dead man, and Zeus struck both with a thunderbolt (Pind. Pyth. 3.47–58). Aeschylus' Apollo says that his father Zeus "made no charms" against death (Aesch. Eum. 645–651), and a chorus of Euripides finds nothing stronger than Necessity, no remedy "in the Thracian tablets that the voice of Orpheus wrote down" (Eur. Alc. 962–972). That last passage suggests that an Athenian audience in 438 BCE already knew of written remedies ascribed to Orpheus.

Doubters and mockers

Doubt in classical Greece took several forms. Herodotus records and declines to believe. The Scythians swear that every one of the Neuri turns into a wolf for a few days each year, "but those who say this do not persuade me" (Hdt. 4.105.1–2). Thucydides is drier. In the plague, prayers and oracles were as useless as medicine and were finally abandoned (Thuc. 2.47.4), and Nicias, "somewhat too much given to divination", insisted on waiting "thrice nine days" after an eclipse, a delay that left the Athenian army at Syracuse to be destroyed (Thuc. 7.50.4).

Comedy aimed at the people who ran the rites. In Aristophanes' Wealth (388 BCE) a slave spending the night in Asclepius' temple watches the priest sweep the cakes and figs off the holy table "into a sack" (Ar. Plut. 653–747). Then the god himself mixes a stinging eye-plaster of garlic, fig-juice, squill and vinegar, all real irritants, and smears it on a politician's eyelids to make them hurt more.

Plato's critique had the widest reach. He calls optical illusion and stage conjuring goēteia, sorcery that preys on a weakness of our nature, and prescribes measuring and counting as the cure (Pl. Resp. 602d). Yet Plato kept the vocabulary for himself. Philosophical arguments are "charms" to sing to the frightened child within us (Pl. Phd. 77e–78a), and the Thracian healer's charm turns out to be "beautiful words" that produce self-control in the soul (Pl. Chrm. 156d–158b).

Aristotle and his pupil Theophrastus sorted the claims one by one. Aristotle accepts that hippomanes, the substance from horses, exists, but says the tales about it "have been invented by women and by the people concerned with incantations" (Arist. HA 8.24, 605a). Theophrastus, writing on plants about 300 BCE, records the root-cutters' rules without endorsing them. They draw three circles round the mandrake with a sword, cut it facing west, and have a companion dance round it talking about sex (Theophr. HP 9.8.7–8). He accepts plants that act on the mind and draws his line at symbols, saying that the claims about "amulets and generally about the so-called protective charms for bodies and houses" are "sillier and less credible" (Theophr. HP 9.19.2–3). In his Characters the superstitious man, whose deisidaimonia is "cowardice towards the divine", purifies his house because he thinks Hecate has been sent against it (Theophr. Char. 16). Theophrastus also told, in a passage that survives only as Plutarch quotes it, how the dying Pericles showed a friend the amulet "the women" had hung round his neck, to prove how ill he was if he put up with such folly. For an elite man the amulet was women's business (Theophr. fr. 146).

Change over time

Three changes stand out in the library's evidence.

The words. Homer has pharmakon, thelgein and a single epaoidē. The words magos, goēs, mageuma, psychagōgos and agyrtēs appear in fifth-century Athenian drama, mostly as abuse. Theseus calls his chaste son a goēs (Eur. Hipp. 1038–1040), and an old man in Euripides' Suppliants says he hates those who stretch out life with diets, drinks and mageumata, "magic practices" (Eur. Supp. 1108–1113). In Euripides' Iphigenia among the Taurians the heroine fakes a purification by chanting "barbarian songs, playing the magus" (Eur. IT 1331–1338). Between about 420 and 400 BCE the Hippocratic polemic and Gorgias turned the word into a category. The fourth century put it into law and invective.

Hecate and Medea. Hesiod's long hymn to Hecate (about 700 BCE) contains no magic at all. She is a goddess of kings, horsemen, fishermen and children, approached "by fine sacrifices according to custom" (Hes. Theog. 411–452). Coins of the Thessalian city of Pherae in the 360s BCE still show her as a civic patron (Dewing 1418). Euripides' Medea in 431 BCE is the earliest figure in the library who keeps her at the hearth as a patron of poison (Eur. Med. 384–398). Medea herself changes in the same way. In Pindar she is a princess who gives Jason a protective oil (Pind. Pyth. 4.213–223). Vases of about 510–470 BCE show her scattering herbs over a cauldron from which a ram leaps out, rejuvenated (London E 163; Harvard 1960.315). In Euripides she is a killer, and her poisoned robe and crown burst into flame on the king's daughter and kill the king who embraces her (Eur. Med. 784–789, 1156–1221).

The full portrait. In the early Hellenistic poets, about two generations after Plato, the picture is complete. Theocritus' second Idyll, titled The Sorceresses (about 275 BCE), shows an ordinary young woman of the city, Simaetha, working a night rite against the lover who left her. She invokes the Moon and Hecate "of the underworld", burns barley and bay, and whirls a bronze rhombos, a spinning disc or bull-roarer whose exact form is disputed.

ὡς τοῦτον τὸν κηρὸν ἐγὼ σὺν δαίμονι τάκω, ὣς τάκοιθʼ ὑπʼ ἔρωτος ὁ Μύνδιος αὐτίκα Δέλφις. χὡς δινεῖθʼ ὅδε ῥόμβος ὁ χάλκεος ἐξ Ἀφροδίτας, ὣς τῆνος δινοῖτο ποθʼ ἁμετέραισι θύραισιν.

As I melt this wax with the goddess's help, so may Myndian Delphis melt with love at once. And as this bronze rhombos whirls by Aphrodite's power, so may he whirl about my door.

She has already gone the round of "every old woman who knew how to chant charms" (Theoc. Id. 2.85–92), and she keeps deadly drugs in a box, learned "from an Assyrian stranger", in case the love-charms fail (Theoc. Id. 2.159–162). In Apollonius' Argonautica (about 250 BCE) Medea, taught by Hecate, can "bind the paths of the stars and the holy moon" (Ap. Rhod. Argon. 3.528–533). The Thessalian moon-witch of Aristophanes' joke had become the witch of literature. Chapters V and VI follow her into the Hellenistic and Roman world.

What is distinctive here

Set beside the other cultures in this report, archaic and classical Greece stands out in four ways.

First, Greek "magic" is tied to substances. The key word, pharmakon, covered the doctor's remedy, the poisoner's dose, the love-charm and the state's hemlock. Small bottles that probably held the hemlock for executions were found in the building identified as the Athenian state prison (Athens, State Prison). That is why Plato's law on magic is part of a law on drugs, and why the classic Greek plot is the love-charm that kills.

Second, the culture that wrote about it was mainly a culture of critics. Egyptian and Mesopotamian texts in this library often preserve the spell itself, spoken from the practitioner's side. Greek literature of this period almost never does. Apart from the Kiln curse and the refrains of Theocritus, it shows the practice through a doctor attacking rivals, a philosopher writing laws, an orator abusing an opponent and a comic poet looking for a laugh. Only the lead tablets, and one curse on papyrus from Memphis (PGM XL), keep the binders' own words, and they bear the critics out on what was done more than on who did it and why (see 'What the binders wrote').

Third, the category itself was made here, out of another people's religion. A Persian priestly title became, within one or two generations, the name for a Greek fraud. In the same decades the same writers kept the old words of song and charm and gave them to philosophy, rhetoric and poetry. Gorgias called speech a drug, and Plato called argument a charm.

Fourth, the Greeks offered, at least in outline, the theory a modern reader reaches for first. Gorgias put the power of incantation in "the soul's opinion", and Plato put the power of binding spells in the belief of the person who casts them and the fear of the victim. Chapter XII starts from that explanation when it asks whether any of it could have worked.

Chapter V

The Hellenistic and Imperial Greek world

Between Polybius and Proclus, Greek writers described magic, sold it, tested it, mocked it and finally turned part of it into philosophy, and they argued at length about where the line ran between magic and legitimate knowledge.

ἐπὶ τὸν νεκρὸν τοῦ παιδὸς προκύψασα καί τινα πρὸς τὸ οὖς ἐπᾴδουσα ἐξήγειρέ τε καὶ ὀρθὸν ἑστάναι τῇ μαγγανείᾳ κατηνάγκαζεν.

She bent over her son's body, chanted something into his ear, woke him, and with her sorcery forced him to stand up straight.

The night after a battle near the Egyptian village of Bessa, an old woman goes out among the dead on the third night after the full moon. She digs a pit, lights a fire, pours honey, milk and wine, throws in a man-shaped figure of dough crowned with laurel and fennel, cuts her own arm, and calls on the moon with names nobody understands. Her son's body rises and speaks. He rebukes her for forcing the laws of the Fates, foretells her violent death, and falls back. Soon afterwards she runs onto a spear and dies. The Egyptian priest Calasiris, close by, refuses to take part (Heliod. Aeth. 6.14–15). The scene comes from Heliodorus' Aethiopica, a novel of disputed date (third or fourth century CE). It is fiction, yet nearly every element in it turns up elsewhere in this library in sources that claim to be fact: the pit and libations, the effigy, the unintelligible names, the charge of forcing fate, the priest who keeps his distance, and the bad end that awaits the practitioner.

This chapter runs from the Hellenistic historians of the second century BCE to the Neoplatonist philosophers of late antiquity. Its authors lived under Hellenistic kings and then Roman emperors, wrote in Greek, and inherited the vocabulary of chapter IV. The library documents them well, with 106 verified passages of imperial Greek prose and some thirty of pagan philosophy, and these writers did something earlier Greeks rarely did. They theorised about magic, and a few of them tested it.

The words and the categories

Imperial Greek had no single word that matches the modern "magic". It had a cluster of overlapping terms, each with its own history and tone, and authors moved between them deliberately.

The most revealing of these is μάγος. Greek authors knew it was a Persian priestly title. Strabo watched magi in Cappadocia chant for about an hour a day before an unquenchable fire, holding a bundle of rods. "These things we have seen," he says (Strabo 15.3.15). Dio Chrysostom says the Persians call magi "men skilled in serving the divine", and that the Greeks, "through ignorance of the word", apply it to goētes (Dio Chrys. Or. 36.41). Diogenes Laertius presents them as philosophers (Diog. Laert. 1.6–9). Philostratus puts the popular Greek sense most sharply. People who call Apollonius a magos, he says, slander him as "wise by force" (βιαίως σοφός) (Philostr. VA 1.2).

The commonest word in practice is γόης, with 255 occurrences across the imperial authors surveyed, a raw count that may include some forms of γοάω, "to lament". A goēs could be a spell-caster, a seductive orator, a quack or a political agitator; Josephus uses it for the men who led crowds to the Jordan or into the desert promising signs of deliverance (Joseph. AJ 20.97; BJ 2.259–263). The word records the speaker's judgement more than any technique.

Who practised magic

The cast is wide, and most of it is foreign. Persian magi light fires without fire in Lydian sanctuaries (Paus. 5.27.5–6) and open the gates of Hades for Lucian's Menippus (Lucian, Nec. 6–10). Babylonian "Chaldaeans" appear as a hereditary priestly caste of astronomers and diviners who avert evils "by purifications, sacrifices and incantations" (Diod. Sic. 2.29), which modern scholars link to the Babylonian namburbi rites of chapter II. A Chaldaean snake-charmer saves a dying slave in Lucian (Philops. 11–13). Egypt supplies the most prestigious experts: Pancrates, a sacred scribe of Memphis said to have learned magic from Isis during twenty-three years underground (Lucian, Philops. 34–36), the Egyptian woman who teaches Leucippe a charm against bee-stings (Ach. Tat. 2.7), and the marketplace performers "taught by Egyptians" whom the pagan philosopher Celsus despised (Orig. Cels. 1.68). Syrians exorcise and make love charms (Lucian, Philops. 16), and a Syrian prophetess, Martha, served Marius in a purple cloak (Plut. Mar. 17). Josephus says that "Arabian women are the most skilled in drugs of all women" (AJ 17.62–63). Within the Greek world, Thessalian women keep their old reputation (Plut. Conj. praec. 48).

Closer to home, wives were suspected of love potions (Conj. praec. 5), and so were Queen Olympias, whose snake cooled Philip's ardour (Plut. Alex. 2), and Aretaphila of Cyrene, tortured for preparing a potion (Plut. Mul. virt. 19). Itinerant professionals included beggar-priests and purifying old women (Plut. De superst. 3, 12), the market diviners Artemidorus spent years among (Artem. 1 pr.), and fate-tellers who seek out people in trouble (Alex. Aphr. Mantissa 24). Holy men and intellectuals were accused of magic: Apollonius of Tyana, the physician Galen, and the sophist Dionysius of Miletus, whose students' memories were credited to "Chaldaean arts" (Philostr. VS 1.22). Last come civic priests and states that did what would elsewhere be called goēteia, and were never called goētes for it.

What they did

Love and binding

Erotic magic is the practice these texts describe most concretely, usually from a distance. Lucian's courtesans recommend a Syrian drug-woman who charges a drachma and a loaf, plus sulphur, a torch and wine. She needs something belonging to the man, a cloak, a shoe or a few hairs, which she fumigates with sulphur and salt while saying both names.

εἶτα ἐκ τοῦ κόλπου προκομίσασα ῥόμβου ἐπιστρέφει ἐπῳδήν τινα λέγουσα ἐπιτρόχῳ τῇ γλώττῃ, βαρβαρικὰ καὶ φρικώδη ὀνόματα.

Then she takes a bull-roarer from the fold of her dress and spins it, rattling off an incantation, foreign and terrifying names.

A hate-charm against a rival goes with it: tread on her footprint, left foot on right. In Lucian's Philopseudes ("The Lover of Lies") a Hyperborean magos, paid four minae down and sixteen on success, sends a clay Eros flying to fetch a married woman (Philops. 13–15). Galen names the genres: love potions (φίλτρα), attraction spells (ἀγώγιμα, literally "leaders"), dream-senders and hate-spells. He refuses to record them, and he calls "ridiculous" the spells for binding legal opponents "so that they cannot utter a word in court" (Gal. SMT 10.1). That is the genre of the lead curse tablets (κατάδεσμοι, Plato's word, which the tablets themselves never use), which survive by the hundred. On the Attic tablets the tongue is the part bound most often (chapter IV), and a Latin tablet of the early second century from Upper Germany asks the Manes below that a man's adversary be struck mute when he comes before the consularis, probably the governor (EDH HD004630).

Love potions were also understood as drugs with doses. In Achilles Tatius' novel a potion given undiluted drives Leucippe mad for ten days (Ach. Tat. 4.15), and Josephus reports a "love potion" that was really poison (AJ 17.62). Plutarch warns wives that love won by drugs is like fish caught with poison, quick to land and spoiled to eat (Conj. praec. 5).

Healing, amulets and the doctors

Medical handbooks record folk healing. Dioscorides, the first-century pharmacologist, keeps magical uses at arm's length by putting them in indirect speech: "it is thought", "they say". Spleenwort tied on with a mule's hoof "is thought" to prevent conception, and "they say" it must be dug on a moonless night (Diosc. MM 3.134). Hellebore-gatherers face east, pray to Apollo and Asclepius and watch for an eagle, and first eat garlic against the heady fumes (MM 4.162). Mandrake, "Circe's plant", is a love-potion root and also the surgeon's anaesthetic for patients about to be cut or cauterised (MM 4.75).

Galen, a century later, went further and tested. He scorned the grammarian Pamphilus' herbal for its "silly Egyptian sorceries" and the incantations said over herbs as they are dug (Gal. SMT 6 prooem.). But he did not dismiss amulets as a class. Of a peony root worn against childhood epilepsy he writes:

καὶ οἶδά γέ ποτε παιδίον ὀκτὼ μησὶ μηδ’ ὅλως ἐπιληφθὲν ἐξ ὅτου τῆς ῥίζης ἐφόρει, ὡς δ’ ἀπεῤῥύη πως ἀπὸ τοῦ τραχήλου τὸ περιάπτον, εὐθὺς ἐπελήφθη

I know of a child who had no fit at all for eight months while he wore the root; when the amulet somehow slipped off his neck, he had a fit at once.

A new root was hung on and the fits stopped. Galen then took it off again "for the sake of experiment", saw another fit, and hung on a large fresh piece, after which the boy had no more. He blamed particles of the root, breathed in. He separated a scorpion-killing charm into its parts and found that "the scorpion died from the saliva alone, without the incantation" (SMT 10.2.16). He tried a colleague's amulet of wolf dung for colic and was surprised to see most patients helped, insisting that what is hung on is "a substance, not barbarous names, as some of the goētes do" (SMT 10.2.21). Chapter XII weighs what these trials showed. The peony and wolf-dung results are what the natural remission of epilepsy and colic, and expectation, would produce; the scorpion test rightly removed the incantation as a cause but does not show what killed the scorpion.

The other side of the line is written from within. The Cyranides, a compilation of uncertain date (first to fourth century CE), calls itself a book of "natural powers, sympathies and antipathies" revealed by Hermes Trismegistus, and tells the reader to engrave an eagle on an eagle-stone, set a grape-pip and the tip of an eagle's or a hawk's feather beneath it, and wear it (Cyran. 1.1.39). Philosophers kept a respectable place for healing by song. Porphyry says Pythagoras treated those sick in soul "with incantations and magic rites" (ἐπῳδαῖς καὶ μαγείαις), using the word with approval (Porph. VP 33), and Iamblichus derives ἐπῳδή itself from Pythagorean music therapy (Iambl. VP 114).

Driving out daimones

A δαίμων (daimōn) in Greek is any lesser divine power, good or bad; the hostile sense of "demon" is only one of its meanings. Josephus, a Jewish priest writing in Greek, calls exorcism an art God taught Solomon, "very strong among us to this day", and saw a fellow Jew, Eleazar, perform it before Vespasian:

προσφέρων ταῖς ῥισὶ τοῦ δαιμονιζομένου τὸν δακτύλιον ἔχοντα ὑπὸ τῇ σφραγῖδι ῥίζαν ἐξ ὧν ὑπέδειξε Σολόμων ἔπειτα ἐξεῖλκεν ὀσφρομένῳ διὰ τῶν μυκτήρων τὸ δαιμόνιον

He held to the possessed man's nose a ring with one of the roots Solomon had prescribed set under its seal, and as the man smelled it he drew the demon out through his nostrils.

Eleazar then ordered the departing demon to overturn a basin of water as proof. Elsewhere Josephus calls demons "spirits of wicked men" who enter the living, and describes a glowing root near Machaerus that drives them out, so deadly to touch that a dog is used to pull it up (BJ 7.180–185). Pagan accounts are cooler. Lucian's "Syrian from Palestine" questions the daimōn, which answers "in Greek or in a barbarian tongue", and drives it out with oaths and threats for a large fee (Philops. 16). In Philostratus, Apollonius expels a daimōn from a laughing youth at Athens with a stare and a command (VA 4.20) and exposes the rich bride of his pupil Menippus at Corinth as an ἔμπουσα, a shape-shifting female demon "fattening Menippus with pleasures in order to eat his body" (VA 4.25).

The dead and the future

Necromancy stood in an odd position. Strabo lists Persian "magi and necromancers" among the honoured national seers who were "thought worthy even of kingship", together with Chaldaeans and Etruscan augurs (Strabo 16.2.39). Plutarch reports that Sparta, on Delphi's instruction, hired "soul-guides" (ψυχαγωγοί) from Italy to draw the ghost of their regent Pausanias away from a sanctuary (De sera num. vind. 17). Josephus treats the woman of Endor's raising of Samuel as real (AJ 6.330–332), and the emperor Caracalla had the best magi consult the dead about his own end (Herodian 4.12). Yet the dream-interpreter Artemidorus files necromancers with sieve-diviners and cheese-diviners as frauds (Artem. 2.69). The great oracles counted as religion, though Plutarch, himself a priest at Delphi, explains them in nearly physical terms: the earth sends up a prophetic breath that acts on the soul "like a drug" and can weaken or fail, and a Pythia forced to prophesy on a bad day rushed screaming for the exit, and died a few days later (Plut. De def. or. 50–51).

The stars

Astrology in this period is an applied science to its practitioners and a fraud to its critics. Claudius Ptolemy, the astronomer of the Almagest, grounds it in physics: the sun drives the seasons, the moon moves tides and living things, and sailors and farmers already forecast from the sky (Ptol. Tetr. 1.2). Because the stars' influence is strong but not strictly necessary, it can be counteracted, and he praises the Egyptians for joining astrology to medicine:

οὐ γὰρ ἄν ποτε ἀποτροπιασμούς τινας καὶ φυλακτήρια καὶ θεραπείας συνίσταντο πρὸς τὰς ἐκ τοῦ περιέχοντος ἐπιούσας ἢ παρούσας περιστάσεις

Otherwise they would never have devised averting rites, amulets and treatments against the conditions that the surrounding heavens bring on, or have already brought.

The same Ptolemy makes "goētes, magoi, poisoners" a personality type produced by Mars joined with Mercury in a bad position, listed with housebreakers and forgers (Tetr. 3.13). The sceptic Sextus Empiricus accepted weather forecasting from the sky and attacked birth horoscopes, observing that the Chaldaean on his hilltop learns of a birth by a gong and can never fix the exact rising degree (Sext. Emp. M. 5). Plotinus found a middle way, treating the planets as letters "always being written in the sky" that signify events without causing them (Plot. Enn. 2.3.7).

Weather, fire and public wonders

Some of the most striking evidence concerns rites that were public and lawful. At Titane near Sicyon a priest sacrificed one night each year into four pits to tame the winds, and "they say he even chants the spells of Medea" (Paus. 2.12.1). At Methana men ran round the vineyards with the halves of a white cockerel against a withering wind, and Pausanias adds: "I have myself seen men turning away hail with sacrifices and incantations" (Paus. 2.34.3). He also saw a magos in Lydia read from a book, in a language unintelligible to Greeks, until dry wood on an altar caught fire by itself (Paus. 5.27.6). Plutarch's sources called the Roman rite of evocatio, which summoned an enemy city's gods to Rome, "bewitchments of gods" (γοητεῖαι θεῶν), and Rome kept its own guardian god's name secret for fear of the same treatment (Plut. Quaest. Rom. 61).

How it was supposed to work

Imperial Greek writers mostly explained magic in the language of natural philosophy, and several explanations ran side by side.

Emanation. Plutarch, at a dinner where the other guests laughed at the evil eye, built a physical theory for it:

μάλιστα δὲ τοῦτο γίγνεσθαι διὰ τῶν ὀφθαλμῶν εἰκός ἐστι· πολυκίνητος γὰρ ἡ ὄψις οὖσα μετὰ πνεύματος αὐγὴν ἀφιέντος πυρώδη θαυμαστήν τινα διασπείρει δύναμιν

This most likely happens through the eyes. Sight moves very readily, and together with the breath, which sends out a fiery ray, it scatters a remarkable power.

Bodies give off smell, voice and breath, Plutarch argues, and envious eyes send out their vice "like poisoned arrows". Grotesque amulets against the eye (προβασκάνια) work by catching the gaze with their oddness, so that it presses less on its victim (Quaest. conv. 5.7.3). Heliodorus' Calasiris gives the same account, comparing envy in the air to the spread of plague (Heliod. Aeth. 3.7).

Sympathy. The cosmos was widely held to be one living body whose parts affect each other by likeness and contrariety (συμπάθεια, ἀντιπάθεια). Plotinus, writing in Rome in the 260s, made sympathy the basis of a complete theory of magic:

καὶ γὰρ μηδενὸς μηχανωμένου ἄλλου πολλὰ ἕλκεται καὶ γοητεύεται: καὶ ἡ ἀληθινὴ μαγεία ἡ ἐν τῷ παντὶ φιλία καὶ τὸ νεῖκος αὖ. καὶ ὁ γόης ὁ πρῶτος καὶ φαρμακεὺς οὗτός ἐστιν

Even when no one contrives anything, many things are drawn and bewitched. The true magic is the Love in the universe, and the Strife as well. This is the first sorcerer and drug-maker.

For Plotinus a sorcerer only exploits connections that nature makes anyway. A charm's melody moves the irrational part of the soul as music does, and a prayer is answered without the star "hearing" it. The wise man's reasoning soul cannot be bewitched, though his body can, and he can undo spells "by chanting back" (Enn. 4.4.43–44). He also explained why statues can hold divine presence: whatever imitates something is receptive to it, "like a mirror able to catch a form" (Enn. 4.3.11). Sallustius, of the emperor Julian's circle, extended this "receptivity" to altars, herbs, stones and the magical signs called χαρακτῆρες (Sallust. De deis 15).

Daimones. Porphyry, Plotinus' pupil, gave a darker account. Evil daimones cause plagues and droughts and then persuade people the good gods are angry, and "all goēteia is accomplished through the opposite kind" (Porph. Abst. 2.41). Their bodies are made of vapour and fattened on the smoke of sacrifices, and through them "the wretched prepare philtres and love charms" (Abst. 2.42). Eusebius later quoted this for its closeness to Christian demonology. Celsus reported the Egyptian doctrine that thirty-six daimones each rule a part of the body and heal it when invoked by name (Orig. Cels. 8.58–60).

Names and tokens. Iamblichus, writing about 300 CE under the persona of an Egyptian priest, defended the "barbarous names" of ritual. If names were conventional, one could swap them, "but the barbarian names have great expressive force and great concision"; Greeks, being "innovators by nature", wear their prayers out by changing them (Iambl. Myst. 7.4–5). A Hermetic treatise asks that it never be translated, because Egyptian speech consists of "sounds full of deeds" (CH 16.2). Iamblichus' central claim goes further still:

οὐδὲ γὰρ ἡ ἔννοια συνάπτει τοῖς θεοῖς τοὺς θεουργούς … καὶ γὰρ μὴ νοούντων ἡμῶν αὐτὰ τὰ συνθήματα ἀφ’ ἑαυτῶν δρᾷ τὸ οἰκεῖον ἔργον

It is not thinking that joins theurgists to the gods … even when we are not thinking of them, the tokens do their own work by themselves.

The συνθήματα (synthēmata) are tokens the gods themselves planted in stones, plants, sounds and rites, which they recognise and answer. On this view ritual works independently of the practitioner's understanding. Plotinus would not have accepted it.

Compulsion. The most troubling idea was that rites could force the gods. Porphyry's collection of oracles, preserved by Eusebius, has a god complain, "Hear me, unwilling as I am, since you have bound me with necessity" (Eus. PE 5.8 = Porph. Phil. ex orac.). Egyptian operators threatened "to dash down heaven", "to stop the sacred barque" or "to scatter the limbs of Osiris". Iamblichus defended the threats by saying they were aimed only at lower, irrational powers, and that the theurgist speaks in the person of a god (Myst. 6.5–7). Since most authors defined illicit magic by this claim, the theurgists had to explain it away.

Where the line ran, and what happened to those who crossed it

Nearly every author in this chapter draws a line between magic and something respectable, and they draw it in different places. Philostratus draws it at fate:

οἱ γόητες, ἡγοῦμαι δʼ αὐτοὺς ἐγὼ κακοδαιμονεστάτους ἀνθρώπων, οἱ μὲν ἐς βασάνους εἰδώλων χωροῦντες, οἱ δʼ ἐς θυσίας βαρβάρους, οἱ δὲ ἐς τὸ ἐπᾷσαί τι ἢ ἀλεῖψαι μεταποιεῖν φασι τὰ εἱμαρμένα

The goētes, whom I consider the most wretched of men, claim to alter what is fated, some by torturing images, some by barbarous sacrifices, some by chanting over something or anointing it.

Apollonius, by contrast, followed the Fates and foreknew what the gods revealed. When he slipped his leg out of a prison shackle, his companion took it as proof of a divine nature, because he did it "without sacrifice, prayer or word" (VA 7.38). Heliodorus draws the line between two kinds of Egyptian wisdom:

ἣ μὲν γάρ τις ἐστὶ δημώδης, καὶ ὡς ἄν τις εἴποι, χαμαὶ ἐρχομένη, εἰδώλων θεράπαινα καὶ περὶ σώματα νεκρῶν εἰλουμένη, βοτάναις προστετηκυῖα καὶ ἐπῳδαῖς ἐπανέχουσα

One kind is common and, so to speak, crawls on the ground. It waits on phantoms and creeps about dead bodies, clings to herbs and relies on incantations.

The higher wisdom of priests and prophets looks upward, studies the stars and aims at the good. Porphyry says purity belongs "to divine and god-wise men", which sorcerers cannot keep because they trouble evil daimones for lust (Porph. Abst. 2.45). Iamblichus rejects image-makers because their maker "is not a god but a human being" (Myst. 3.28). Josephus' Moses tells Pharaoh that his signs come "by God's providence and power" (AJ 2.284–287). Galen draws the line at substance against words, Artemidorus inside his own profession (Artem. 2.69), and Plutarch between piety and superstition, whose signs are "sorceries and magic rites, runnings-around and drummings, unclean purifications" (De superst. 12).

The state drew it in political terms. In the speech Cassius Dio composed for Maecenas, advising Augustus, the historian of the Severan age sets out a policy:

μαντικὴ μὲν γὰρ ἀναγκαία ἐστί, καὶ πάντως τινὰς καὶ ἱερόπτας καὶ οἰωνιστὰς ἀπόδειξον, οἷς οἱ βουλόμενοί τι κοινώσασθαι συνέσονται· τοὺς δὲ δὴ μαγευτὰς πάνυ οὐκ εἶναι προσήκει.

Divination is necessary, so by all means appoint haruspices and augurs for anyone who wants to consult them. But magicians ought not to exist at all.

The reason given is that such men, by telling a few truths among many lies, stir people to revolution. Dio also records that Agrippa expelled astrologers and goētes from Rome (Cass. Dio 49.43.5). The goētes of Josephus met the sword: the procurator Fadus' cavalry killed Theudas' followers and beheaded him (AJ 20.97). Philostratus says the art "is condemned both by nature and by law" (VA 7.39), and that many goētes on trial confessed their skill (VA 5.12). Apollonius' defence speech answers a charge of magic that included sacrificing a boy (VA 8.7); the Roman laws behind such trials belong to chapter VI. In 371–372 pagan philosophers who had used a laurel tripod to spell out Valens' successor were burned or beheaded, and men stopped wearing the philosopher's cloak (Soz. HE 6.35; see chapter VI).

Private accusations had force too. Galen did not advertise his accurate prognoses "lest the physicians and philosophers hate me even more, babbling that I am a goēs and a seer" (Gal. De praenot. 10). Aretaphila, caught preparing a potion for the tyrant of Cyrene, saved herself under torture by saying it was only a love charm, "foolish perhaps and womanish, but not worth death" (Plut. Mul. virt. 19). The pagan Celsus turned the same accusation on Jesus, who he claimed had learned "certain powers on which the Egyptians pride themselves" while working in Egypt (Orig. Cels. 1.28). Meanwhile the wind-priest at Titane, Spartan ghost-laying and Roman evocatio did what goētes were accused of doing, and nobody prosecuted them. The line followed who performed a rite and on whose authority more closely than it followed the rite itself.

Doubters and mockers

Educated scepticism was widespread, though it rarely amounted to complete disbelief. Marcus Aurelius thanks his tutor for teaching him "disbelief in what wonder-mongers and goētes say about incantations and the sending-away of daimones" (M. Aur. 1.6). Polybius called claims that certain statues were never rained on "childish", tolerable only where they kept the masses pious (Polyb. 16.12). Plutarch tells young wives to laugh at anyone who claims to draw down the moon, since Aglaonice of Thessaly simply knew when lunar eclipses would fall and used the knowledge to fool other women (Conj. praec. 48; De def. or. 13).

Lucian was the great mocker. His Philopseudes sets a sceptic among philosophers who swap cures (a shrew's tooth tied on in a lion skin, or must it be a doe's?) and tall tales, among them the first sorcerer's apprentice, who sets a pestle carrying water and cannot stop it (Philops. 7–8; 34–36). His Alexander exposes a real oracle-founder at Abonuteichos on the Black Sea coast:

μεμηνέναι προσποιούμενος ἐνίοτε καὶ ἀφροῦ ὑποπιμπλάμενος τὸ στόμα· ῥᾳδίως δὲ τοῦτο ὑπῆρχεν αὐτῷ, στρουθίου τῆς βαφικῆς βοτάνης τὴν ῥίζαν διαμασησαμένῳ

Sometimes he pretended to be mad and filled his mouth with foam. That came easily: he chewed the root of soapwort, the dyers' plant.

Lucian goes on to explain how sealed questions were opened with a hot needle (Alex. 19–21) and how the god's "self-spoken" oracles came through a tube of cranes' windpipes (Alex. 26). Sextus notes that goētes treat lamp wicks with verdigris to change the colour of onlookers' faces (Sext. Emp. PH 1.46). Chapter XIII covers these tricks.

More interesting than the exposés are the ancient explanations of why people believed. Philostratus describes something close to what modern psychologists call confirmation bias:

ἢν μὲν γὰρ παθόντων τι τῶν παιδικῶν πρὸς τοὺς ἐρῶντας ἢ δώροις ὑπαχθέντων προβαίνῃ τὰ ἐρωτικά, ὑμνεῖται ἡ τέχνη ὡς ἱκανὴ πάντα, εἰ δʼ ἀποτυγχάνοι ἡ πεῖρα, ἐς τὸ ἐλλειφθὲν ἡ ἀναφορά

If the beloved warms to the lover, or is won over by gifts, and the affair goes ahead, the art is praised as all-powerful. If the attempt fails, the blame goes to something left out.

The client says, "if only I had burned the thing", and returns. Alexander of Aphrodisias describes fate-tellers who approach people in crises and make shrewd guesses from experience "of what usually follows in such cases", then present them as a science of fate (Mantissa 24); the technique resembles modern cold reading. An Egyptian musician told Celsus that magic works on "the uneducated and those of corrupted character" and can do nothing to philosophers (Orig. Cels. 6.41). Plotinus attacked Gnostic teachers who claimed to drive out disease-demons by speech, replying as a physician would that diseases come from "fatigue, surfeit, deficiency or putrefaction" and yield to purging and drugs (Enn. 2.9.14).

Doubt about one practice often sat beside belief in another. Plutarch laughed at moon-pulling and theorised the evil eye; Galen mocked binding spells and tested a peony amulet; Philostratus called goēteia the vilest trade and told Apollonius' miracles with conviction. Genre mattered as much as temperament. Satire mocks, handbooks instruct, histories hedge with "it is said", and defences of a hero or a nation insist that their wonders were of another kind.

Change over time

The Hellenistic historians mainly rationalise. Polybius rejects marvels. Diodorus makes Hecate a Taurian princess who tested aconite on strangers (Diod. Sic. 4.45) and treats goētes as figures of myth, such as the Idaean Dactyls who invented incantations (Diod. Sic. 5.64). Strabo, under Augustus, suggests that the sorcerer Telchines were skilled metalworkers slandered by rivals (Strabo 14.2.7), and treats magi and Chaldaeans as honoured foreign priesthoods.

In the first century CE the subject turns political and practical: Josephus' goētes are revolutionary prophets, Solomonic exorcism is an inherited science, and Dioscorides catalogues folk uses at a distance. The most intense debate falls under the Antonine emperors, roughly 150–190, the decades of Lucian, Galen, Ptolemy, Sextus, Artemidorus, Marcus and Celsus. From the Severan age the emphasis shifts to definition and defence. Philostratus keeps his holy man clear of goēteia, Cassius Dio argues for banning magicians, and Heliodorus' two wisdoms, if the novel is this early, anticipate the philosophers. Plotinus in the 260s explains magic by cosmic sympathy, Porphyry assigns sorcery to evil daimones, and around 300 Iamblichus makes ritual the summit of philosophy under its own name, theurgy. In the 360s the emperor Julian speaks of "the blessed theurgists" who know what the "rabble" do not (Jul. Or. 5), and in the fifth century Proclus still cites a "soul-drawing wand" that drew a sleeping boy's soul from his body before Aristotle (Procl. In Remp. II.122 Kroll). Over six centuries magic moves from an exotic and mythical topic to a moral, legal and finally theological one.

What is distinctive here

Set beside the other cultures in this report, the Hellenistic and imperial Greek world stands out first for the amount of theory it produced. Many texts elsewhere in the library state what a rite does; these authors ask how it could work, and answer with emanations, breath, sympathy, daimones, the nature of names and the tokens of the gods. Second, the period contains real testing: Galen's on-off-on trial of the peony, his saliva without the spell, and his trial of the wolf-dung amulet have few parallels anywhere in the library. Third, it has the richest literature of exposure, from Lucian's forensic notes to Sextus' lamp wicks, written by men who often believed in something close by.

Fourth, magic in these texts is almost always somebody else's: Persian, Chaldaean, Egyptian, Syrian, Jewish, Arabian or Thessalian. Yet Strabo, Dio Chrysostom and others said plainly that at home the magos was a priest, and the practitioners advertised the same map themselves, crediting charms to a Thessalian and a Syrian woman (PGM XX 2.21–36) and their books to the temples of Egypt (Suppl. Mag. 72 i 1–5; see 'The practitioners' own books'). Fifth, the label was a weapon feared by doctors, sophists and holy men, while the civic rites that most resembled sorcery went unchallenged. Finally, no culture in the library took what its own writers called magic so far into philosophy: Plotinus explained it, Iamblichus elevated it, and the theurgists practised it as the highest piety. Chapter VII follows the Christian writers of the same centuries, who took over many of these categories and redrew the line again.

Chapter VI

Rome: the law of the evil song

Roman magic was defined less by what was done than by who did it, where and for whom, and the city's law, poetry and history show the line between rite and crime moving from the Twelve Tables to the treason trials of the fourth century.

Lead tablets in the walls

In the autumn of 19 CE Germanicus Caesar, Tiberius' adopted son and heir, lay sick at Antioch, convinced that the governor of Syria, Gnaeus Piso, had poisoned him. Tacitus reports what was found in the house.

reperiebantur solo ac parietibus erutae humanorum corporum reliquiae, carmina et devotiones et nomen Germanici plumbeis tabulis insculptum, semusti cineres ac tabo obliti aliaque malefica quis creditur animas numinibus infernis sacrari.

Dug out of the floor and the walls they kept finding human remains, spells and curses, Germanicus' name cut into tablets of lead, half-burnt ashes smeared with gore, and the other instruments of harm by which, it is believed, souls are handed over to the powers below.

Germanicus died. A lead tablet with a name is what archaeologists have recovered from Roman graves, springs and shrines in large numbers, and the Latin tablets in the library hand such names to the powers below, one of them under the heading inimicorum nomina ad inferos, names of enemies to the underworld (EDH HD076474; see 'Curses from the provinces' below). Tacitus does not say whether the curse worked; he sets it beside the suspicion of poison, and the two charges ran together because Latin used one word, veneficium, for both. And the victim was the heir to the throne, so private sorcery had become a matter of state. Roman magic is known to us mostly at moments like this, when it was found and prosecuted.

The chapter draws on two bodies of verified texts: Latin poetry from Plautus (about 205 BCE) to the Christian poets of Gaul (about 507 CE), and non-Christian Latin prose from Cato the Elder (about 160 BCE) to Ammianus Marcellinus (about 390 CE). Christian prose belongs to chapter VII, and the Greek writers of the empire to chapter V. The Latin curse tablets of the provinces, added to the library since, are read in 'Curses from the provinces' below.

The Roman vocabulary

Early Rome had no word for "magic". The noun magia is rare in the library before Apuleius in the 150s CE. Romans had instead words for particular acts, most of them neutral until context made them sinister.

• Carmen is any formal, rhythmic utterance: poem, hymn, legal formula, prayer or spell. With its verbs incantare and excantare ("charm away") it is the oldest Roman word for magic. Horace says of a children's choir praying for rain that "by song the gods above are appeased, by song the shades below" (Hor. Epist. 2.1.132–138).
• Venenum is any potent drug, even a hair dye. Veneficium is the giving of harmful drugs and, from the late Republic, the main legal name for sorcery. A venefica is poisoner and witch at once.
• Saga, "wise woman", and anus, "old woman", name the female practitioner. Petronius' freedmen call such women mulieres plussciae, "women who know more" (Petron. Sat. 63).
• Magus first means a Persian priest, as still in Catullus (Catull. 90). By the first century BCE it is an insult for anyone dealing in foreign, secret rites.
• Devovere, to consecrate someone to the gods below, and defigere, to pin or transfix, are the words of cursing. On the curse tablets themselves both words are uncommon: defigo is on 12 of the library's 287 Latin tablets (EDH HD024366) and devoveo or devotio on 7 (EDH HD051474), while the British writers more often give (dono) the culprit to a god (EDH HD000090; see 'Curses from the provinces' below).
• Maleficium means "wrongdoing" in Cicero. Only from Tacitus and Apuleius on does it narrow to "sorcery".
• Superstitio, religion that is excessive, fearful, private or foreign, is the frame into which Romans fitted most of this, set against religio, correct ancestral worship.

Two more words name creatures. The strix is a screech-owl that eats children's entrails, or a woman turned into one; Plautus' cook already jokes about rivals who season their dishes "with screech-witches" (Plaut. Pseud. 819–821). The versipellis, "skin-changer", is a werewolf (Pliny NH 8.80–84).

Romans also had a map on which magic always came from elsewhere. Witches were Thessalian, love-poisons Colchian (Medea's country), astrologers Chaldaean, entrail-readers Etruscan, snake-charmers Marsian or Libyan. Plautus' Amphitryo blames his household's madness on "that Thessalian sorcerer" (Plaut. Amph. 1043), and Sidonius still recites the map in 458 (Sidon. Carm. 5.259–263).

Who practised it

The first practitioner to consider is the head of a Roman household. Cato's farming manual gives the landowner a charm for a dislocated joint.

Luxum siquod est, hac cantione sanum fiet. harundinem prende tibi viridem P. IIII aut quinque longam, mediam diffinde, et duo homines teneant ad coxendices. incipe cantare [...] 'motas uaeta daries dardares astataries dissunapiter', usque dum coeant.

If there is a dislocation, this chant will make it sound. Take a green reed four or five feet long and split it down the middle, and have two men hold it to the hips. Begin to chant [...] "motas uaeta daries dardares astataries dissunapiter" until the halves come together.

The words mean nothing in Latin. Cato goes on to have the reed cut and bound to the dislocation or fracture, where it would serve as a splint, though nothing in the rite puts a joint back. The same Cato forbids his slave farm-manager to consult a haruspex, an augur, a soothsayer or a Chaldaean (Cato Agr. 5.4), and prescribes a lustration in which a pig, a sheep and a bull are led round the fields while Mars is asked to keep off disease and storm (Cato Agr. 141.2–3). The master's own charm and sacrifice were household religion; the paid outsider was the danger. Two centuries later Columella repeats the rule and adds the saga (Columella 1.8.6).

Below the master stood ordinary female specialists. Plautus' old bachelor will not marry because a wife would want money at Minerva's festival for the charm-woman, the dream-reader, the soothsayer, the female entrail-reader and "the woman who reads by the eyebrow" (Plaut. Mil. 690–696). The joke is about expense, and nobody treats these women as criminals. In Persius a grandmother dabs spittle on a baby's forehead against the evil eye (Pers. 2.31–40), and when Tibullus' Delia was sick an old woman sang a charm over her (Tib. 1.5.9–16).

Men appear mostly as professionals with ethnic titles: Persian or Armenian magi, Chaldaean astrologers, Etruscan haruspices, Marsian snake-charmers like Virgil's priest Umbro, who can soothe vipers but not heal a spear wound (Verg. Aen. 7.750–760). The best clients sat at the top of society. The augur Appius Claudius consulted oracles of the dead (Cic. Div. 1.132); the senator Nigidius Figulus used entranced boys to find a lost purse (Apul. Apol. 42); a leading senator wore an amulet of two Greek letters (Pliny NH 28.22–29). Tiberius kept the astrologer Thrasyllus (Tac. Ann. 6.21), and Nero had magi try to raise the ghost of the mother he had murdered (Suet. Nero 34.4).

In literature the witch is a woman, usually old, poor and urban, and often also a bawd. Pliny calls drawing down the moon "this one female science" (Pliny NH 25.9–11), and Quintilian tells students that one would more readily believe banditry of a man and veneficium of a woman (Quint. Inst. 5.10.25). Many of the accused in real trials were women, from the matrons of 331 BCE to Martina, sent from Antioch after Germanicus' death, which suggests that the stereotype shaped who was suspected.

What they did

Love and binding

Latin poetry's model love spell is Virgil's eighth Eclogue, adapted from Theocritus. A woman whose lover has gone to the city ties three threads of three colours in three knots, because "the god delights in an odd number" (Verg. Ecl. 8.75), and puts a clay figure and a wax one in the same fire: "as this clay hardens and this wax melts" (Ecl. 8.80), so may Daphnis harden to others and melt for her.

Carmina vel caelo possunt deducere Lunam; / carminibus Circe socios mutavit Ulixi; / frigidus in pratia cantando rumpitur anguis. / Ducite ab urbe domum, mea carmina, ducite Daphnim.

Songs can pull the Moon itself down from the sky; with songs Circe changed Ulysses' crew; chanting bursts the cold snake in the meadows. Bring him home from the city, my songs, bring Daphnis home.

(The edition misprints pratis as pratia.) Horace's Canidia uses two dolls, a large woollen one to punish a small wax one (Hor. S. 1.8.17–50). Ovid, impotent with a girl, wonders whether a witch "has pinned my name in red wax and driven fine needles into the middle of the liver" (Ov. Am. 3.7.27–36). The magic wheel was spun to draw a lover and reversed to release one (Hor. Epod. 17).

Binding also had a place in the official year. At February's festival of the dead, Ovid's old woman performs the rite of Tacita, the Silent Goddess, among a group of girls.

tunc cantata ligat cum fusco licia plumbo / et septem nigras versat in ore fabas, / quodque pice adstrinxit, quod acu traiecit aena, / obsutum maenae torret in igne caput;

Then she binds enchanted threads with dark lead, turns seven black beans over in her mouth, and roasts in the fire the head of a small fish, stitched shut, sealed with pitch and run through with a bronze needle.

Leaving, tipsy on the leftover wine, she announces that hostile tongues are bound. Lead, piercing and the silencing of opponents are the stuff of real curse tablets, and here they sit in the calendar of state religion. Three Latin tablets call on Muta Tacita, the Cambodunum tablet perhaps on Mutae in the plural, to strike their writers' opponents dumb (EDH HD019906; EDH HD027805; EDH HD082820; see 'Curses from the provinces' below).

The dead

Necromancy was the most feared practice and the one most often charged. Cicero accuses his enemy Vatinius of "calling up the souls of the dead and appeasing the gods below with boys' entrails" (Cic. Vat. 14). Horace's fifth Epode makes a poem of the charge. Canidia and her helpers bury a kidnapped freeborn boy to his chin and set food before him that he cannot reach.

exsucta uti medulla et aridum iecur / amoris esset poculum, / interminato cum semel fixae cibo / intabuissent pupulae.

so that his marrow, sucked out, and his dried liver might make a love potion, once his eyes, fixed on the food he was forbidden, had wasted away.

The ingredients (grave fig, a strix feather, eggs smeared with toad's blood), and the setting of Horace's other Canidia poem, the old paupers' cemetery on the Esquiline, became the stock of every later Latin witch. Under Nero the figure grew into a monster. In Lucan's civil-war epic, Pompey's son Sextus consults the Thessalian Erichtho, whom Lucan defines by everything Roman worship required.

Nec superos orat, nec cantu supplice numen / Auxiliare vocat, nec fibras illa litantes / Novit: funereas aris imponere flammas / Gaudet,

She does not pray to the gods above or summon a helping power with a suppliant hymn, and she knows nothing of entrails that show a sacrifice accepted. She delights in putting funeral fire on the altars.

She picks a fresh corpse with sound lungs so that its voice will be full, fills it with hot blood and "moon-froth", and threatens the underworld gods with their true names until the dead man rises and foretells Pompey's ruin (Luc. 6.667–830). The same rite is pious when Thebes' royal seer performs it for the king (Sen. Oed. 548–607). Real Romans tried it. In 16 CE the young noble Libo Drusus was ruined when a man he had asked "to lure up the shades of the underworld with chants" informed on him (Tac. Ann. 2.27–30).

Healing and protection

Most Roman practice was protective and mundane. Pliny lists blessing sneezes and not sweeping while a guest leaves (Pliny NH 28.22–29), and spitting into one's own lap to excuse a boastful hope (28.35–38). Against quartan fever, he says, bedside medicine "is almost powerless", so he gives amulets: dust a hawk has rolled in, a black dog's tooth, a lizard's right eye (Pliny NH 30.98–99). Myrrha's nurse in the Metamorphoses offers three remedies for unexplained suffering: for madness, a woman who heals with chant and herbs; for harm done by someone, a magical purification; for divine anger, sacrifice (Ov. Met. 10.394–399). Folk healing, counter-magic and cult were three services in one market.

The stars and the future

From the second century BCE Chaldaean astrology was the foreknowledge Rome's rulers most wanted and most feared. Juvenal's women trust an astrologer as if he spoke from Ammon's spring, "since the oracles at Delphi have fallen silent" (Juv. 6.553–556). Tacitus describes the astrologers urging Otho on in 69.

urgentibus etiam mathematicis, dum novos motus et clarum Othoni annum observatione siderum adfirmant, genus hominum potentibus infidum, sperantibus fallax, quod in civitate nostra et vetabitur semper et retinebitur.

The astrologers pushed him on as well, declaring from the stars that there would be upheavals and a glorious year for Otho: a kind of men that betrays the powerful and deceives the hopeful, and that in our state will always be forbidden and always be kept.

The most elaborate device in the library is described at a treason trial in 371: a laurel tripod, a metal dish with the alphabet round its rim, and a ring on a thread that swung from letter to letter spelling out hexameters (Amm. Marc. 29.1.29–32). Asked who would succeed Valens, it spelled THEOD, and a man named Theodorus was executed with many others.

How it was supposed to work

Exact words. Roman state religion held that formulae failed if misspoken. Pliny notes that a magistrate's prayer needed a reader, a checker and a piper to drown ill-omened sounds, and asks outright "whether words and the incantations of charms have any power" (Pliny NH 28.10–13). His evidence is Roman: the Vestals were believed to root runaway slaves to the spot by prayer, and the consul Decius destroyed an army with a formula.

Likeness and contact. Clay hardens and wax melts; a doll is punished, a name pierced; a split reed closes and so does the joint. Modern scholars, following Frazer, call this sympathetic magic. The Latin texts perform it without explaining it. The curse tablets state the comparison outright, as I write this backwards, so may everything he does go backwards, but give no reason why it should hold (EDH HD054658; see 'Curses from the provinces' above).

Potent substances. Herbs cut at night with bronze sickles, the hippomanes discharge of a mare in heat (Verg. G. 3.280–283), menstrual blood that drives pests from a field (Pliny NH 28.77–80), corpse parts. Medea's rejuvenating cauldron in Ovid holds moon-frost, a werewolf's entrails and the head of a crow that has outlived nine generations, so that the animal's long life passes to the old man (Ov. Met. 7.179–293).

Compulsion of the gods. Lucan alone among the poets turns this into theory. "What bargain, what pact, holds the gods bound? Must they obey, or do they enjoy it?" (Luc. 6.492–499). He offers a pact, an unknown piety, threats and a supreme god who compels the rest, and leaves the question open. He also sets a limit. Erichtho admits that her art can delay or hasten one man's death, but when the whole chain of causes from the world's beginning is at stake, fate is stronger (Luc. 6.589–623). Stoic fate has been put into a witch's mouth.

Daemons and the soul. Apuleius the Platonist argues that a child's soul, lulled by chant or scent, can return towards its divine nature and foresee the future, with daemons as go-betweens (Apul. Apol. 43).

The human mind. Pliny's own theory is psychological and historical.

fraudulentissima artium plurimum in toto terrarum orbe plurimisque saeculis valuit. [...] natam primum e medicina nemo dubitabit ac specie salutari inrepsisse velut altiorem sanctioremque medicinam

The most fraudulent of all the arts has had the greatest power over the whole world and for the most ages. [...] No one will doubt that it was born from medicine and crept in looking like a cure, as a higher and holier kind of medicine.

Magic then absorbed religion and astrology, and by holding together the hope of health, the fear of the gods and the wish to know the future it gained its hold over the mind. Pliny goes on to trace it from Zoroaster through Osthanes (Pliny NH 30.3–11). Written in the 70s CE, this is the earliest Latin text in the library that treats magic as a single thing with a history.

Where the line ran, and how Rome punished

The Twelve Tables

Rome's first law code, of about 450 BCE, survives only in quotation. Pliny preserves two clauses.

quid? non et legum ipsaruin in duodecim tabulis verba sunt: qui fruges excantass it, et alibi: qui malum carmen incantassit ? [...] defigi quidem diris precationibus nemo non metuit.

Are these not the very words of the laws in the Twelve Tables: "whoever has charmed away crops", and elsewhere "whoever has chanted an evil song"? [...] And there is no one who is not afraid of being transfixed by dread curses.

(The library's text has the misprints ipsaruin and excantass it.) Scholars dispute whether malum carmen first meant a harmful spell or a defamatory song, and the library does not give the penalty. The crop clause is clear: charming a neighbour's harvest into one's own field was theft by magic. The poets remembered it. Chant "lures the crops from neighbours' fields", says Tibullus (Tib. 1.8.17–24), and Virgil's Moeris can move sown crops to another field (Verg. Ecl. 8.99).

Pliny also shows the law in use. In 191 BCE the freedman Furius Cresimus was harvesting far more from a small farm than his neighbours from large ones, and was prosecuted for luring their crops with veneficia. He brought his sturdy slaves, iron tools and fat oxen into the Forum and said "these, citizens, are my sorceries", adding that he could not bring his late nights and sweat. He was acquitted by every vote (Pliny NH 18.41–43). An outsider who did unusually well was the natural suspect.

The same act on both sides of the line

No technique in itself made an act magical. In 340 BCE the consul Publius Decius Mus repeated after the chief priest a formula giving himself and the enemy army to the gods of the dead.

dique Manes, vos precor, veneror, veniam peto oroque, uti populo Romano Quiritium vim victoriamque prosperetis hostesque populi Romani Quiritium terrore, formidine morteque adficiatis.

And you, gods of the dead, I pray to you, I worship you, I ask and beg your favour: give strength and victory to the Roman people of the Quirites, and strike the enemies of the Roman people with terror, dread and death.

Then he rode into the enemy and died. In form this devotio is a curse; Livy treats it as the most glorious act of Roman religion, and Germanicus' tablets bore the same word. King Numa lured Jupiter down from the sky with charms and haggled with him over sacrifice, though how it was done is "forbidden for a man to know" (Ov. Fast. 3.323–346); his successor Tullus performed the rite wrongly and was killed by lightning (Livy 1.31.8). At the Lupercalia a bride is told she will not become a mother "by potent herbs, nor by prayer, nor by magic song", only by the blows of the priests' goatskin thongs (Ov. Fast. 2.425–428).

What made a rite a crime was its setting and aim: privacy, night, graveyards, payment, foreign origin, a woman or stranger as agent, harm or coercion as purpose. In the war crisis of 213 BCE, Livy says, refugees crowded into Rome and "petty sacrificers and seers" made "easy profit from other people's error" (Livy 25.1.6–12), and the senate ordered all prophetic and ritual books handed in. Augustus burned more than two thousand prophetic books and kept only the state's Sibylline oracles (Suet. Aug. 31.1).

Poison and sorcery

The first mass trial in the library is for veneficium. In the plague year 331 BCE a slave woman revealed that matrons were "cooking" drugs; two insisted they were medicines, were made to drink them, and died, and about 170 women were condemned (Livy 8.18.4–13). Livy wishes the story were false. More inquiries ran through the plague years 184–180 BCE beside the suppression of the Bacchic cult, which began, Livy says, with "a Greek of low birth, a petty sacrificer and seer" (Livy 39.8.3–8); he reports thousands condemned, though he doubts one figure of 2,000 (Livy 39.41.5–6).

In 81 BCE Sulla created a standing court for murder and poisoning; Cicero quotes the law against anyone who "made, sold, bought, possessed or gave" poison (Cic. Clu. 148). Whether spells counted was for lawyers to argue. Quintilian's classroom example asks whether "the spells of magi" are veneficium, since everyone knows spells are not a deadly drink (Quint. Inst. 7.3.7).

Treason

Under the emperors the gravest magic was asking about the emperor's death, which counted as maiestas, treason. After Libo's trial the senate expelled astrologers and magi from Italy and threw one from the Tarpeian Rock (Tac. Ann. 2.32). When the decree was renewed in 52, Tacitus judged it atrox et inritum, "harsh and ineffectual" (Tac. Ann. 12.52). In 66 Servilia, not yet twenty, was charged with selling her jewels to pay for "magic rites" after asking magi about her father's fate (Tac. Ann. 16.30–31).

Apuleius on trial

About 158 CE, at Sabratha in North Africa, the philosopher Apuleius was accused of using magic to win a rich widow, Pudentilla. His Apologia is the only surviving speech by a man charged with magic. In the learned sense, he says, a magus is just the Persian word for a priest, so "what crime is it to be a priest" (Apul. Apol. 25.9–11)? In the popular sense a magus can do anything by converse with the gods, in which case his accusers were brave to prosecute (Apul. Apol. 26.6–9). Then he describes what the law punished.

magia ista, quantum ego audio, res est legibus delegata, iam inde antiquitus XII tabulis propter incredundas frugum inlecebras interdicta, igitur et occulta non minus quam tetra et horribilis, plerumque noctibus uigilata et tenebris abstrusa et arbitris solitaria et carminibus murmurata

This magic, so I am told, is a matter for the laws, forbidden long ago by the Twelve Tables because of unbelievable ways of luring crops. So it is secret as well as foul and horrible: kept up mostly at night, hidden in darkness, done alone without witnesses, and muttered in spells.

The prosecution said fifteen slaves had watched his rite; Apuleius replies that so many witnesses rule out magic. They said he worshipped a skeleton carved from rare wood. He produced it in court, a pretty boxwood Mercury (Apul. Apol. 61–64). The verdict is not recorded; scholars usually infer acquittal from the speech's publication.

His novel, the Golden Ass, is full of the magic he denied. The innkeeper Meroe can "bring down the sky, hang the earth in the air, harden springs, dissolve mountains" (Apul. Met. 1.8). Pamphile's rooftop workshop holds inscribed metal plates and the limbs of corpses (Apul. Met. 3.17–18), and when Lucius copies her ointment to become a bird he becomes a donkey (Apul. Met. 3.21–25). The cure is Isis, whose priest tells Lucius he has passed from blind Fortune to "a Fortune who sees" (Apul. Met. 11.15).

Ammianus and the fourth-century trials

Under the Christian emperors Constantius II and Valens, the treason charge reached down to ordinary healing. Ammianus, a pagan officer writing about 390, reports that at Scythopolis in 359 anyone who wore a remedy for quartan fever round his neck, or was said to have passed a tomb at dusk, was condemned as a poisoner (Amm. Marc. 19.12.14). Pliny had listed such amulets without a hint of crime. An old wives' pain charm, "which medical authority also allows", could be fatal (Amm. Marc. 16.8.2), and a governor of Asia executed an old woman who cured fevers "with a gentle charm" after she had cured his own daughter (Amm. Marc. 29.2.26).

Visus adulescens in balneis admovere marmori manus utriusque digitos alternatim et pectori, septemque vocales litteras numerasse, ad stomachi remedium prodesse id arbitratus: in iudicium tractus, percussus est gladio post tormenta.

A young man was seen in the baths touching the marble and then his chest with the fingers of both hands in turn, counting off the seven vowels, in the belief that it would help his stomach. He was dragged to court, tortured and beheaded.

The seven Greek vowels, linked to the seven planets, are common in the Greek magical papyri, written in rows on handbooks and amulets alike (PGM I 11–12; Suppl. Mag. 10). After the Antioch trials owners across the East burned their own libraries, and Ammianus insists that most seized books were ordinary learning (Amm. Marc. 29.1.41). The persecutors used what they punished: Maximinus, who ran the Roman trials, had kept a Sardinian necromancer and then had him killed (Amm. Marc. 28.1.7).

Doubt and mockery

Elite Latin writers were mostly sceptical in principle and inconsistent in practice. Cicero, himself an augur, wrote antiquity's fullest attack on divination after Caesar's murder. He remembers the Chaldaeans promising Pompey, Crassus and Caesar that each would die "in old age, at home, and in glory" (Cic. Div. 2.99); all three died violently. Throw dice often enough, he adds, and the best throw comes up by chance (Cic. Div. 2.121). Shown votive paintings of sailors saved by their vows, the atheist Diagoras asks where the paintings of the drowned are (Cic. ND 3.89).

Pliny tested magic against history. Nero, with unlimited wealth and a wish to command the gods, studied it under the Armenian king Tiridates and gave it up.

inmensum, indubitatum exemplum est falsae artis, quam dereliquit Nero [...] proinde ita persuasum sit, intestabilem, inritam, inanem esse, habentem tamen quasdam veritatis umbras, sed in his veneficas artes pollere, non magicas.

That Nero abandoned it is a vast and certain proof that the art is false. [...] So let us be persuaded that it is detestable, useless and empty. It does have some shadows of truth, but in those it is the art of drugs that has power, and not the art of magic.

If the magicians' herbs routed armies, Rome would not have needed siege engines at Carthage (Pliny NH 26.18–20); if a salamander quenched fire, fire-prone Rome would have used it (Pliny NH 29.74–76). Yet Pliny filled five books with magical remedies.

The poets mocked from inside the tradition. Horace counts laughing at "dreams, magic terrors, marvels, wise-women, ghosts of the night and Thessalian portents" among the marks of a free mind (Hor. Epist. 2.2.208–209), and his wooden Priapus scatters Canidia and Sagana with a fart, so that they drop false teeth and wig as they flee. Ovid, the fullest describer of rites, was also their plainest debunker. A girl's hair fell out because of her own dye and no rival's spell (Ov. Am. 1.14.39–46), love philtres only drive people mad (Ov. Ars 2.99–107), and Marsian chants split no snakes; the moon "will never be shaken from her horses", gongs or no gongs (Ov. Medic. 35–42). The gongs were real. Seneca and Martial describe the whole city beating bronze during a lunar eclipse to drown the witches dragging the moon down (Sen. Phaedr. 785–794; Mart. 12.57.15–17), while Pliny praised astronomers who predicted eclipses to the hour (Pliny NH 2.53–55).

Satire made magic a market of dream-expiators, philtre-sellers and astrologers (Mart. 7.54; Juv. 6.592–626). The particular tricks exposed by the medical writer Celsus and by Livy belong to chapter XIII. Some doubters half-believed. Tacitus blames failed predictions on incompetent astrologers rather than on astrology (Tac. Ann. 6.22).

Change over time

Two shifts run through the whole span. Maleficium narrowed from "crime" to "sorcery". And the line between harmless folk healing and punishable magic moved inward, until what Plautus treated as a household expense and Pliny as a folly could cost a man his head. Christian poets kept the old vocabulary and reversed its values: for Prudentius around 402, augury, haruspicy, the Sibyl and astrology are all a demon's tools for leading souls astray (Prud. C. Symm. 2.887–895). That story belongs to chapter VII.

What is distinctive about Rome

Every culture in this report separates acceptable ritual from unacceptable. Rome did it unusually through law. Its first code names magical crimes, its poisoning statute stretched to spells, and its treason law made the stars a police matter. In Egypt and Mesopotamia ritual experts worked within the temple; at Rome the official experts, augurs, pontiffs and haruspices, belonged to the state, and the private specialist was suspect by definition. Its writers gave each foreign people its own art and took credit for ending Druidic human sacrifice (Pliny NH 30.12–13). And Cicero, Pliny and Ovid built a tradition of empirical doubt, with failed predictions, counterfactuals and natural causes found for supposed bewitchments, that has few parallels in the rest of the ancient library. The same writers preserved magic in bulk, and the same elite that banned it kept consulting it. Tacitus' line about astrologers, always forbidden and always kept, fits Roman magic as a whole.

Chapter VII

Early Christianity: miracle, magic and the demons

From the gospels to Bede, Christians mostly took it for granted that magic worked, traced its power to demons, and spent seven centuries deciding which of their own healings, blessings and lots stood on the right side of the line.

A bonfire at Ephesus

The Acts of the Apostles, written late in the first century, sets one of its liveliest scenes in Ephesus, a city famous for spells. Cloths that have touched Paul cure the sick. Seven sons of a Jewish chief priest named Sceva, travelling exorcists, try the formula "I adjure you by the Jesus whom Paul preaches" on a possessed man; the spirit replies that it knows Jesus and Paul but not them, and the man beats them out of the house naked (Acts 19.13–16). Believers come forward to confess their practices, and then:

ἱκανοὶ δὲ τῶν τὰ περίεργα πραξάντων συνενέγκαντες τὰς βίβλους κατέκαιον ἐνώπιον πάντων· καὶ συνεψήφισαν τὰς τιμὰς αὐτῶν καὶ εὗρον ἀργυρίου μυριάδας πέντε.

A good many of those who had practised the curious arts brought their books together and burned them in front of everyone. They added up what the books were worth and found it came to fifty thousand pieces of silver.

Whether the fire happened cannot be checked, but the story holds this chapter in small. Christian power travels by touch and by a name. The same name fails in the mouths of people without the right allegiance. And magic is a thing with a price, written down, owned, and destroyed in public by the convert. The phrase τὰ περίεργα, "the over-elaborate things", was a common Greek euphemism for sorcery, and nobody in the story doubts that the books worked.

The words for it

Greek-speaking Christians inherited a ready vocabulary. A μάγος (magos) was first a Persian priest, and Matthew's magoi who follow a star are honoured visitors (Matt 2.2); in Acts the word marks a rival. A γόης (goēs) was a sorcerer as cheat, and the Pastoral Epistles warn that goētes "will go from bad to worse, deceiving and deceived" (2 Tim 3.13). Φαρμακεία (pharmakeia), drug-craft, covered medicine, poison and spell in one word. Paul lists it among the "works of the flesh" next to idolatry (Gal 5.20), and the church manual called the Didache, of disputed date around 100, sets "you shall not practise magic, you shall not practise pharmakeia" directly before abortion, which suggests that drugs to end pregnancy were part of the meaning (Did. 2.2). Against these stood the approved words δύναμις, σημεῖα and τέρατα, "power, signs and wonders". Mark has Jesus feel dynamis leave him when a sick woman touches his clothes (Mark 5.30).

Latin writers used magus and ars magica, but their sharpest word was maleficus, "evil-doer", which became the legal name for the sorcerer and his crime, maleficium. Around it gathered the praecantator who sang charms over the sick, the sortilegus who cast lots, the ariolus who diagnosed and foretold, the mathematicus, who was an astrologer, and in sixth-century Gaul the caragius, a local word whose sense is lost. Their objects were ligaturae, things tied on the body, and phylacteria, amulets. Jerome's definitions on Daniel were copied for centuries. Enchanters, he says, work "by words", magi "philosophise about each thing", and malefici "use blood and victims, and often touch the bodies of the dead" (Hier. In Dan. 1.2.2).

In the Christians' own terms the deepest division ran between the power of God and the work of demons. A technique that works by itself hardly existed for them. Someone always acted behind the rite, and the question was who.

Jesus accused

The earliest charge of magic in this literature is aimed at Jesus, and it comes from inside the gospels:

καὶ οἱ γραμματεῖς οἱ ἀπὸ Ἰεροσολύμων καταβάντες ἔλεγον ὅτι Βεεζεβοὺλ ἔχει, καὶ ὅτι ἐν τῷ ἄρχοντι τῶν δαιμονίων ἐκβάλλει τὰ δαιμόνια.

The scribes who had come down from Jerusalem kept saying, "He has Beelzebul," and, "It is by the prince of the demons that he drives the demons out."

Nobody in the scene doubts that the exorcisms work. The quarrel is over the source of the power. Jesus' retort in Matthew, "by whom do your sons cast them out?", assumes that other Jewish exorcists were at work and were accepted (Matt 12.27), and in Mark the disciples find a stranger already using Jesus' name against demons (Mark 9.38). Some of Mark's cures follow a procedure. Jesus takes a deaf man aside, puts his fingers in his ears, spits, touches his tongue, groans, and says the Aramaic word Ephphatha, "Be opened" (Mark 7.34). Many modern scholars think Matthew and Luke dropped this story because it looked too much like a magician's routine; that is an inference from their silence.

Outsiders said it plainly. The pagan philosopher Celsus, around 175–180, gave the charge to a Jewish speaker, and Origen preserved it seventy years later in order to answer it:

καὶ ὅτι οὗτος διὰ πενίαν εἰς Αἴγυπτον μισθαρνήσας κἀκεῖ δυνάμεών τινων πειραθεὶς, ἐφ᾿ αἷς Αἰγύπτιοι σεμνύνονται. ἐπανῆλθεν ἐν ταῖς δυνάμεσι μέγα φρονῶν, καὶ δι᾿ αὐτὰς θεὸν αὑτὸν ἀνηγόρευσε.

…and that because he was poor he went to Egypt to work for hire, and there tried his hand at certain powers the Egyptians pride themselves on. He came back full of himself on account of these powers, and on their strength proclaimed himself a god.

Egypt was the classic school of magic, and Celsus compared Jesus' wonders with those of Egyptian-trained performers who sell their lore in the marketplace "for a few obols", drive out demons, blow diseases away and lay out banquets that do not exist (Orig. Cels. 1.68). The charge was older than Celsus, for Justin, in Rome around 155, already rebuts the question "what prevents the one we call Christ from having done his powers by magic art?" (Just. 1 Apol. 30.1). It also lasted. Carthaginian pagans around 411 still ranked Apollonius and Apuleius above Christ, which Augustine found barely "worth a laugh" (Aug. Ep. 138.18). Julian granted the cures and dismissed them as village work (Jul. Gal. 191E).

How Christians answered

Since both sides agreed that wonders happened, Christians needed tests for telling a miracle from a magical act, and each test drew the line in a different place. The first was the formula:

οὐ γὰρ κατακηλήσεσιν ἰσχύειν δοκοῦσιν ἀλλὰ τῷ ὀνόματι Ἰησοῦ μετὰ τῆς ἀπαγγελίας τῶν περὶ αὐτὸν ἱστοριῶν.

They do not appear to prevail through spells. They prevail through the name of Jesus, together with the telling of the stories about him.

Origen admits that the name sometimes works "even when named by bad men", which makes the line very fine. The second test was the apparatus. Eusebius challenged anyone to name "sorcerers and drug-workers who practise their sorcery without libations and sacrifices and the summoning and presence of demons"; Jesus used none of these, so he was no goēs (Eus. Dem. ev. 3.6.34). Others looked at results. The Pseudo-Clementine Homilies call Simon's walking statues "useless" signs because they heal nobody (Ps.-Clem. Hom. 2.32–34). The last argument was victory. Ignatius of Antioch wrote around 110 that at the Nativity "all magic was dissolved, and every bond of wickedness vanished", using the word for a binding spell (Ign. Eph. 19.3), and Athanasius in Egypt invited doubters to make the sign of the cross and watch "all magic and sorcery" become powerless (Athan. Inc. 48.3).

Simon Magus and the magician-heretic

If Jesus was the accused, Simon of Samaria was the accusers' ideal type:

Ἀνὴρ δέ τις ὀνόματι Σίμων προυπῆρχεν ἐν τῇ πόλει μαγεύων καὶ ἐξιστάνων τὸ ἔθνος τῆς Σαμαρίας, λέγων εἶναί τινα ἑαυτὸν μέγαν,

There was a man named Simon who had for some time been practising magic in the city and amazing the people of Samaria, claiming that he was someone great.

Simon is baptised, sees the apostles give the Spirit by laying on hands, and offers money for the same power. Peter's curse gave the church the word simony and one of its lasting rules, that a divine gift cannot be bought. A few chapters later Paul strikes the court magus Elymas blind, "not seeing the sun for a time" (Acts 13.11). In form this is a curse that disables a rival, exactly what a curse tablet was for; Luke calls it "the hand of the Lord".

By the second century Simon had become the father of heresies. Justin says he was honoured in Rome with a statue inscribed SIMONI DEO SANCTO (Just. 1 Apol. 26.1); modern scholars usually think Justin misread a dedication to the Sabine god Semo Sancus. Every rival teacher could be cast this way. Irenaeus accused the Gnostic Marcus of giving women "love-philtres and attraction spells" (Iren. Haer. 1.13), and the Refutation of All Heresies attributed to Hippolytus, whose authorship is disputed, explains how Marcus turned the eucharistic cup red with a hidden drug while he "babbled at length" ([Hippol.] Ref. 6.39). In 385 the charge became lethal, when Priscillian, bishop of Avila, was "convicted of maleficium" at Trier and executed, the first Christian, it is usually said, to be put to death by a Christian government on a religious charge (Sulp. Sev. Chron. 2.50.8).

Who practised, and what they did

The sources name Jewish exorcists, court magi, a slave girl with "a python spirit" whose fortune-telling earned her owners "much profit" (Acts 16.16), curse-specialists hired for the chariot races, philosophers, senators, and in the sermons and councils of sixth-century Gaul and Spain a whole village profession of charmers, lot-casters and diviners. Women appear mostly as alleged targets of love magic, and clergy often enough that councils had to forbid them to make amulets; the Coptic spell-books bear out both (see 'The spell-books of Christian Egypt' below). On the other side stood Jesus, the apostles, the elders who were to pray over the sick and anoint them "in the name of the Lord" (Jas 5.14), and later monks, bishops and saints' shrines, whose actions closely mirror the ones they condemned.

Exorcism and names

The Gerasene demoniac tries to bind Jesus with the exorcist's own formula, naming him exactly and saying "I adjure you by God" (Mark 5.7). Names crossed religious lines. Origen noticed pagan enchanters using "the God of Abraham" without knowing who Abraham was (Orig. Cels. 1.22), and the late Acts of Philip has the apostle cry "Zabarthan, sabathabat, bramanouch, come quickly!", words like the nonsense names of the magical papyri, while his opponent shouts that this is "true sorcery" (Acta Philippi 23).

Healing, amulets and the preachers

Amulets were where bishops met daily life, and where they lost most often. The Council of Laodicea, probably in the 360s, forbade clergy to "make what are called phylacteries, which are in truth chains for their own souls" (Conc. Laodicea can. 36). Augustine settled for less: "When your head aches, we praise you if you put the Gospel to your head instead of running to a ligature" (Aug. In Ioh. tr. 7.12). The fullest picture comes from Caesarius, bishop of Arles in the early sixth century, whose sermons the library files under Augustine's name. He imagines the sick-bed of a Christian surrounded by helpful neighbours:

Mitte ad illum divinum, transmitte ad illum cingulum aut fasciam tuam, mensuretur et aspiciat; et ipse tibi dicet quid facies,aut utrum evadere possis. Dicit et alius Ille bene novit fumigare

"Send to that diviner. Send him your belt or your headband to be measured and looked over, and he will tell you what to do, and whether you can pull through." Another says, "That man really knows how to fumigate."

Caesarius calls refusing such help a kind of martyrdom. He also lets his parishioners answer back: "What are we to do, since these auguries, and the caragi and diviners, often tell us true things?" (Caes. Arel. Serm. 54). Gregory of Tours shows the same contest from the saint's side. Arioli diagnose a collapsed peasant woman with "the noonday demon" and bring "ligatures of herbs and words of incantations"; his niece takes them off and pours oil from St Martin's tomb into her mouth (Greg. Tur. Virt. Mart.). Yet the word "phylactery" never became purely negative. Around 603 Pope Gregory the Great sent a gift to the Lombard boy-king:

Excellentissimo autem filio nostro Adulouvaldo regi transmittere e phylacteria curavimus, id est cru- cern cum ligno sancte crucis Domini, et r lectionem sancti Evangelii g theca Persica inclusafu.

To our most excellent son King Adaloald we have taken care to send phylacteries, that is, a cross holding wood of the Lord's holy Cross, and a reading from the holy Gospel enclosed in a Persian case.

Laodicea's "chains of the soul" and the pope's gift share a name and a shape, a case worn on the body holding something powerful. What separated them was the content and the giver.

Divination and lots

Christians had a lot-oracle from the start, for the apostles cast lots for Judas' replacement and "the lot fell on Matthias" (Acts 1.26). Opening the Bible for omens, the "lots of the saints", was condemned at Agde in 506 for "clerics and laymen" alike (Conc. Agde can. 42). Seventy years later Gregory of Tours tells, without disapproval, how the prince Merovech laid the Psalter, Kings and the Gospels on St Martin's tomb and kept vigil for a sign; in the same chapter Gregory laughs at a pythoness and opens Proverbs at random himself (Greg. Tur. Hist. 5.14). Divination about the throne was treason, as the trial of the laurel tripod that spelled out ΘΕΟΔ in 371 showed (Soz. HE 6.35; see chapter VI).

Love, binding and the dead

Jerome's Life of Hilarion tells of a young man of Gaza, rejected by a consecrated virgin, who studied a year at Memphis and buried "monstrous figures engraved on a plate of Cyprian copper" under her threshold, so that she tore off her veil and shouted his name (Hier. V. Hil. 21). When a Syrian noblewoman believed a concubine had bewitched her husband, the monk Aphrahat "dimmed the working of the sorcery" by prayer and gave her blessed oil, and the husband's love returned (Thdt. HR 8.18). The holy man did what a love charm promised, inside a marriage and by approved means.

Few Christians doubted that necromancy worked. Justin cited necromancy as proof that souls survive death (Just. 1 Apol. 18). Origen held that the woman of Endor really raised Samuel, against fellow Christians who said "I do not believe the ventriloquist" (Orig. De engastr. 3). Rufinus has the Egyptian abbot Macarius call a murdered man by name at his grave and hear him answer that the accused was innocent (Rufin. Hist. mon. 28), a necromancy used as legal evidence. Transformations were treated more warily. Augustine heard of Italian innkeeper women who turned travellers into pack animals with something in cheese, and decided that the victim's dream-image appeared to others as a beast while his drugged body lay asleep (Aug. Civ. 18.18).

How it was supposed to work

The master explanation was demonic. Tertullian, at Carthage around 200, retold the Jewish myth of the angels who lusted after women and taught them metals, "the powers of herbs", "the force of incantations" and the reading of the stars (Tert. Cult. fem. 1.2). His source was the Book of the Watchers, where Azazel teaches metals, bracelets and eye-paint and Semjaza "enchantments, and root-cuttings" (1 Enoch 8.1; 8.3; see chapter III). Tertullian names the book and knows that some reject it because the Jews do not admit it to their scriptures (Tert. Cult. fem. 1.3), and his calliblepharum, eye-paint, is the καλλιβλέφαρον of the Greek Enoch (Cult. fem. 2.10; 1 Enoch 8.1 Greek). The fallen angels' heirs, the demons, have airy bodies and fly everywhere at once, so they know distant news and feel rain coming. Their cures are a confidence trick:

Benefici plane et circa curas valitudinum. Laedunt enim primo, dehinc remedia praecipiunt ad miraculum nova sive contraria, post quae desinunt laedere, et curasse creduntur.

They are benefactors, to be sure, where health is concerned. First they do harm; then, to make a marvel of it, they prescribe remedies, novel or contrary ones. After that they stop doing harm, and people believe they have cured.

Beside the demons ran theories of signs and names. The Syrian Tatian said roots, sinews and bones have no power of their own; they work like letters, by a convention the demons fixed (Tatian Or. 17). Origen took the Stoic view that names belong to things by nature and defended magic as "a consistent system with principles known to very few", whose names lose their force in translation (Orig. Cels. 1.24).

Augustine's theories

Augustine of Hippo (354–430) made these scattered arguments into a system that the Latin Middle Ages inherited almost whole. He gave demons a physics. Their keen senses, speed and centuries of experience let them "seem to divine" as old men, doctors and sailors do to the inexperienced (Aug. Div. daem. 3.7). They cannot create; when Pharaoh's magicians made frogs and serpents, demons only hurried along "certain hidden seeds" already in the elements (Aug. Trin. 3.8.13).

He also gave magic a theory of signs. Superstition covers "consultations and pacts of signification agreed with demons", including amulets their users call physica, "natural remedies", to make them sound respectable (Aug. Doctr. chr. 2.20.30). Omens work like the letter X, which means one thing to Greeks and another to Latins, so people made them valid by noticing them, "and so they turn out differently for different people" (Aug. Doctr. chr. 2.24.37). This shared code with demons is the root of the later doctrine of the demonic pact. It yielded a practical test for medicine:

Aliud est enim dicere, Tritam istam herbam si biheris, venter non dolebit; et aliud est dicere, Istam herbam collo si suspenderis, venter non dolebit. Ibienim probatur contemperatio salubris, hic significatio superstitiosa damnatur.

It is one thing to say, "If you crush this herb and drink it, your stomach will stop aching," and another to say, "If you hang this herb round your neck, your stomach will stop aching." In the first a wholesome mixture is approved; in the second a superstitious signification is condemned.

A remedy that acts on the body is nature; one that acts by being worn is a message to someone. Augustine admitted hard cases and left them to the user's intention. Finally he refused the Platonists' distinction between base goetia and honourable theurgia, ritual union with the gods, treating "magia, or by a more detestable name goetia, or by a more honourable one theurgia" as labels for one cult of demons (Aug. Civ. 10.9).

Isidore's taxonomy

Around 620 Isidore of Seville gathered all this into the chapter "On the magi" of his Etymologies, the definition the early Middle Ages used. Magic, "handed down by the evil angels", ruled the world for ages, and haruspicy, augury, oracles and necromancy are its inventions (Isid. Etym. 8.9.3). Then comes the decisive equation:

Magi sunt qui vulgo malefici ob facinorum magnitudinem nuncupantur. IIi et elementa concutiunt, turbant mentes hominum, ac sine ullo veneni haustu violentia tantum carminis interimunt.

Magi are those whom people commonly call malefici, evil-doers, because of the scale of their crimes. They throw the elements into turmoil, unsettle people's minds, and kill by the sheer violence of a spell, with no poison drunk at all.

A list follows: diviners by the four elements (8.9.11–13), sortilegi with their "lots of the saints" (8.9.28), amulet-makers whose ligatures "the medical art condemns" (8.9.30), and the praestigium, the trick that "dazzles" the eyes (8.9.33). Because the library holds his sources, the seams show. The magi and malefici come from Jerome on Daniel, the ligatures from Augustine's On Christian Teaching, the sortilegi almost word for word from canon 42 of Agde, and the elemental divination from Augustine's reading of Varro. Isidore's taxonomy carried authority because every piece of it had already been ruled on.

Councils, emperors and kings

Roman law had always punished harmful magic, and Constantine at first kept the old line between public and private. In 319 he forbade any haruspex to "cross another's threshold" on pain of burning while leaving public rites alone (CTh 9.16.1). In 321 he punished magic aimed at health or chastity and protected the rest:

Nullis vero crimina- tionibus implicanda sunt remedia b humanis c quae- sita corporibus aut in agrestibus locis, ne maturis d vindemiis metuerentur imbres, aut ruentis e grandi- nis lapidatione quaterentur

But no charge is to attach to remedies sought for human bodies, or to measures taken in the countryside so that ripe grapes need not fear the rains or be battered by a stoning of hail.

The stray letters are the printed edition's footnote marks, kept as they stand in the library's text. The Visigothic interpretation of the same law in 506 drops the exemption and adds "storm-senders" to the criminals. Within two centuries the healing charm and the hail rite had moved from protected to punishable. Church penalties ran alongside.

The canons punish clients as well as practitioners, and the charge could be turned against the weak. When Queen Fredegund's son died of dysentery in 584, she had Parisian women tortured until they confessed to killing him by sorcery, then executed them; Gregory of Tours adds that part of the confession was something "which I cannot by any means allow myself to believe" (quod nulla ratione credi patior) (Greg. Tur. Hist. 6.35).

Doubt, mockery and exposure

Most Christian writers held two positions at once. Magic was real because demons were real, and particular wonders were often fakes. A few went further. Theophilus of Alexandria attacked Origen for having written that "the magic art does not seem to me the name of any really existing thing; but even if it were, it is not an evil work", a sentence that survives only because an opponent quoted it (Hier. Ep. 96.16). Tertullian himself asks what magic is and answers "what nearly everyone says, a deception" (Tert. An. 57). Eusebius noted that oracles' rare hits are carved on stone while their misses are forgotten, like marvelling when two people drawing lots from ten thousand get the same number (Eus. PE 4.2.2). The physician Vindicianus told Augustine that astrologers' hits were chance (Aug. Conf. 4.3.5). Cassiodorus, writing for Theodoric about a charioteer whom fans called a sorcerer, explained that magic gets the credit "when victory cannot be credited to the horses" (Cassiod. Var. 3.51).

Exposures of temple machinery after 391 (Rufin. HE 11.23; Aug. Civ. 21.6) and the Hippolytan catalogue of conjuring tricks ([Hippol.] Ref. 4.28) are treated in chapter XIII. Christian impostors got the same scrutiny; one wandering holy man's bag held herb roots, moles' teeth and mouse bones (Greg. Tur. Hist. 9.6). Debunking left the demonic theory intact and gave Christians a way to win arguments while keeping the demons in reserve.

Change over time

The first-century texts already include the basic categories of magic and pharmakeia in vice lists, exorcism "in the name", the honoured Magi and the magician-heretic. Second-century apologists built the demonic theory while Celsus systematised the charge against Jesus. After 312 Christianity gained the state. Eusebius accused the defeated Maxentius of divining from the entrails of babies (Eus. HE 8.14.5), a stock charge against fallen rulers. Law and canons took over, and Augustine gave the West its theory. From about 500 the target moved from temples and philosophers to the village, and the vocabulary grew local words such as the Sicilian canterma (Greg. Reg., to Cyprian), while Christian counter-practices multiplied, among them tomb oil and dust, Gospel books on aching heads and relic phylacteries.

Bede, writing in 731, marks the end of the period. When the mission of Augustine of Canterbury landed in Kent in 597, King Æthelberht met them in the open air, "following an ancient augury", in case they had "any magic art" that would work indoors (Bede HE 1.25). In the plague of the 660s baptised Northumbrians ran back to "incantations or phylacteries" (Bede HE 4.27). And when the chains of a captured thegn named Imma kept falling off whenever his brother said Mass for him, his captor asked the obvious question:

an forte litteras solutorias, de qualibus fabulae ferunt, apud se haberet, propter quas ligari non posset. At ille respondit nil se talium artium nosse

…whether he perhaps had about him loosening letters, of the kind the stories tell of, that made it impossible to bind him. He answered that he knew nothing of such arts.

Scholars often link the "loosening letters" with runic charms, though Bede does not say so. Bede gives a known magical effect a Christian cause, much as the gospel writers had claimed exorcism for the Spirit of God.

What is distinctive here

Taking up the Jewish story that fallen angels taught mankind sorcery (1 Enoch 7.1; see chapter III), this is the first body of texts in the library to apply that one cause people by people, with Persian magi, Egyptian priests, Etruscan haruspices, Greek theurgists, Roman state cult and Germanic charms all explained by the same demons. It fused magic with heresy, so that a rival teacher was by definition a magician. It appealed unusually often to physicians as judges of a lawful remedy, and its sermons describe village practitioners and their clients in close detail. Above all, its holy men did much of what the condemned specialists did. They cast out demons by names, healed with touch and oil, sent protective objects, cast lots on tombs, made the dead speak and undid love spells. The texts place the difference in who acted and by whose power, and seldom in technique. That was the ground on which Celsus attacked Jesus, and it is the ground the Christian authors chose to hold.

Chapter VIII

The North: seiðr, galdr and the runes

In the Old Norse and Old Swedish texts of the library, magic is a learnable craft of song, carving and trance, sharply divided by gender and ethnicity, and told mostly by Christian Icelanders looking back on a pagan past they both remembered and reshaped.

A song outside the door

In western Iceland, some time before the year 1000 as the saga reckons it, a family of Hebridean immigrants came by night to the farm of the old chieftain Hrútr. Kotkell, his wife Gríma and their sons had already chanted a storm onto the sea from a raised platform and drowned Guðrún's husband (Laxd. 35). Now they turned on Hrútr's household.

Þau fóru á bǿ Hrúts. ok gerðu þar seið mikinn. En er seiðlǽtin kómu upp, þá þóttusk þeir eigi skilja, er inni váru, hverju gegna mundi. En fǫgr var sú kveðandi at heyra.

They went to Hrútr's farm and worked a great seiðr there. When the seiðr-sounds rose, the people indoors could not make out what was happening. But the chanting was lovely to hear.

Only Hrútr knew what the sound meant. He ordered everyone to stay awake and not look out, but sleep came over them all. His twelve-year-old son Kári, the one the rite was aimed at, went out into the singing and fell dead. Hrútr and his kinsman Óláfr pái hunted the family down. Kotkell and Gríma were stoned to death and buried under a cairn. Their son Hallbjǫrn Whetstone-eye had a bag pulled over his head so that his gaze could do no harm; when it was taken off at sea before he was drowned, he looked at the land and cursed it, "and the look of his eyes was not good" (Laxd. 37).

The scene holds much of this chapter. Outsiders work a named technique through sound and sleep, the eye is dangerous, and the community kills without trial, because, as the saga had said after the storm, "people thought those who practised such sorcery deserved to die" (Laxd. 35).

It also shows the problem of the evidence. Laxdæla saga was written around 1245, two and a half centuries after the events, by a Christian for Christians. A few skaldic verses and Völuspá may reach back to pagan times; the rest is thirteenth-century sagas and histories, Norwegian Christian laws, sermons, translations, and late Swedish devotional texts.

The words for it

Old Norse has no single word for what we call magic. The most general term is fjǫlkynngi, "much-knowing", and a person skilled in it is fjǫlkunnigr. The word puts knowledge at the centre. Gunnhildr, found in a Sámi hut, says she has come "to learn arts" from two Finns who are the wisest in the region (HHárf. 32), and Snorri says that all the sorcery of the North descends by teaching from Óðinn and his priests and "has spread widely and lasted long" (Yng. 7). Within that field the sources name particular crafts.

Some of these words were ordinary before they were magical. In the Norwegian Strengleikar, a woman travelling alone hears dogs barking far off and the hana galdr, the "galdr of the cock", which is simply cock-crow (Strengleikar, 22v.23). Gerningar means only "deeds" in the Norwegian sermons, so King Sverrir's Christian law has to say "evil gerningar" when it means sorcery (Sverris kristinréttr 79).

Who practised magic

Óðinn and Freyja

The oldest datable mention of seiðr in the library is a half-stanza by the skald Kormákr, composed in the 960s and quoted in the Codex Wormianus. It alludes to Óðinn working seiðr toward Rindr, the woman he had to win to father an avenger for his son Baldr (Skáldsk., Kormákr, Sigurðardrápa 3). Its syntax is notoriously difficult.

The fullest account is Snorri Sturluson's in Ynglinga saga, written around 1230, where Óðinn is a human chieftain from Asia whose skills made people take him for a god. When he changed shape, "his body then lay as if asleep or dead" while he travelled as a bird or beast, fish or serpent, and "with words alone" he could quench fire, still the sea and turn the winds (Yng. 7). He made the embalmed head of Mímir speak by galdrar (Yng. 4) and woke the dead from their mounds (Yng. 7). He taught these arts "with runes and with songs, those that are called galdrar", and so the Æsir are called "smiths of galdrar" (Yng. 7). Snorri gives seiðr a separate, female origin among the Vanir, the other family of gods.

Dottir Nia-rþar var Freyia, hon var blot-gyðia hon kende fyrſt með Áſom ſeið ſem Vǫnom var títt.

Njǫrðr's daughter was Freyja. She was a priestess of sacrifice, and she was the first to teach seiðr among the Æsir, as was the custom among the Vanir.

Óðinn learned it, Snorri says, and it was the art "which carries the greatest power". It could reveal men's fate, cause death, misfortune or sickness, and take wits or strength from one person and give them to another. "But such great ergi follows this sorcery when it is performed that men did not think it fitting to deal with it, and so the art was taught to the priestesses" (Yng. 7). Ergi was the gravest insult a man could receive: unmanliness, with a strong suggestion of sexual passivity.

The seeresses

The most authoritative magical voice in the library is a woman's. In Völuspá, usually dated around 1000 though the date is disputed, a vǫlva tells Óðinn the history of the world. The frame is a paid consultation. She "sat out alone" when the old god came and looked into her eyes, and he "chose for her rings and necklaces" in return for her speech (Vsp. 28–29). After the gods have three times burned a woman called Gullveig in Óðinn's hall, and she still lives (Vsp. 21), the poem portrays a wandering seeress.

Heiði hana hétu hvar til húsa kom völu velspá vitti hon ganda seið hon kunni seið hon leikinn æ var hon angan illrar brúðar.

Heiðr they called her, wherever she came to a house, a seeress who foretold well. She worked with wands. She knew seiðr; in seiðr she was entranced. She was always the delight of a wicked woman.

Almost every word is disputed. Ganda may mean staffs or spirits, and leikinn "skilled" or "possessed"; Hauksbók reads hugleikin, "mind-entranced" (Vsp. 22 (Hauksbók)), and modern readings of seiðr as trance lean on that variant. The figure is clear enough, a woman who goes from house to house, paid for what she knows.

The sagas fill out the type. Huld the seiðkona is hired by a Finnish wife to draw the faithless Swedish king Vanlandi back to her or kill him. He feels an overpowering longing for Finland, falls asleep, and cries out that a mara, a nightmare-being, is trampling him; when his men hold his head it tramples his legs, and when they hold his legs it crushes his head until he dies (Yng. 13). The seeress persisted even beneath the church. Late in Laxdæla saga, Guðrún has the floor dug up where she knelt to pray, and under it lie blackened bones, a brooch and a great staff, "and people thought they knew that a seeress's grave had been there" (Laxd. 76).

Women of the household, and a queen

Many saga practitioners are not professionals. Þorgerðr brák, the slave-woman who fostered Egill, was "strong as a man and very skilled in magic" (Egils saga 40). The widow Þórveig, whose sons Kormákr had killed, cursed him so that he "should never enjoy Steingerðr", and the narrator explains his later coldness toward his bride by her seiðr (Kormáks saga 5). At the top of society stands Gunnhildr, queen of Eiríkr Bloodaxe, trained by Sámi masters. In Egils saga she has seiðr worked so that Egill "should never find rest in Iceland until she saw him" (Egils saga 59).

Men, shame and the fire

Men did practise seiðr, and the texts treat them with contempt or violence. The sharpest episode concerns a son of Haraldr Fairhair.

hann nam fiolkyngi oc gerðiz ſeiðmaðr . Haralldi konongi þotto illir ſeiðmenn. ... hann breɴdi iɴi Rognvald bróðor sinn með Lxxx ſeið manna oc var þat verc lofat mioc.

He learned sorcery and became a seiðr-man. King Haraldr thought seiðr-men evil. ... [Eiríkr] burned his brother Rǫgnvaldr in his house, together with eighty seiðr-men, and that deed was much praised.

Between those sentences another seiðmaðr, Vitgeirr, answers the king's order to stop with a verse: small harm if the children of old men and old women do seiðr when the king's own son does it (HHárf. 34). Snorri sets the hostility under a pagan king. Whether that reflects pre-Christian attitudes or a Christian writer projecting his own cannot be settled from the texts.

Men's rune-magic, by contrast, was honourable. In Rígsþula, which tells how slave, farmer and earl came to be, the boy Konr ungr (his name plays on konungr, "king") surpasses his father because he knows "runes of eternity and runes of life": he can save men, blunt sword-edges, calm the sea, understand birds, quench fires and allay sorrows (Rígsþula 43–44). Rune-knowledge makes the king. The poem's date is disputed (tenth to thirteenth century), but it is the library's best list of what runes could do.

The Sámi and other outsiders

The texts repeatedly place magic at the edge of Norse society. The Finnar of the sagas are mainly the Sámi of the far north. Gunnhildr's two Sámi masters track like dogs, ski so well that neither man nor beast escapes them, and when they are angry "the earth turns over before their gaze, and if any living thing comes before their gaze it falls dead" (HHárf. 32). Eyvindr kinnrifa, tortured for refusing baptism, says he is "a spirit quickened in a human body by the sorcery of Finns" (ÓT 76).

What they did

Foreknowledge

Knowing the future was the most respectable aim of Northern magic. The seeress of Völuspá "sits out", the posture the later laws call útiseta, sitting out at night to learn fate or wake the dead. When the Althing split between pagans and Christians around 1000, the pagan lawspeaker Þorgeirr "lay down and spread his cloak over himself" for a day and a night without a word, then proclaimed that all should be Christian, though people might still sacrifice in secret (Íslendingabók 7). Many scholars read the cloak as a seer's seclusion; Ari, writing around 1125, says nothing of the kind.

Compulsion at a distance

The usual work of seiðr in the sagas is to make someone long for a place, lose desire, find no rest or die, as with Vanlandi, Kormákr and Egill. A twelfth-century Norwegian law shows what such harm was thought to leave behind. If a destroying witch's kit is found "in people's beds or bolsters", hair, nail-clippings, perhaps frogs, perhaps body parts, "or those things thought fit for sorcery", the accused must carry hot iron; the worst fordæða is she "who destroys a cow or calf, a woman or a child" (Borgarþ. kristinréttr I 16). Several readings are uncertain.

Runes

Rune magic is literate magic, and its texts insist on competence. Served a poisoned drink, Egill stabs his palm, carves runes on the horn, reddens them with his blood and speaks a verse, and the horn bursts (Egils saga 44). Later he visits a farm in Värmland where the daughter, Helga, has long been ill and sleepless.

þar fann hann talkn, ok váru þar á rúnar. Egill las þǽr, ok síðan telgði hann af rúnarnar. ok skóf þǽr í eld niðr. ... Skalat maðr rúnar rísta, nema ráða vel kunni

There he found a piece of whalebone with runes on it. Egill read them, then whittled the runes away and scraped them down into the fire. ... "A man should not carve runes unless he knows well how to read them."

A farmer's son had tried to carve love-runes for her and got them wrong. Egill burns the bone, has her bedclothes aired, and puts new runes under her pillow, and she wakes as if from sleep. Fresh bedding and ritual go together in the cure, and the saga does not separate them; the panel reads the illness and recovery as distress and reassurance (see chapter XII). Its lesson is that runes are a skill, and a skill done badly does harm.

Scorn that strikes

Níð was defamation, usually sexual insult, in verse or carved image. It was a legal offence, and in the sagas it works as a curse. Outlawed in Norway by Eiríkr and Gunnhildr, Egill set up a pole with a horse's head on it.

Hér set ek upp níðstǫng . ok sný ek þessu níði á hǫnd Eiríki konungi ok Gunnhildi dróttningu. Hann sneri hrosshǫfðinu inn á land. Sný ek þessu níði á landvǽttir þǽr, er land þetta byggja.

"Here I raise a scorn-pole, and I turn this scorn against King Eiríkr and Queen Gunnhildr." He turned the horse's head toward the land. "I turn this scorn against the land-spirits who live in this land."

The landvættir, spirits of hills, rocks and waterfalls, are to wander lost until they drive the king and queen out. Egill also carved the formula in runes on the pole, joining speech, writing and image in one act, and Eiríkr was soon driven from Norway. Snorri records níð as state policy. After the Danish king seized an Icelandic ship, "it was made law in Iceland that a scorn-verse should be composed about the Danish king for every nose in the land" (ÓT 33). The king's spy, sent in the shape of a whale, was driven off by the island's land-spirits (ÓT 33).

Wind, fog and shape

In a seafaring society weather was an obvious field for magic. Kotkell's family chanted galdrar from a seiðr platform, so the two native categories could be used together. The chieftain Rauðr the Strong, a great sacrificer with many Sámi followers, "always had a fair wind wherever he wanted to sail, and that came from his sorcery" (ÓT 78). Eyvindr kelda landed a shipload of seiðmenn under a fog so thick the king could not see them, but at daybreak the fog turned on its makers and they wandered in circles until caught (ÓT 63).

The idea of the hamr, the "skin" or shape a person wears, explains a family of phenomena. A man in battle-fury was said to hamask, to "go into his shape". Óðinn's warriors went without mail, "mad like dogs or wolves", and neither fire nor iron harmed them, which Snorri calls berserksgangr (Yng. 6). Egils saga adds an almost clinical note: "it is said" that while the fury lasted nothing could withstand such men, "but as soon as it had passed they were weaker than usual" (Egils saga 27).

Objects with a nature

Some magic lived in things. The sword Skǫfnungr had a "nature" (náttúra): the sun must not shine on its hilt, it must not be drawn where women are present, and its wounds healed only if its own healing-stone was rubbed on them (Laxd. 57). The most interesting case attaches magic to a real traded object. The Norwegian chieftain Þórir hundr had twelve reindeer-hide coats made by the Sámi "with such great sorcery that no weapon could take hold on them, far less than on a ring-mail coat" (ÓH 193). At Stiklestad in 1030 he wore one against King Óláfr himself.

Oláfr konongr hio til Þóris hundz um herþarnar ſverþit beit ekki, eɴ ſva ſyndiz ſem dyſt ryki ór hreínbialbanom ... Mildr faɴ gerſt hve galdrar gramr ſialfr megin rammir fiolkuɴigra fiɴa fullſtórum barg þóri

King Óláfr struck at Þórir hundr across the shoulders. The sword did not bite, but it looked as if dust flew from the reindeer coat. ... The generous prince found out for himself how the mighty spells of the sorcerous Finns gave Þórir great protection.

The verse is by Sigvatr, Óláfr's own skald, composed within a decade or so of the battle, so the magical explanation is almost contemporary. In Snorri the king's marshal Bjǫrn then turns his axe and strikes with its back, a practical answer to armour that resists a blade, and Þórir is wounded on the hand, where the coat does not cover him. The earlier Legendary Saga of St Óláfr has wolfskin coats instead, and gives Bjǫrn a great club (Legendary Óláfs saga). Thick hide is a plausible soft armour against cuts, but nothing supports the saga's claim that it outdid ring-mail.

How it was supposed to work

The texts imply several theories at once. Magic is knowledge, learned and passed on. It is sound, chanted in galdr and seiðr. It is writing, if the runes are rightly carved. It rests on a self that can leave the body. It works through the gaze, which is why Hallbjǫrn's head was bagged. And it engages spirits who can be coerced, woken or sent. Against all of this stands luck. When the young Óláfr Haraldsson's fleet was caught by a storm the Finns had raised, "the king's hamingja proved stronger, as it often did, than the Finns' sorcery" (ÓH 9). Snorri asks the question outright. In Gylfaginning the Swedish king Gylfi wonders whether the Æsir's power "came from their own nature, or whether the divine powers they worshipped caused it" (Gylf. 1–2). Snorri's own answer, as we shall see, was a third possibility.

Where the line fell, and what was done

Kings and communities

In the sagas, punishment of sorcerers does not begin with Christianity. Eiríkr's burning of the eighty seiðmenn happens at a pagan court, and when the ancient Swedish king Aun, having sacrificed nine sons to Óðinn for extra decades of life, lay drinking from a horn like an infant, the Swedes refused to let him sacrifice the tenth (Yng. 25). The conversion kings turned hostility into policy. Óláfr Tryggvason decreed that all men "known and proven to deal in galdrar and gerningar, or seiðmenn" must leave the land, then invited them to a feast, got them drunk and set fire to the hall (ÓT 62). The one who escaped, Eyvindr kelda, was later caught in his own fog and drowned on a tidal skerry. Rauðr the Strong, refusing baptism, was tied down with a peg between his teeth while the king tried to force an adder into his mouth (ÓT, the death of Rauðr). Snorri reports these acts without condemning them.

The Christian laws

The Norwegian provincial Christian laws, compiled in the twelfth century and copied in the fourteenth, draw a graded line. The older Borgarþing law makes it "an unatonable deed to make journeys to the Finns to ask prophecy", but fines a woman three marks if she "bites a finger or toe off her child for long life" (Borgarþ. kristinréttr I 16). Consulting a foreign seer was far worse than a domestic rite. The same law attends to procedure. A woman reputed "troll-ish" could clear herself with the oaths of six women, or else leave the district with her goods, and the clause adds that "it is not she herself who causes it that she is a troll" (Borgarþ. kristinréttr I 16). The Frostaþing law outlaws "as a murderer" anyone who sacrifices to heathen spirits or deals in prophecy or sorcery, and those who house him; men cleared themselves with hot iron, women with boiling water (Frostaþingslǫg III 15). King Sverrir's law orders that a woman proved a "troll-woman or man-eater" be taken out to sea and her back broken, but requires the charge to come from three households, and turns her penalty on an accuser from a single household whose charge fails (Sverris kristinréttr).

In the 1260s and 1270s King Magnús the Law-mender codified the field. His Christian law counts as "heathen error" not only galdrar and prophecy but belief "in land-spirits, that they dwell in groves, mounds or waterfalls", and "sitting out to ask about fate" (Magnús, kristinréttr 1–2). The ban now reached belief about the landscape itself. His national law of 1274 lists magic among the crimes that cost a man his life.

En þeir er líf sitt láta fyrir þýfsku sakir eða útilegu ... ok svá fyrir morð ok fordǽðuskap, fyrir spár ok útisetur, at vekja troll upp, at fremja með því heiðni

And those who forfeit their lives for theft or banditry ... and likewise for murder and witchcraft, for prophecies and sitting out to wake up trolls, to practise heathenism by it.

Sermons, medicine and the licit counterparts

The Old Norwegian Homily Book of around 1200 puts women's remedies on the forbidden side.

lauſung. mein æiðar. mutu fe. ok trva a liflyf qvenna eða gerningaʀ. eða a ſpa ſꜵgumſpa ſꜵgur. þat er allt diofulſ craftr.

... loose living, false oaths, bribes, and believing in women's life-herbs or in sorcery or in soothsaying. All that is the devil's power.

The word lyf meant both medicine and charm. An Icelandic medical fragment of about 1300 tells the healer, to stop bleeding, to pound a red-seeded field herb in wine or ale "and sing the Pater noster meanwhile" (AM 655 XXX 4to, 1v), and cites Dioscorides (whose text in the library has no such story; the closest match is Galen's, see chapter V) for a peony root hung on an epileptic boy's neck, who fell into a fit whenever it dropped off (AM 655 XXX 4to, 3r–v). Amulet and prayer were acceptable with Latin learning and a Christian formula behind them.

The sagas show the Church fighting ritual with ritual. Against Rauðr's storm Bishop Sigurðr, in full vestments, brought candles, incense, a crucifix, the Gospel and holy water to the prow, and the ship rowed through a lane of calm (ÓT 79). In Iceland a berserk who "feared neither fire nor edge" waded through a pagan-consecrated fire but shrank from the missionary Þangbrandr's, and when the priest struck him with a crucifix "such a miracle (jarteign) happened" that his sword fell (Njáls saga 103). A feast in Trøndelag shows the ambiguity openly. When the Christian king Hákon made the sign of the cross over the toasting cup, Jarl Sigurðr told the suspicious farmers he had made the sign of Þórr's hammer (Hák. góða 17). The gesture was the same either way.

Doubt, mockery and the art of illusion

The sagas mostly tell magic as true, but they hedge with "it is said", and characters sometimes refuse to believe. Kormákr's foster-mother Þórdís, a seeress, secretly tries to undo the curse that keeps him from Steingerðr.

hefir hon skorit tvǽr gǽs ok látit renna saman blóðit í bolla. ... Hafða ek nú ǽtlat at fyrirkoma þeim óskǫpum , er Þórveig hafði á lagt ykkr Steingerði. ... Kormákr mǽlir: Ekki trúi ek á slíkt.

She had cut the throats of two geese and let the blood run together into a bowl. ... "I meant to undo the evil fate that Þórveig laid on you and Steingerðr." ... Kormákr says, "I don't believe in such things."

His curiosity spoils the rite, and she had also taken a fee from his opponent. Earlier, told how to draw Skǫfnungr, he scoffs, "You have a lot of fuss, you charm-men!", and the sword howls and sticks in its scabbard (Kormáks saga 9). The plot punishes the sceptic, but the saga gives him the line.

Snorri's doubt is selective and learned. He mentions that on one journey "trolls and evil beings teased" Óláfr Tryggvason's men, then declines to tell it: "we would rather write about how King Óláfr christianised Norway" (ÓT 80). His larger move is to explain pagan religion as human skill and deception. In the expanded Prologue of the Codex Wormianus, Saturn is a man of Crete who "knew things before they happened" by the "pythonic art" and "found the red ore in the earth from which he smelted gold", so his people believed him a god (Edda, Prologue). The red ore, rauði, is the bog-iron Icelanders actually smelted. Gylfi's whole visit to the Æsir takes place inside an illusion, and within it the giant Útgarða-Loki explains to Þórr how he was fooled.

enn ſionhvꝼingar heꝼi ek goꝛt þer ok ꝼẏrſta ſinn a ſkoginum kom ek til ꝼunꝺar við ẏðꝛ ok þa er þu ſkẏllꝺer læẏſa neſtbaggan þa haꝼða ek bunꝺit hann með gꝛeſſ iarní

"But I worked illusions on you. The first time, in the forest, I came to meet you, and when you tried to untie the food-bag I had bound it with iron wire."

His hammer had struck a hidden mountain; the end of his drinking horn lay in the sea, and his draughts made the ebb tide. At the end the hall vanishes and Gylfi stands on an empty plain, while the Æsir take the names of the gods in their own stories "so that people should not doubt" they were the same (Gylf. 53). Christian translators used the same word. In the Icelandic Gregory, St Benedict tells monks who see their kitchen ablaze that it is a "sight-turning fire", dispelled by the cross (Greg. Dial. II.10).

Natural causes sit beside marvellous ones. The Norwegian King's Mirror explains Irish battle-madness as the terror of young men at the war-cry, then adds that the sufferers grow feathers and fly (Konungs skuggsjá 11). Gunnhildr disposes of her Sámi teachers by strewing what was taken for ash and by wholly practical means. Jealousy has kept them watching each other for three nights, and when they finally sleep she pulls sealskin bags over their heads and Eiríkr's men kill them (HHárf. 32). In the Swedish Flores och Blanzeflor (1312) the hero, accused of entering a guarded tower by sorcery, says he "cannot do truldom" (Flores och Blanzeflor); he had been carried in hidden in a basket of flowers.

Change over time

One medieval author names the change himself. Of the widow Gríma in early eleventh-century Iceland, Fóstbræðra saga says this.

Þat var mǽlt um Grímu, at hon kynni sér margt, ok þat tǫluðu menn, at hon vǽri fjǫlkunnig. Nú fyrir því at kristni var var ung ok vangǫr, þá sýndisk þat mǫrgum mǫnnum atgervi , at maðr vǽri fjǫlkunnigr.

It was said of Gríma that she knew a great deal, and people said she was skilled in magic. Now because Christianity was young and still unformed, many people thought it an accomplishment for someone to be skilled in magic.

The rough sequence runs from the verse of the 960s and Völuspá, through Sigvatr's magical reading of a real battle around 1040, to the twelfth-century laws and sermons. Between about 1200 and 1280 the sagas and Snorri projected magic back onto the Viking Age, mostly as the practice of outsiders and enemies of conversion, and explained it through euhemerism and ergi. By then seiðr needed glossing. Þiðreks saga, around 1250, explains that its sorceress queen "went to do seiðr, as was done in olden times, when sorcerous women, the ones we call völur, used to work seiðr" (Þiðreks saga, Ostacia).

In the Old Swedish texts of the fourteenth and fifteenth centuries the vocabulary and the conception change. Sorcery is truldom, the sorcerer a trolkarl or koklare (related to Old Norse kukl, "hocus-pocus"), and his art swartekonst, the "black art" of learned necromancy. A hired "devil's man" skilled in black art sends a devil to tempt a nun, and a picture of St Jerome keeps it off (Holm A 10, 87v). The converted magus of the Swedish Barlaam burns "all his books" of truldom and galdra (Fornsvenska Barlaam 25). Magic has become a matter of books and devils, on the continental model, and folk practitioners make things worse.

Sidha̅ komo qͥnno̅na frænder oc kalladho koklara at the ſkullo m; kokelſkap diæwlin wtkøra Ok tha koklarene willo diæwlin wtkøra tha pintis ho̅ ſo̅ ho̅ hafdhe thuſanda diæffla

Then the woman's kinsmen came and called in conjurors to drive the devil out with their conjuring. And when the conjurors tried to drive the devil out, she was tormented as if she had a thousand devils.

Only a bishop's prayer frees her. Around 1500 a prayer book from the Birgittine house at Vadstena promises that "no one is so great a sinner" that he will not receive grace from Mary at his end if he reads a certain prayer daily (Holm A 80). A fixed text on a fixed schedule guarantees an effect, and the Church called it devotion.

What is distinctive about the North

In the North shame attached to a single technique. Seiðr brought ergi on men, while galdr and runes were honourable for them. A neighbouring people, the Sámi, serve as the reservoir of expertise, consulted, traded with, feared and legislated against, down to coats bought on a trading trip. Runes make magic depend on literate competence, so that a badly carved love-charm sickens a girl. In níð, insult, lawsuit and curse are one act, addressed even to the spirits of the land. The single idea of the hamr explains both berserk fury and Óðinn's journeys in animal form. Communities kill sorcerers by stoning, drowning on tidal skerries and burning in halls, before the Church's laws as well as after them. And Snorri explained his ancestors' gods as men who won worship by skill, prophecy and illusion, and then preserved their stories with great care.

Nearly every Northern source on pagan magic was written by Christians two centuries after the conversion, fascinated by the old ways and committed to their end. What they show most reliably is how thirteenth-century Icelanders and Norwegians imagined their past, as a world of chanting women, shape-strong men, Sámi masters and land-spirits in the hills.

Chapter IX

Medieval Europe: blessings, fairies and frauds

Between the eleventh century and the fifteenth, German monks copied charms beside their herbals, French poets gave a fairy king a cup that judged sinners, and Dante put the sorcerers and alchemists of Italy in Hell among the frauds.

A wound stopped with a thumb

Somewhere in twelfth-century Franconia a scribe copied a few lines of German into a Latin medical book now in Bamberg. They tell a story nobody would find in the Gospels.

Krist unde. Judas spileten mit spieze. dô wart der heilige Krist wunt in sîne sîten. dô namer den dûmen. unde verdûhte sie vorne. sô verstant dû bluot. sôse Jordanis ahe verstuont.

Christ and Judas were playing at spears. Then holy Christ was wounded in his side. He took his thumb and pressed the wound shut at the front. So stand still, blood, as the river Jordan stood still.

Christ and his betrayer are playing with spears, and the Saviour gives himself first aid. Then the speaker turns from the story to the patient's blood and gives it an order. The charm goes on with a rhyme about Christ's wound that neither bled nor festered, three paternosters and an adjuration by the five holy wounds (Bamberger Blutsegen). Inside the sacred tale sits a real technique, pressure on the wound, but the charm never tells the healer to press. Chapter XII puts the charms' successes down to the spontaneous clotting of small wounds and nosebleeds, since a recitation of a minute or two is far shorter than the pressure a nosebleed needs.

This chapter covers the medieval material in two parts of the library. The first is the Reference Corpus of Middle High German (ReM), which holds about 400 German manuscripts written between about 1050 and 1350: charms, medical books, Bible poems, chronicles, sermons and courtly romances. The second is a mixed group of French, Italian and Latin texts: the Old French chansons de geste (heroic epics about Charlemagne's wars against the Saracens, as medieval Christians called Muslims), Dante's Commedia, and the line-by-line manuscript transcriptions of the CATMuS dataset, which reach from a Carolingian chronicle to a printed book of about 1500.

What medieval people called it

None of these medieval languages has a single word that matches the modern "magic". German speakers kept two families of words well apart. A segen was a blessing, the word for a priest's blessing; the modern titles of many charms (Blutsegen, Wurmsegen) borrow it. None of the German charms in the library calls itself zouber. A few call what they do a segen, others call themselves "these words", and many open with a Latin heading instead. Zouber was sorcery, the thing a confessor asked about and a saint denied doing. A third word, goukel, meant a conjuring trick or illusion. The working verb of the charms is büezen, to make good, the same verb that gives buoze, penance. A cured horse is "made good" of its foundering (Ad equum errehet).

The Old French epics use enchanter (in the Picard dialect encanter), a word that slides from casting a spell to simply beguiling someone. In Doon de la Roche a traitor "enchants" a court by flattery and slander, and no sorcery is involved (Doon de la Roche 115–116, 2956–2958). Faerie, from Latin fatum (fate), is both the fairy realm and its power. Nigromance, the art of the "black books", is Greek νεκρομαντεία (divination by the dead) reshaped by Latin niger, black. Dante's Italian adds malia (a spell), affatturare (to bewitch) and his own phrase magiche frode, "magical frauds".

Some absences matter as much as the words that occur. The German word hexe, "witch", never appears in the 1050–1350 corpus, and unholde usually means an ogre or a fiend. Old French sorcier and sorcerie do not occur in the library's epic editions at all.

Who practised magic

The German charms were written down by clerics. They survive in Latin codices from monasteries such as Prüll near Regensburg and in medical manuscripts at Bamberg, Innsbruck and Zürich. Their Latin rubrics leave a blank marked N. for the patient's name, and in a cattle charm for the colour of the beast (Contra vermes pecus edentes). One wound charm implies a lay healer. In it Christ makes three brothers swear that they will neither keep his remedy secret nor take money for it (Münchener Wundsegen). A fever charm gives its patient the ordinary names Heinrich or Hildegunde (Gothaer Fiebersegen). Travellers spoke their own protective charms (Münchener Ausfahrtsegen). Nobody in the corpus calls any of these people a zouberære.

The sorcerers of German narrative are a different cast. Most are male villains taken from the Bible and from saints' legends: Simon Magus, Pharaoh's magicians, the Ethiopian sorcerers Zaroes and Arphaxat whom St Matthew defeats (Sermon on St Matthew), and the hired seer Balaam (Vorauer Balaam). Beside them stand ambiguous masters of learning: Solomon, Klingsor, and a priest who had read the black books in his youth. The female sorcerer belongs almost entirely to courtly romances adapted from French and Latin, where Medea, Circe and Queen Isolde are learned women skilled in drugs.

The French epics give magic to fairies and their households, to Saracens and their servants, and to learned clerks. Auberon, the dwarf king of Faerie in Huon de Bordeaux (c. 1260–1268), says he is the son of Julius Caesar and Morgue la fée, and that fairies gave him gifts at his birth (Huon de Bordeaux 3494–3517). Healers include physicians, one a former pagan from overseas (Aymeri de Narbonne 4411–4418), and the women of Femenie, the poets' land of Amazons (Mort Aymeri 1984–2006). Dante's sorcerers are partly classical seers and partly astrologers, prophets and alchemists of his own century. The CATMuS manuscripts add clerics with lot-books, wives in the penitentials, possessed women at a saint's tomb and a royal astrologer defending his trade.

What they did

Charms for bodies, beasts and journeys

The largest single body of medieval European magic in the library is the German charm. Most follow a two-part pattern that scholars call the historiola, "little story". First comes a short narrative of a sacred precedent. Then the speaker applies it to the case in hand with "as... so". The Paris Jordan charm against nosebleed tells how Christ and John came to the Jordan and Christ told the river to stand still while they crossed. "As the Jordan stood then, so stand you, blood of N." The Latin rubric says to say it three times and tie a knot in the patient's hair each time (Pariser Jordansegen). The story changes with the illness. Job on his dunghill, eaten by worms, serves against worms, and the charm ends "the worm is dead, dead is the worm" (Prüler Wurmsegen). A chain of holy births asks a fever to leave the body as gently as Mary bore Christ (Gothaer Fiebersegen).

Animals were patients too. A horse charm has Christ meet a man leading a foundered horse and tell him to whisper in its ear and tread on its right foot (Ad equum errehet). A cattle charm addresses the most powerful thing in the sky.

contra vermes pecus edentes. ich beswere dich sunne. bîsancto Germano. daz dû hiute ne schîn. êr deme dic colore vihe die würme ûzsîn

Against worms that eat cattle. I adjure you, sun, by Saint Germanus, not to shine today until the worms are out of the [name the colour] beast.

Charms also protected. A morning charm for a man setting out asks that his enemies' weapons "lie and sleep today" and that his head be of steel, while his own sword alone cuts. It invokes Christ "by the wonderful name tetragrammaton", the four-letter Hebrew name of God (Münchener Ausfahrtsegen). A Cologne morning blessing binds every enemy, "visible or invisible", limb by limb (Kölner Morgensegen). A very old house blessing taunts a demon it calls wiht ("wight") with a word, chnospinci, that the creature supposedly cannot say. Nobody knows what the word means (Zürcher Haussegen).

The charm tradition was not confined to German. A twelfth-century Latin manuscript, BL Egerton 821, which also holds the lot-book discussed below, has a charm that names the fevers as seven sisters (Lilia, Reptilia, Fugalia, Africa, Vilia, Sufocalia, Ignea) and adjures them by the Trinity, the Virgin and the archangels (Egerton 821, fever charm).

The Arzneibücher: pharmacy, amulets and rites

The German Arzneibücher ("medicine books") translate Latin pharmacy, much of it from the medical school of Salerno, and credit it to Hippocrates and Galen. Most of the Arzenîbuoch Ipocratis is plain pharmacy, such as willow bark boiled in old wine for the spleen (Arzenîbuoch Ipocratis 2v). Yet the same books also give a dead horse's bone to hang on a live horse's neck (Innsbrucker Arzneibuch 77v–78r), a live lizard bound on the navel to draw out illness (Innsbrucker Arzneibuch 78r), and an egg laid on a Thursday against the stone (Zürcher Arzneibuch 45v).

The fullest mixture is the book of "Meister Bartholomäus" (c. 1200). It gives a test for a wounded man that a modern surgeon would recognise. If he digests a drink of burnet he will recover, and if it is found in the wound he will die. The drink can reappear only where the wound reaches the gut, and before modern surgery a pierced gut was usually fatal (Bartholomäus 9v). It fumigates "tooth-worms" with a candle of henbane oil and wax and promises that the worms will fall into a bowl of water, though tooth-worms do not exist and the recipe claims no relief of pain (Bartholomäus 7v). It staunches bleeding at a distance with water drunk by the messenger (Bartholomäus 8v). Verbena must be circled with gold and silver, adjured "by the seventy-two names of almighty God", dug before sunrise without iron and blessed on a feast of the Virgin. It then protects against "all zouber" and the nightmare-demon, the alp (Bartholomäus, verbena). For epilepsy, the book binds the illness into a knotted strap and sends it to the grave with a stranger.

In nomine patris et filii et spiritus sancti begrabe ich mit diseme riemen den siechtuom ditze mennischen. mit dem gedinge. daz disem mennischen dirre siechtuom niemer mêre gewerre. unz dirre lîchname andem jungestem tage erstê.

In the name of the Father and of the Son and of the Holy Spirit I bury this person's sickness with this strap, on condition that this sickness never trouble this person again until this corpse rises on the Last Day.

The Resurrection works here as a legal deadline. The compiler saw no contradiction between this rite and the burnet test a few leaves earlier.

Divination: lots, moons, ordeals and stars

Medieval divination in the library is often Church business. A twelfth-century manuscript copied in a religious house contains a dice oracle that calls itself the "lots of the saints".

Hae sũt sortes scõꝵ q̃ nũꝙͣ fallũt᷑ id̾oq dm̃ ... IIII IIII III Recede sortes ñ dãt tͥ respõsũ alt̾a die ueni.

These are the lots of the saints, which never fail; therefore [pray to] the Lord... Four, four, three: Go away. The lots give you no answer. Come another day.

Church councils had condemned the sortes sanctorum since late antiquity, yet monks still copied them. In German there is a lunar calendar that tells, day by day of the moon's month, whether a sick man will live and when dreams come true (Tegernseer Lunar). The judicial ordeal was divination run by priests. A Trier liturgy for a cold-water ordeal over a stolen horse has mass said, the water blessed, and a boy sent into it as proxy for the accused (Trierer Gottesurteil).

Astrology sat on the border. A life of St Elisabeth (c. 1300) has the Hungarian master Klingsor, "who knew all things: necromancy, and astronomy", read from the stars the birth of the future saint, and explains that God "sent his spirit" to him as he once did to Balaam (Leben der hl. Elisabeth 201ff.). In Anseïs de Metz astrology is a liberal art with biblical credentials. Joseph rose under Pharaoh by it and the Magi found Christ by their star (Anseïs de Metz, U 33r.301–324). In the epics a learned priest may foretell a hero's death without blame (Mort Aymeri 490–491).

Necromancy and the black books

From the twelfth century on, both literatures know a learned art of conjuring spirits from books. The Kaiserchronik tells how young Astrolabius put his ring on the finger of a Venus statue and was bound to the devil inside it. The court chaplain Eusebius, who "while he was a young man, read in the black books", summons the devil by reading from a book and forces him to give the ring back. The narrator approves (Kaiserchronik 13216–13250). Herbort von Fritzlar's Trojan poem (c. 1190–1210) gives the art a home address.

sie kunde arzedîen unt von nigromanzîen daz man heizet swarze buoch ... noch sô lernet man die list in einer stat zuo Tolêt ... des was niht man hette es wân daz iz alsô wære

She knew medicine and necromancy, which is called the black book... that art is still learned in a city, at Toledo... It was not so. People only imagined that it was so.

The last sentence refers to Medea's most famous feats. She made rivers run back to their source and the moon move backwards, and Herbort tells his audience that these things existed only in people's belief. Elsewhere he says that this "zouberlist called nigromantia" came down to us from the devil, who spoke out of pagan idols and foretold the future (Herbort 9365–9373). The mid-thirteenth-century Vorauer Novelle takes two young clerics to a city where "the books of necromancy" are read openly in a school of magic. The master warns that whoever learns the magic books "must take God's swift curse upon his soul". They learn anyway, use the art to seduce, and one dies "on the magic book" (Vorauer Novelle 128–165).

The French Anseïs de Metz (late thirteenth century) puts nigromance at the end of a clerk's studies. It is "the most hated of the arts, for the devil holds full power over it", and yet it lets the clerk name thieves, understand the speech of dogs and birds, turn himself into a partridge and take a castle without bloodshed. The barons at once wish he were on their side (Anseïs de Metz, U 33r.325–347).

Objects of virtue: fairy and Saracen magic in the chansons de geste

In the French epics magic lives mostly in things. Auberon owns an ivory horn made by four fairies that heals whoever hears it and banishes hunger (Huon 3231–3255); a seat that reveals poison (Huon 3609–3621); a hauberk whose wearer cannot drown or burn (Huon 5055–5073); and a golden cup.

De se main destre l’a trois fois enviré , / Puis fist crois sus de Dieu de maïsté ; / Lors devint plains de vin et de claré ... Nus n’i puet boire s’il n’est preudom par Dé , / Et nes et purs et sans pecié mortel .

With his right hand he turned it round three times, then made over it the cross of God in majesty, and it filled with wine and spiced wine... No one can drink from it unless he is a good man before God, clean and pure and free of mortal sin.

The cup is a fairy ordeal. Its powers, like the hauberk's, depend on the moral state of whoever uses it. The hauberk's inscribed collar even requires that the wearer's mother never desired any man but her husband (Huon 5055–5073). The horn comes with bans on needless blowing and on lying (Huon 3706–3729), and Huon's breaches drive the plot. Auberon himself works by wishing. He wishes two traitors and a lying abbot and monk onto the gallows, and there they hang (Huon 10355–10368).

Saracen magic in the chansons is exotic, technical and often criminal. In the emir's palace at Babylon in Aymeri de Narbonne (c. 1210–1220) stands a mechanical tree.

Fet fu de coivre si l ot en fet dorer ... Li enchanteres fist forment a loer ... Par nigromance i fait le vent entrer / Encontremont par le tuel monter / Qant li vanz sofle les oisiax fet chanter

It was made of copper, and he had it gilded... The enchanter deserved great praise... By necromancy he makes the wind go in and climb up through the pipe, and when the wind blows it makes the birds sing.

The poet calls it necromancy and describes plumbing. The tree closely resembles the air-driven singing-bird trees of the palaces at Constantinople and Baghdad. The same poem has Saracens call Charlemagne's overnight siege tower "a great enchantment" (Aymeri 1043–1048), and in Huon two copper men with flails guard a castle until a girl "throws down the engien", the machine (Huon 4563–4571, 5236–5239). The library's fragment of Fierabras has the one real burglar's spell in these poems.

une chandele a prise au feu l a alumee / vínt a l us de la chambre si la trova fermee / et il a dit son charme tantost est deffermee ... La chandele au larron est adonc anublee / Onc plus l enchantement n ot la dedenz duree

He took a candle and lit it at the fire, came to the chamber door and found it locked. He said his charm, and at once it stood unlocked... Then the thief's candle went dark, and the enchantment lasted no longer in that room.

The thief Malpin, hired by the emir, uses the charm and a sleep-candle to steal Floripas's girdle, which keeps the besieged French from starving (Fierabras V 71–81). When Gui de Bourgogne cuts him in two the candle goes out and the sleepers wake. Other marvels are plain dyes, as when a knight blackens his skin with a herb to pass as a Saracen (Floovant 1216–1223).

How it was supposed to work

The texts offer several explanations, sometimes in the same manuscript. The historiola works by precedent. A second model treats the illness as a person who can be addressed and commanded. "I command you, worm, who lie in the flesh, whether you are one or two or however many you are" (Contra vermem edentem). Its Latin note warns that if it is ever used on a beast of burden it will afterwards be useless for humans, as if a charm's power could be spoiled.

A third model is binding and transfer. Illness is tied into a knot in the hair or a strap, or drawn into a lizard, or taken by the white caladrius bird of the Physiologus, which looks a sick man in the face and draws his sickness into itself (Physiologus, Caladrius). A fourth is the power of names. Prayer books contain words nobody can translate, zanuel, paruc, domello, astac, efranis (Muri prayer book), and lists of the hidden names of God such as "On, Ione, Ot, Zel, Melchion, Adonay" (Vatikanische Gebete). A Benediktbeuern guide tells the reader to keep Gabriel in mind during thunder and Uriel against enemies, though Uriel was among the angel names a Roman synod had condemned in 745 (Benediktbeurer Ratschläge).

A fifth model gives stones, herbs and animal parts hidden natural powers, which ritual gathering could strengthen. Jasper staunches blood and helps the eyes if one blesses oneself with it on three mornings (Prüler Steinbuch, jasper). Every limb of a vulture killed without iron is a medicine, by authority of "Jerome's Chaldean books" (Bartholomäus, vulture letter). The Wiener Genesis (c. 1060–1080) explains the monstrous races in a similar way. Adam forbade the women certain herbs that harm the unborn, and they disobeyed (Wiener Genesis 1282–1297). Wolfram tells the same story in Parzival (Parzival 518.1–20).

Behind the hostile cases stands a theory of demons. The German Lucidarius, a vernacular encyclopedia, places evil spirits in the air below the moon until Judgement Day and says they take bodies of air when they appear to people (Lucidarius I 1v). Simon Magus in the Kaiserchronik boasts that he can make lead into gold and escape any captor (Kaiserchronik 2082–2094). St Peter reveals the source and tells him to bring out "your dead... an unblemished child, with which you have worked your sorcery" (Kaiserchronik 2565–2572).

The fairy world of Huon follows its own logic. Power was given at birth by fairies, sits in objects as their vertu or disnité (inherent power or worth), and is exercised by wishing. It can also produce mere appearances. Auberon raises storms and makes a river that a fleet could sail on, but "all you see there will be phantoms" (Huon 3161–3190). Dante adds a psychological model of his own. In a dream an ugly, stammering woman becomes beautiful under his gaze and sings that she is the siren who bewitches sailors. The beholder's desire makes the enchantment (Purg. 19.1–33).

Where the line was drawn

The boundary between licit and illicit did not follow technique. Lawful baptism used the same verbs as the charms. The priest "blows the devil away and adjures him in the name of the Father, Son and Holy Spirit" (Hoffmannsche Predigtsammlung 27r). A Latin collect to St Blaise becomes a throat charm once a German rubric adds "bless" and three paternosters (Münchner Halssegen). The texts drew the line by other tests. The first was the source of power. In the Kaiserchronik's disputation scene, the Jewish master Zambri kills a wild bull by whispering a secret word in its ear (Kaiserchronik 10043–10051). Pope Silvester answers that killing is the devil's work, like manslaughter, and sets a test.

habeez Zambrî mit gote getân. sô heize in ouch mit gote wider ûf stân.

If Zambri did it with God, let him with God command it to stand up again.

God gives life and the devil can only destroy. A thirteenth-century sermon applies the same test to pagan healing. The devil in the idol Astaroth made people deaf, blind and lame by zouber, then "took the magic off them" when they sacrificed, so that they took him for a god (Sermon on St Bartholomew).

The second test was who acted and why. The priest Eusebius may conjure the devil to free a bewitched young man. The student clerics of the Vorauer Novelle learn the same art for seduction and are damned. The third was reality against appearance. The Altdeutsche Exodus (c. 1120–1130) calls Pharaoh's magicians goukelære who make their serpents "by the art of illusion" (Altdeutsche Exodus 1242–1250). A sermon says Simon's carved heads only "seemed" to laugh (Sermon on Peter and Paul). The French epics draw the same line for fairy power with a recurring formula. Auberon acts "par faerie, par le Diu volonté", by fairy power and by the will of God (Huon 8996–9001). At the end of the poem he leaves Faerie for a seat at God's right hand in Paradise (Huon 10452–10458).

The penalties were mostly spiritual. A twelfth-century model confession from Bamberg lists the sins a lay person was to confess.

in zoubere. in goukele. in heilsôde. in getrügede. an deme verwâzenen merrsale. des gotes rehtes. in alleme tiuvelheite.

...in sorcery, in conjuring tricks, in omen-taking, in deceit, in the accursed obstruction of God's judgement, in all devilry.

"Obstruction of God's judgement" most likely means tampering with the ordeal, though that reading is an inference. Latin penitentials copied in a twelfth-century French manuscript set fifteen years of penance for "magicians and soothsayers" who practise "the magic art" (BnF lat. 12270, penitential canons). Other canons in the same book price the domestic magic of wives, such as mixing a husband's semen into his food to keep his love or burning a man's skull to give as medicine (BnF lat. 12270, canons on women). None of the German law-book texts in ReM, the Schwabenspiegel among them, has an article on sorcery in the parts ReM holds, though the full law books, like the Sachsenspiegel (Landrecht II 13), prescribe burning for it.

Secular punishment appears in a few places, and usually as politics. The oldest manuscript of Nithard's history records that in 834 Lothar had Gerberga, sister of his enemy Bernard of Septimania, "drowned in the Saône in the manner of sorcerers", more maleficorum (Nithard, Hist. 1.5). In the Kaiserchronik a rejected courtier calls the empress Crescentia's old attendant an unholde, and later Crescentia herself is accused of bewitching her husband. The narrator presents both charges as slander (Kaiserchronik 12179–12195).

Dante gave sorcery a fixed place in his moral map of Hell. In the eighth circle, the first of the two given to fraud, "nest hypocrisy, flattery and whoever casts spells", beside falsifiers, thieves and simonists (Inf. 11.52–60). The diviners walk with their heads twisted backwards, and among them stand the professionals of his own age.

Vedi Guido Bonatti; vedi Asdente, / ch'avere inteso al cuoio e a lo spago / ora vorrebbe, ma tardi si pente. / Vedi le triste che lasciaron l'ago, / la spuola e 'l fuso, e fecersi 'ndivine; / fecer malie con erbe e con imago.

See Guido Bonatti. See Asdente, who now wishes he had kept to his leather and his thread, but repents too late. See the wretched women who left the needle, the shuttle and the spindle to make themselves fortune-tellers. They worked spells with herbs and with images.

Michael Scot, astrologer at the court of Frederick II, "truly knew the game of magical frauds" (Inf. 20.106–117). Whether that means he mastered the game or saw through it is disputed. The alchemists lie lower still, among the falsifiers. Griffolino of Arezzo was burned because he joked that he could fly and could not teach a fool to do it, but Minos damned him "for the alchemy I practised in the world" (Inf. 29.109–120). His companion Capocchio, also burned according to Dante's early commentators, "falsified metals with alchemy" and was "a good ape of nature", an imitator who could not truly transform anything. Next to them sits Master Adam, burned for counterfeiting the florin (Inf. 29.136–139; 30.73–75). Dante puts alchemy in the same class of fraud as coining false money.

Astrology was allowed within limits. Dante has Marco Lombardo explain that the heavens start our movements, "I do not say all of them", and that free will remains (Purg. 16.67–81). In the heaven of the fixed stars Dante thanks Gemini, his birth sign, for "all my genius, whatever it may be" (Par. 22.112–120). Nearly two centuries later the royal astrologer Simon de Phares, condemned by the Paris theologians in 1494, drew the same line from the practitioner's side. In what seems to be his working copy, astrology is "the principal and dean of the seven liberal arts", set apart from "invokers, necromancers, deceivers or diviners" (BnF fr. 1357, Simon de Phares).

Doubt, irony and mockery

The epics reverse the usual direction of the charge. In older Christian writing Christians accuse pagans of magic. In the chansons the guilty and the pagans accuse Christians of it. When Auberon's cup empties in the hand of the emir Gaudise, and at the end in Charlemagne's own, both cry enchantment.

Li rois le prent , k’il ne l’ose véer ; / Lues qu’il le tint , li vins en est alés . / Vasal , dist Karles , vous m’avés encanté » / Dist Auberons : « Ains fait vo mauvaisté

The king takes it, for he dares not refuse. As soon as he held it, the wine was gone. "Vassal," said Charles, "you have bewitched me." Auberon said, "Your own wickedness has done this."

Huon himself, caught in a lie, blames it on the devil who "enchanted" him (Huon 5531–5534). In Otinel the Holy Spirit descends as a white dove and converts the Saracen champion (Otinel 615–625). The Saracen Clarel has another explanation: "You are enchanted. You have drunk a draught of forgetting, by which these physicians cause madness" (Otinel 1319–1324). French saints' lives repeat the move, with sceptics dismissing miracles as enchanterie (BnF fr. 412).

Courtly German poets were openly sceptical. Herbort's Medea, when she falls in love, at first suspects that "the elves are deceiving me" (Herbort 756–760, 845–856). His Ulysses says of Circe that "magic art is but wind against a man's heart", and that he escaped her drinks by resolve (Herbort 17631–17682). The German Alexander poets reject as a lie the story that Alexander was "a juggler's son", meaning the Egyptian magician Nectanabus of the Greek romance, whom they never name (Straßburger Alexander 83–88, 262–269). Gottfried von Straßburg (c. 1210) turned his irony on the ordeal. Isolde carries the hot iron unhurt because she has framed her oath so that it is technically true.

daz der vile tugenthafte Krist wintschaffen als ein ermel ist. ... erist allen herzen bereit. ze durhnehte unde ze trügeheit. ist ez ernest ist ez spil. er ist ie swie sô man wile.

...that Christ, for all his virtue, flutters in the wind like a sleeve... He is at the service of every heart, for honesty and for deceit. Be it earnest or be it play, he is always just as one wants him.

Gottfried calls the oath gelüppet, a word otherwise used for poison and magic drugs. Watching the diviners' twisted necks, Dante wonders whether palsy could ever turn a head so far round, "but I never saw it, nor do I believe it is so" (Inf. 20.10–18). He also has Virgil insist that Mantua was named "without any other casting of lots", taking augury out of the Aeneid's account (Inf. 20.52–99). At the end of the period the Évangiles des quenouilles (c. 1470–1480), a comic book of old women's "gospels" told at their spinning evenings, makes household magic a joke. A woman who wants her lover's devotion should feed him catnip, and two of the gossips are called "la fée" (Évangiles des quenouilles).

Change over time

The earliest German layer, roughly 1050–1150, includes charms whose figures are not obviously Christian. In the Strasbourg blood charm "Genzan and Jordan went shooting together" and Genzan shot Jordan in the side, and a "holy Tumbe" is asked to charm the wound away. The manuscript burned in 1870, and the names are disputed (Straßburger Blutsegen). An epilepsy charm opens with the word "Donerdutigo" and tells of the devil's son on Adam's bridge (Donerdutigo charm). The early Bible poems keep native words. The magicians of Egypt copy the plague of blood mit galsters liste, by the art of incantation, a cognate of Old Norse galdr (Altdeutsche Exodus 1288–1291), and Joseph's divining cup is rendered with the old verb liezen, to cast lots (Wiener Genesis 4669–4671). The twelfth-century monastic collections are thoroughly Christian.

Around 1150 the Kaiserchronik brought in full sorcerer narratives. Around 1200 courtly romance imported learned necromancy, Toledo, female enchantresses and a new scepticism, while Bartholomäus gathered herbal rites into a medical handbook. In the thirteenth century sermons carried the apostles' contests with magicians to lay audiences. The French chansons moved from exotic Saracen machines and poisons around 1200 to the fully Christian fairy kingdom of Huon in the 1260s, and by the late thirteenth century nigromance had become an academic subject in Anseïs. Dante (c. 1306–1321) combined Aristotelian astrology with a classification of magic as fraud. By the fifteenth century the CATMuS manuscripts show courtly astrology on the defensive, Merlin's prophecies read as political texts in the Hundred Years War (BnF fr. 1984), and women's folk magic turned into comic print.

What is distinctive here

Medieval Europe in this library differs from the ancient cultures earlier in the report in three ways, though the spell-books of Christian Egypt (chapter VII) share much of the first. First, almost all its healing and protective magic was written by people who considered themselves orthodox Christians and called their work blessing or medicine. Charm, prayer, herbal and lapidary sit in the same codices, and the historiola puts Christ, Mary, Job and the Jordan to work on nosebleeds, worms and foundered horses.

Second, the charge of magic had become an argument about the source of power, and it was often turned round. The pagan, the persecutor and the guilty king call God's work enchantment, and the narrator corrects them. Third, poets of around 1200 to 1320, Herbort, Gottfried and above all Dante, brought a scepticism closer to Lucian than to the charm books. They called Medea's reversed moon a belief, the ordeal a game and the diviner's craft a fraud, and yet Dante thanked his stars for his genius. The fairy magic of Huon has no real parallel elsewhere in the library. Its powers are held under God and test the sins of those who touch them.

Chapter X

Renaissance England: witches, conjurers and the art laid down

In the chronicles, treatises, chapbooks and plays of Elizabethan and Jacobean England, magic is a crime a king must never pardon, a weapon in court faction, a fairground cheat and, on the stage, a learned art partly built out of an ancient witch's prayer.

A wax king at Forres

Raphael Holinshed's History of Scotland, in its 1587 edition, tells of King Duff, a tenth-century ruler wasting from a sickness his physicians could not name. He sweated at night and could not sleep. Rumour blamed witches at Forres, hired by rebel nobles of Moray, and soldiers broke into their house at midnight.

one of the witches rosting vpon a woodden broch an image of wax at the fier … as the image did waste afore the fire, so did the bodie of the king breake foorth in sweat.

One of the witches was roasting a wax image on a wooden spit at the fire … and as the image wasted before the fire, so the king's body broke out in sweat.

A second witch recited words over the image and basted it. The image was broken, the witches burned, and the king recovered at that very hour. Holinshed hedges: the image was made "as is to be thought" by the devil's craft Holinshed, Duff 7–9. The story holds much of this chapter in small: a sympathetic image, a plot against a king, women hired by noblemen, a fire, and a chronicler who will not vouch for the wonder. Twenty years later Shakespeare put his witches on the road to Forres.

The library's Renaissance English collection supplies seventy surveyed passages. Most are from Shakespeare (c. 1590–1611). The rest come from Holinshed's Chronicles, the English Faust Book (c. 1592), the works of James I, Sidney's Defence of Poesie, a 1910 study that quotes North's 1579 Plutarch, and Arthur Golding's 1567 English Ovid. For several plays the library holds both the chronicle source and the play, so the reader can watch the material change hands. Beside them the chapter reads the treatises in which an English gentleman, two preachers and a Scottish king argued about witches and conjurers, the conjurers' own books, and both printed texts of Marlowe's Doctor Faustus.

The words for it

Elizabethan English had no neutral word for what this report calls magic. It had a cluster of terms, most of them accusations, each naming a different kind of person or act.

Several of these words, juggler and cozenage among them, cover magic and fraud at once. The word "spirits" was also double. It could mean demons, or the fine vapours of Galenic medicine that carried life through the body, so when the doctor in Cymbeline says his drug only locks up "the spirits a time", he is talking physiology Cym. 1.5.33–44. The period's own categories turned less on technique than on where a power came from: God, nature or the devil.

Who practised it

The texts sort practitioners by sex, class and learning. Village witches are old, poor, female and quick to take offence. Macbeth's First Witch avenges a refused gift of chestnuts by tormenting a sailor's ship and his sleep Mac. 1.3.8–25. Learned magi are men with books, staff and circle: Prospero, Glendower, the conjurer Bolingbroke, Faustus. Faustus called himself an astrologer and, "for a shadow", a physician Faust Book ch. 1. Mixed teams link court and street. In 1441 Eleanor Cobham, Duchess of Gloucester, was accused together with a necromancer, a priest and "Margerie Iordeine, surnamed the witch of Eie" Holinshed, Henry VI 252–253. Foreigners carry occult power: Joan of Arc, the "Egyptian" charmer of Othello's handkerchief, Sycorax of Algiers, the Welsh Glendower, and an Ireland "much giuen to such dailie practises" Hooker, Irish Chronicle, Elizabeth 115. Then come frauds such as Pinch, and licit healers such as Friar Laurence and the king who touches for "the evil".

One real Elizabethan practitioner appears in the chronicle. Simon Penbrooke of Southwark was "a figureflinger, and vehementlie suspected to be a coniurer". Summoned before the church court at St Saviour's in January 1578, he collapsed and died with money in his hand. In his clothes were five books of conjuration and a tin man holding three dice, engraved "Chance dice fortunatlie" Holinshed, Elizabeth 2.6. Across the corpus, harmful village magic is women's work and learned ritual magic is men's, and the plays like to set them side by side, Prospero against Sycorax and Bolingbroke beside Jourdain.

What they did

Foretelling

Divination is the commonest practice in the collection. Its most famous case began in Holinshed as something other than witchcraft. Three women in "strange and wild apparell" hail Macbeth and Banquo and vanish. At first this was taken for illusion. When everything they said came true, opinion changed.

the common opinion was, that these women were either the weird sisters, that is (as ye would say) the goddesses of destinie, or else some nymphs or feiries, indued with knowledge of prophesie by their necromanticall science

The general opinion came to be that these women were either the weird sisters, which is to say the goddesses of destiny, or else nymphs or fairies endowed with knowledge of prophecy by their necromantic science.

Holinshed never calls them witches. The later prophecies about a man "borne of anie woman" and Birnam Wood come in his account from "a certeine witch, whome hee had in great trust", and their "vaine hope" drives Macbeth to tyranny Holinshed, Macbeth (Duncane 53). Shakespeare merged Fates, fairies and trusted witch into one set of bearded crones, and Banquo casts them as seers: "If you can look into the seeds of time, / And say which grain will grow" Mac. 1.3.48–69.

The chronicles return often to prophecy that fulfils itself. A seeress on Iona tells a royal messenger the king will die by a friend's hand, "Euen by thine saith she", and the frightened messenger does the deed Holinshed, Natholocus 2–3. Peter of Wakefield, a hermit "either inspired with some spirit of prophesie as the people beléeued, or else hauing some notable skill in art magike", told King John he would lose his crown by Ascension Day, and was hanged when the day passed Holinshed, John 168. Lower down, divination was a trade; Ford says the old woman of Brentford "works by charms, by spells, by the figure, and such daubery as this is" Wiv. 4.2.159–165, the figure being a chart cast to answer a client's question.

Harm at a distance

The wax image is the standard instrument of harmful magic in the chronicles, and it is political. Eleanor Cobham's associates "deuised an image of wax representing the king, which by their sorcerie by little and little consumed" Holinshed, Henry VI 252–253, so that her husband would inherit. On the stage the image gives way to the cauldron.

Round about the cauldron go; / In the poison'd entrails throw. … Eye of newt and toe of frog, / Wool of bat and tongue of dog … Root of hemlock digg'd i' the dark … Then the charm is firm and good.

Circle the cauldron and throw in the poisoned guts … newt's eye and frog's toe, bat's fur and dog's tongue … hemlock root dug up in darkness … then the charm is strong and sound.

The recipe mixes real poisons (hemlock, yew, toad) with matter whose force is sacrilege and horror, such as "witches' mummy" and the finger of a "birth-strangled babe", gathered at charged times. Macbeth asks what they are doing, and they answer "A deed without a name" Mac. 4.1. English trial records seldom mention cauldrons. The English cases that Gifford and Glanvill print have none: their witches work with familiars, curses and images of wax Gifford, Dialogue 50 Saducismus Triumphatus, Relation III. The one cauldron in Scot's Discoverie belongs to the continental demonologists he summarises, whose witches steal unbaptised children from their graves and "seeth them in a chaldron, until their flesh be made potable" to make a flying ointment Scot, Discoverie III.1. The brew owes more to the classical witches of Ovid and Seneca.

Raising spirits

The fullest staged conjuration is in 2 Henry VI. The hour is chosen ("Deep night, dark night, the silent of the night"), a "hallow'd verge" is drawn, a priest reads "Conjuro te", and the spirit is commanded "By the eternal God, whose name and power / Thou tremblest at". Its answer can be read two ways: "The duke yet lives that Henry shall depose" 2H6 1.4.15–31. Ritual magic used God's name to force a demon. The English Faust Book drops that claim. Faustus draws "with a wand a Circle in the dust, and within that many more Circles and Characters" and charges the spirit "in the name of Beelzebub to appeare" Faust Book ch. 2. A few chapters later he puts the bargain in writing.

giuen both bodie & soule, vpon such condition, that they shall learne me, and fulfill my desire in all things … at the end of 24. yeares … I haue written it with mine owne hand and blood

… have given both body and soul, on condition that they teach me and fulfil my desire in everything … at the end of twenty-four years … I have written this with my own hand and blood.

The pact is drafted like a legal deed, "by these presents" Faust Book ch. 6, with a term of years. Its English stage counterpart is bodily. Joan of Arc, as the French lose, calls on her familiars, "Where I was wont to feed you with my blood, I'll lop a member off and give it you" 1H6 5.3.1–29, and the fiends hang their heads and leave. The Faust Book is careful about what a conjurer can raise. When Faustus shows students Helen of Troy, they stay calm only because "they perswaded themselues she was a Spirit" Faust Book ch. 45. In the Protestant teaching of Marlowe's day the dead did not return. The second part of Lavater's Of Ghostes and Spirites (English 1572) sets out to show that apparitions "be not the Soules of Men, but be either good or euill Angels, or else some secret and hyd operations" Lavater, Of Ghostes and Spirites, Part II, and Perkins writes that "Dead men doe neither walke, nor appeare in bodie or soule after death" Perkins, Damned Art 3.3.12. So the Helen who appears is a demon in her shape.

Love, herbs and drugs

Oberon's love-in-idleness, the wild pansy, works when its juice is "on sleeping eye-lids laid" MND 2.1.165–172, on the eye because love was thought to enter there. Othello's handkerchief is darker. An Egyptian charmer gave it to his mother to keep her husband faithful, and he warns that "there's magic in the web of it … And it was dyed in mummy" Oth. 3.4.55–75, the embalmed-flesh powder that apothecaries really sold. Critics still argue over whether the play believes this history.

Much of the collection's "magic" is pharmacy. Friar Laurence states the theory, "mickle is the powerful grace that lies / In herbs, plants, stones, and their true qualities" Rom. 2.3.15–26, and his sleeping draught is a "distilled liquor" with no spell attached Rom. 4.1.93–106. The line between these and the witch's brew is thin. The poisoner in Hamlet's play-within-the-play addresses his "mixture rank, of midnight weeds collected, / With Hecate's ban thrice blasted" and invokes its "natural magic and dire property" Ham. 3.2.255–260. "Natural magic" was a technical term for the hidden physical powers of things, and here it shares a sentence with a goddess's curse.

Healing by holiness

The same means, an amulet and spoken words, sit on the right side of the line when the healer is sacred.

'Tis call'd the evil: / A most miraculous work in this good king … The mere despair of surgery, he cures, / Hanging a golden stamp about their necks, / Put on with holy prayers

It is called "the evil": a truly miraculous work in this good king … people whom surgery has given up he cures, hanging a gold coin round their necks, put on with holy prayers.

Edward the Confessor cures scrofula with a touch-piece and prayer. English kings still performed the ceremony, and the play compliments James I's line as its heirs. Fighters in a judicial combat under Richard II swore they carried no "herbe or stone" of the kind magicians used Holinshed, Richard II 64.

Illusion and spectacle

The Faust Book's third part is a jest-book. Faustus sells a horse that must never be ridden into water; the buyer does so, "and he sate on a bundell of strawe" Faust Book ch. 34. At Frankfurt fair "foure Iuglers" cut off one another's heads and send them to the barber Faust Book ch. 47. In The Tempest magic is described as theatre. Prospero praises Ariel's harpy as a part well "Perform'd" Tmp. 3.3.83–91, and after the masque he says "These our actors, / As I foretold you, were all spirits" Tmp. 4.1.148–149.

How it was supposed to work

The texts carry several theories at once. In the Forres witches' own account Holinshed, Duff 8, sympathy makes the wax king sweat; James VI explained the same effect otherwise (see 'The king's theory' below). Holy names compel spirits, and the circle protects the conjurer while they rage outside it. Blood, in a pact or fed to a familiar, binds a spirit to serve. Books store the power, as Caliban knows: "Remember / First to possess his books; for without them / He's but a sot, as I am" Tmp. 3.2.86–93.

Over all of these lies the Protestant conviction that the devil works by deceit. Macbeth's Hecate, probably added later by Thomas Middleton (the authorship is disputed), promises "artificial sprites / As by the strength of their illusion / Shall draw him on to his confusion" Mac. 3.5.6–29. Her apparitions tell the truth in a way built to mislead. When Macduff turns out to have been "Untimely ripp'd" from the womb Mac. 5.8.16, Macbeth calls them "juggling fiends … / That palter with us in a double sense" Mac. 5.8.8–22, giving demons the name of fairground cheats.

Where the line ran, and how it was punished

Three distinctions do most of the work. The first is the source of the power. The royal touch and the cauldron use comparable means, yet one is a "miraculous work" and the other "a deed without a name". Shakespeare's Joan protests that she was "chosen from above, / By inspiration of celestial grace", and that her accusers think it "a thing impossible / To compass wonders but by help of devils" 1H6 5.4.36–53. Holinshed sides with her judges. Her saints were "wicked spirits", and she was condemned for "diuelish witchcraft and sorcerie" Holinshed, Henry VI 128–129. The second is natural against demonic virtue. A friar may keep drugs, while Mantua's law kills an apothecary who sells "mortal drugs" Rom. 5.1. Rosalind, joking, claims a teacher "most profound in his art and yet not damnable" AYL 5.2.58–70, a joke that needs the distinction to be common knowledge. The third is law. Brabantio arrests Othello as "a practiser / Of arts inhibited and out of warrant" Oth. 1.2.62–79. When Hermione's statue moves, Leontes answers "If this be magic, let it be an art / Lawful as eating" WT 5.3.85–111. These echo the statutes, though the Witchcraft Acts of 1563 and 1604 are not themselves in the library; the second is cited once, in a prosecution described below Mompesson to Collins (1674), in Saducismus Triumphatus.

Two documents of enforcement are. Cardinal Pole's visitation articles under Mary I ask churchwardens to report anyone who practises "anie arts of magike, or necromancie … or be vehementlie suspected thereof" Holinshed, Mary I 341, between drunkards and irregular marriages. Suspicion alone could be reported. The other is James VI and I's manual of kingship for his son.

so is there some horrible crimes that yee are bound in conscience neuer to forgiue: such as Witch-craft, wilfull murther, Incest, (es- pecially within the degrees of consanguinitie) Sodomie, poisoning, and false coine.

There are some horrible crimes that you are bound in conscience never to pardon: witchcraft, wilful murder, incest (especially between blood relations), sodomy, poisoning and counterfeiting.

Witchcraft heads the list, ahead of murder. Later James tells the prince to consult "no Necromancier nor false Prophet, vpon the successe of your warres, remembring on king Saules miserable end" Basilikon Doron 2.31.

Punishment in the chronicles is graded by rank and sex. In the Cobham case the duchess did public penance and was imprisoned for life, the witch Jourdemaine was burned, and the cleric Bolingbroke was hanged, drawn and quartered, swearing to the end that "there was neuer anie such thing by them imagined" Holinshed, Henry VI 252–253. Joan and the Forres witches were burned; Peter of Wakefield was hanged with his son. The judge read Penbrooke's sudden death as "the iust iudgement of God towards those that vsed sorcerie" Holinshed, Elizabeth 2.6. Below the courts, Ford beats the "witch" out of his house, and Falstaff, in her clothes, nearly lands "i' the common stocks, for a witch" Wiv. 4.5.

Religion could move the line. James's Premonition turns the charge on Rome: "For their Sorcery consider of their Agnus Dei, that will slocken fire; of the hallowed shirts", and of prayers against shot and thunder, "And iudge, if this be not very like to Sorcerie and incantation of charmes" James I, Premonition 1.87. In Protestant eyes, Catholic sacramentals were charms.

Doubt, mockery and exposure

Nearly every play that stages magic also stages doubt. Banquo wonders whether the witches were earth's "bubbles" or whether "we have eaten on the insane root / That takes the reason prisoner" Mac. 1.3.79–85, a plant traced to North's Plutarch, where Antony's starving soldiers ate a herb that "made them out of their witts" North's Plutarch, Antony, in MacCallum. Glendower boasts "I can call spirits from the vasty deep", and Hotspur asks "But will they come when you do call for them?" 1H4 3.1. The sharpest attack is on astrology, and Shakespeare gives it to a villain.

This is the excellent foppery of the world, that, when we are sick in fortune … we make guilty of our disasters the sun, the moon, and the stars: as if we were villains by necessity; fools by heavenly compulsion

This is the world's great stupidity: when our luck turns bad … we blame our disasters on the sun, the moon and the stars, as if we were villains because we had to be and fools because heaven forced us.

Then Edmund uses the same jargon to deceive his brother Lr. 1.2. The comedies of the 1590s treat professional magicians as cheats. Antipholus arrives in Ephesus expecting them.

They say this town is full of cozenage, / As, nimble jugglers that deceive the eye, / Dark-working sorcerers that change the mind, / Soul-killing witches that deform the body, / Disguised cheaters, prating mountebanks

They say this town is full of fraud: quick-handed tricksters who fool the eye, sorcerers working in the dark who alter the mind, soul-killing witches who deform the body, cheats in disguise, chattering quacks.

The list runs from magic into swindling without a break, and every wonder in the play turns out to be a pair of twins. The exorcist Pinch diagnoses possession by "gazing in mine eyes, feeling my pulse" Err. 5.1.237–246, and Feste in Twelfth Night exorcises a sane man TN 4.2.25–34. Edgar in King Lear feigns possession by devils whose names, "Flibbertigibbet" among them Lr. 4.1.57–64, scholars trace to Samuel Harsnett's 1603 exposé of Catholic exorcism, which the library does not hold.

Three famous charges in the plays are shown to be fabricated. Richard III bares a withered arm and blames "Edward's wife, that monstrous witch" and Jane Shore R3 3.4.59–75. His source, Holinshed copying Thomas More, adds five words, "as it was neuer other" Holinshed, Edward V 95. In 2 Henry VI the priest Hume admits he was hired "to undermine the duchess / And buz these conjurations in her brain" 2H6 1.2.87–107, and in the next act Duke Humphrey exposes a false miracle by asking a man "born blind" to name colours 2H6 2.1.124–135. Othello answers a charge of love magic with a story.

She loved me for the dangers I had pass'd, / And I loved her that she did pity them. / This only is the witchcraft I have used

She loved me for the dangers I had come through, and I loved her because she pitied them. That is the only witchcraft I have used.

Brabantio cannot accept that his daughter chose a Black man freely, so he concludes it was witchcraft, and the charge measures the accuser. The chroniclers doubt in their own register. Hooker rejects bewitchment as the cause of the Earl of Essex's death in Dublin as unchristian, "and therefore not to be credited" Hooker, Elizabeth 115. Harrison says King John abolished the ordeals because "sorcerous and artificiall practises" kept fire and water from working Harrison, Description II.9; historians credit the Fourth Lateran Council of 1215. Sidney calls the Roman habit of opening Virgil for omens "a verie vaine and godlesse superstition" Sidney, Defence 8.1. His poet, by contrast, "neuer maketh any Circles about your imagination, to coniure you to beleeue for true, what he writeth" Defence 48.1.

Medea's prayer in Prospero's mouth

England met ancient magic mostly in translation, and the library lets one transfer be followed through three texts. In Ovid's Metamorphoses the witch Medea goes out at midnight for the herbs that will make old Aeson young. She calls on Night, on "triceps Hecate", the three-formed goddess of witches Ov. Met. 7.194, and on the powers of earth, air and water.

quaeque magos, Tellus, pollentibus instruis herbis, / auraeque et venti montesque amnesque lacusque / dique omnes nemorum, dique omnes noctis adeste.

And you, Earth, who furnish the magi with potent herbs; you breezes and winds, mountains, rivers and lakes; all you gods of the groves, all you gods of the night: be present.

With their help she has turned rivers back, moved forests, shaken mountains, and made "manesque exire sepulcris", the dead come out of their tombs Ov. Met. 7.206. Arthur Golding, a Puritan who also translated Calvin, Englished it in 1567.

Ye Charmes and Witchcrafts, and thou Earth which both with herbe and weed / Of mightie working furnishest the Wizardes at their neede: / Ye Ayres and windes: ye Elves of Hilles, of Brookes, of Woods alone, / Of standing Lakes

You charms and witchcrafts, and you Earth who supply wizards in their need with herbs and plants of mighty working; you airs and winds; you elves of hills, of brooks, of woods, of still lakes …

Golding made Ovid's magi into "Wizardes" and the gods of the groves into English elves, and he made Ovid's lakes "standing Lakes". Some forty years later Prospero, about to give up his art, speaks the same prayer.

Ye elves of hills, brooks, standing lakes and groves … I have bedimm'd / The noontide sun, call'd forth the mutinous winds … graves at my command / Have waked their sleepers, oped, and let 'em forth / By my so potent art.

You elves of hills, brooks, still lakes and groves … I have darkened the midday sun and called out the rebellious winds … at my command graves have woken their sleepers, opened and let them out, by my most powerful art.

The structure is Ovid's: an invocation of the spirits of places, then "by whose aid", then a catalogue of feats. Golding's elves and standing lakes survive word for word, and the darkest feat, opening graves, comes straight from the witch. Shakespeare drops Hecate and the herbs and adds fairy rings and Jove's thunderbolt. Many critics take this to mean that Prospero's art and Medea's are one kind. The renunciation follows at once.

But this rough magic / I here abjure … I'll break my staff, / Bury it certain fathoms in the earth, / And deeper than did ever plummet sound / I'll drown my book.

But this crude magic I here renounce … I will break my staff, bury it many fathoms in the earth, and drown my book deeper than any sounding-line has ever reached.

"Abjure" was the word for formally renouncing heresy. Prospero, who prized his books "above my dukedom" Tmp. 1.2.168, can return to Milan only once they are drowned. Reading the speech as Shakespeare's farewell to the stage is a modern inference.

Other ancient material travelled the same way. Medea's Hecate reappears in Hamlet and Macbeth. North's Plutarch gave Shakespeare Caesar's sacrifice, in which "one of the beastes which was sacrificed had no hart" North's Plutarch, Caesar, in MacCallum, and the soothsayer's "Beware the ides of March" JC 1.2.15–24. North, following Amyot's French, also Christianised Plutarch's Greek: Antony's guardian daimon becomes "the good angell and spirit that kepeth thee" North's Plutarch, Antony, in MacCallum, which Macbeth echoes in "My Genius is rebuked" Mac. 3.1.56.

Change over time

The collection runs from Golding (1567) and Holinshed (1577, revised 1587) to The Tempest (c. 1610–11). Three shifts are visible. The first is from chronicle to stage, as when the Cobham wax image becomes a spirit-raising with thunder. The second is a change of tone. The 1590s comedies treat conjurers and exorcists as frauds, while the Jacobean plays give magic real if ambiguous power. Scholars link this to the accession in 1603 of a king who had written on witches. His Daemonologie (1597) shows what that king believed Daemonologie I.5 (see 'The king's theory' above), but not that the playwrights wrote to please him, so the link itself still rests on outside evidence. The third shift is a theme the plays name. In Henry V (1599) a churchman says "It must be so; for miracles are ceased" H5 1.1.67. A few years later, after Helena's cure of the King of France, Lafew goes further.

They say miracles are past; and we have our philosophical persons, to make modern and familiar, things supernatural and causeless. Hence it is that we make trifles of terrors

People say the age of miracles is over, and we have our natural philosophers to make supernatural and inexplicable things seem ordinary and everyday. That is why we make light of things that should terrify us.

The doctrine that miracles had ended with the apostles, together with a natural philosophy that explained wonders away, left a gap. The late romances fill it with wonders that prove natural: Hermione alive all along, Thaisa revived by medicine, Imogen's death a drug. And the last great magician of the English stage lays down his art of his own accord.

What is distinctive here

Beside the other cultures in this report, Renaissance England stands out in five ways. Its sources are visible. For several plays the library holds chronicle and drama, and for Prospero it holds the Latin, the English intermediary and the play, and for Faustus the prose source and both printed texts of the play, so the reworking of magical material can be watched at a resolution no other culture here allows. The blood-fed familiar is an English motif with no parallel in the Greek and Latin corpora. Magic charges are weapons inside the royal family, against Cobham, Hastings and Jane Shore, and the texts themselves expose two of them as pretexts. The frame is confessional. A Protestant king calls Catholic sacramentals sorcery, and Lear borrows its devils from an anti-Catholic exposé. And the theatre itself is treated as magic, with spirits as actors and revels that melt "into air, into thin air". Its chroniclers hedge, its playwrights stage the spell and the sceptic together, its king writes the theory and legislates, and a Kentish gentleman argues that the whole crime is a cheat Scot, Discoverie XVI.2. What its texts most often show is the accusation of magic, who makes it and why, and what it costs the accused.

Chapter XI

How magic differs across the cultures

Set side by side, the library's cultures share most of their techniques and differ mainly in who was allowed to use them, what the power was thought to be, and who wrote the record.

The chapters so far have taken one culture at a time. This one asks the same questions of all of them at once. Who called what "magic"? Who practised it? Where did each society draw the line between acceptable and forbidden ritual, and how did it punish those who crossed it? How was the thing thought to work? And which techniques, figures and phrases can be followed from one culture to the next inside this one library?

The comparison rests on the verified survey: 1,343 passages from fifteen collections, from the Pyramid Texts carved around 2350 BCE to Shakespeare's last plays. Each passage was tagged for its techniques, the sex of the practitioner and the author's stance. Counting those tags cannot say what ancient people did. It can say what kinds of text each culture left in this collection, and that turns out to explain a good deal of the apparent difference between cultures.

1. The words, and who used them

In every culture of the library one set of words names the speaker's own rituals and another names the rituals of people the speaker fears or despises. Those who used the first set seldom meant by it anything like our "magic". In the Pyramid Texts ḥkꜣ (heka) is a power that gods possess and the king eats and becomes. A crown goddess is called "Great of Magic" (PT 220–221), and nothing in the corpus treats heka as suspect except when an enemy holds it.

č̣(ṭ) mṭw ꜣwꞽ p.t śṭꜣ tꜣ ṭp-ꜥw(.wꞽ) Ppy pn / Ppy pw ḥkꜣ.w / Ppy pw ẖr(.ꞽ) ḥkꜣ

Words to be spoken. The sky stretches out (the verb is uncertain), the earth quakes before this Pepi. Pepi is heka. Pepi is the one who holds heka.

The Mesopotamian exorcist's craft, āšipūtu, was likewise a branch of scholarship, and its patron Marduk is himself called "exorcist of the gods" (RINAP 3 153 o 15). No medieval German charm in the library calls itself zouber; a few call what they do a segen, a "blessing". Old Norse galdr was an Odinic art of chanted verse before Christian law paired it with gerningar and condemned both. Latin carmen covered a hymn, a legal formula and a spell, and was sinister only by context.

The second set of words is a vocabulary of accusation. Akkadian kišpū, the secret harmful work of the kaššāpu and above all the female kaššāptu, is always somebody else's doing. Its Hebrew relative, by scholars' usual etymology, is the mĕkaššēpâ of Exodus, the "sorceress" who may not live. Greek γόης (goēs) and μάγος (magos), Latin veneficus and maleficus, German zouberære, Norse fordæða and English "witch" all work this way. They are labels applied from outside, in court, in sermons and in insults.

The Greek case shows a word moving from the first set to the second within a few decades. Herodotus, writing around 430 BCE, describes the Persian magus as a priest whose presence makes a sacrifice lawful.

Μάγος ἀνὴρ παρεστεὼς ἐπαείδει θεογονίην, οἵην δὴ ἐκεῖνοι λέγουσι εἶναι τὴν ἐπαοιδήν· ἄνευ γὰρ δὴ Μάγου οὔ σφι νόμος ἐστὶ θυσίας ποιέεσθαι.

A Magus stands beside it and sings a song of the gods' birth, which is what they say the incantation is. Without a Magus it is not their custom to sacrifice at all.

A dialogue attributed to Plato defines the magus's art, μαγεία, as "the service of the gods" ([Pl.] Alc. 1, 122a). By the late fifth century BCE the Hippocratic author of On the Sacred Disease was lumping μάγοι with purifiers, begging priests and charlatans (Hippoc. Morb. Sacr. 1), and Gorgias paired μαγεία with γοητεία as two arts of deceiving the mind (Gorg. Hel. 10). The foreign priest's title had become an insult for a Greek rival.

The few learned writers who use the outsiders' word without hostility attach a definition. Apuleius, on trial for magic in Roman Africa around 158 CE, argued that in Persian a magus is simply a priest (Apul. Apol. 25), and Origen, a Christian with no wish to practise it, describes magic as "a consistent system with principles known to very few" (Orig. Cels. 1.24). Outside Egypt, no culture in the library has a neutral word for magic as a whole. The general category comes into being as the name for other people's rituals.

2. Who practised

State scholars

The oldest corpora show magic in the hands of literate, salaried, mostly hereditary men attached to a king or a temple. The Mesopotamian evidence is the clearest. Exorcists, diviners, dream-interpreters and lamentation priests appear as servants of the gods and of the king. Colophons name the families that held the craft, such as the Assur house of Kiṣir-Nabû and Šamaš-ibni. The exorcist claims no power of his own.

ša₂ {⸢d}e₂⸣-a a-na-ku ... ša₂ {d}DAM.KI.NA a-na-⸢ku⸣ ... DUMU šip-ri ša₂ {d}AMAR.UTU a-na-ku ... ša₂ qa-ti {d}e₂-a ina qa-ti-ia i-ba-aš₂-ši

I belong to Ea. I belong to Damkina. I am Marduk's envoy. Ea's own hand is the hand I work with.

Egypt's Pyramid Texts are recited by a lector priest on behalf of a king who is himself the magician. Esarhaddon, after taking Memphis around 671 BCE, carried off Egyptian ḫarṭibu experts together with exorcists and snake charmers as war booty (RINAP 4 9 i' 9'–11'). The word is usually derived from the Egyptian title of the chief lector priest, and it is the word behind the Hebrew ḥarṭummîm who fail to read Pharaoh's dream (Gen 41.8). Herodotus' Persian magi and Rome's Etruscan haruspices fill the same niche. Claudius set the haruspices up as Italy's oldest discipline against "foreign superstitions" (Tac. Ann. 11.15), and even Constantine allowed them to practise in public (CTh 9.16.1). Israel had its own sanctioned diviners in the priests who carried the Urim (Exod 28.30).

An outsider would call what these specialists did magic. Inside their own culture it was public service, recorded in believing, institutional texts. No passage in the Mesopotamian collection mocks the exorcist. The nearest thing to doubt is the righteous sufferer of Ludlul, who complains that dream-interpreter and exorcist both failed him (Ludlul II, comm. K.3291).

Freelancers

Beside the state experts, every literate culture in the library describes freelancers, and it is to them that hostile words stick. Plato's begging priests and seers go to rich men's doors offering to settle a family's guilt or harm an enemy for a fee (Pl. Resp. 364b–c). Cato tells his slave farm-manager not to consult a haruspex, an augur or a Chaldaean (Cato Agr. 5.4). Ezekiel's women work for handfuls of barley and pieces of bread (Ezek 13.18–19). Caesarius of Arles, in the sixth century, scolds his peasants for running to caragii, diviners and lot-casters when they fall ill (Caes. Arel. Serm. 54). The Norse seeress Heiðr goes from farm to farm (Vsp. 22). The recurring marks of the suspect specialist are payment, mobility and foreignness.

Women, men and the counts

The image of magic as a woman's art is strongest in the corpora that are most literary. The surveyors recorded the practitioner's sex for 942 of the 1,343 passages. Grouping those records into male, female and mixed gives 606 passages with male practitioners and 247 with female ones. The share varies from corpus to corpus far more than the overall figure suggests.

Two further cuts sharpen the pattern. By the author's stance, women are 63 per cent of named practitioners in literary fiction and 48 per cent in satire, but 6 per cent in practical instruction. The only two women who give instructions in the whole set are Circe and Medea, both inside poems (Hom. Od. 10.508–540; Ap. Rhod. 3.1026–1062). By technique, women are 71 per cent of named practitioners of love magic and 66 per cent for drugs and poisons. For astrology the figure is 6 per cent, for exorcism 10, for theurgy 15. Rune-magic has thirteen male practitioners and no female ones.

So the library holds two magicians. One is male, learned and institutional, the writer of manuals or the hero of learned tales: the āšipu, the lector priest, Óðinn, the theurgist, Pancrates of Memphis with his animated pestle (Lucian Philops. 34–36), Faustus with his blood-signed contract (English Faust Book 6) and Prospero with his books. The other is female and usually seen from outside, through poetry, satire or accusation: Circe, Medea, Horace's Canidia burying a boy to make a love-potion from his liver (Hor. Epod. 5), Lucan's Erichtho, the kaššāptu and Macbeth's sisters. Pliny states the Roman assumption plainly.

durat tamen tradita persuasio in magna parte vulgi, veneficiis et herbis id cogi eamque unam feminarum scientiam praevalere.

Still, most ordinary people keep the old belief that eclipses are compelled by potions and herbs, and that this is the one science in which women excel.

Euripides' Medea had already claimed that women are "by nature most skilled" in poisons and sworn by Hecate at her hearth (Eur. Med. 384–398). The witch as a type is far older than Euripides, though. In the Assur commentary on Maqlû the female witch binds her victim, and her kind pluck his hair and gather the dust of his footprints (KAR 94 7'–11'). She is the only female practitioner in the Mesopotamian passages surveyed, and she appears there as the enemy. The anti-witchcraft rituals added to the library since also have women perform protective rites at the expert's direction: a pregnant woman slips three times beneath a pregnant jenny and tells it "May your baby die, and may my baby live!", and on the day of her labour a young girl makes the figurines (CMAwR 5.5.2:11–16).

The North complicates the picture. Snorri gives seiðr to Óðinn himself and then explains why it passed to women.

Oðinn kɴe þa iþrott er meſtr mattr fylgðe. oc framþi ſialfr er ſeiðr heitir. ... En þessi fiolkyɴge er framit er fylgir ſva micil ergi at eigi þótte karl mǫnnom viþ at fara. oc var gyðionom kend ſ iþrót.

Óðinn knew the art that carries the most power, called seiðr, and practised it himself. But such great unmanliness (ergi) goes with this sorcery that men thought it shameful to meddle with it, and so the art was taught to the priestesses.

Here the shame attaches to one technique, seiðr. Rune-carving and galdr stay male and honourable: Egill cuts runes into a poisoned horn, reddens them with his own blood and the horn bursts (Egils saga 44). Male seiðmenn are killed. King Haraldr's son Rögnvaldr learned sorcery and was burned with eighty seiðmenn, "and that deed was much praised" (HHárf. 34). The Norse texts police a man doing a woman's ritual, which is a different worry from the Greek and Roman fear of women's drugs.

In the medieval German collection the female sorcerer appears almost only in courtly romance adapted from French and Latin. The sorcerers of religious narrative are men such as Simon Magus and the Jewish master Zambri (Kaiserchronik 10043ff.), and the word hexe does not yet occur. Where texts look inside households, women reappear as ordinary practitioners. The Norse law fines a mother who bites off her baby's finger "for long life" (Borgarþ. kristinréttr I 16). The Norwegian homilist condemns belief in women's "life-herbs" (Homily Book). Athenian vases show babies wearing strings of amulets (Harvard 1960.342), and Dante shows women who "left the needle, the shuttle and the spindle" to make spells with herbs and images (Inf. 20.121–123). The near-absence of women from the instruction texts says more about who could write those texts (temple scholars, monks, physicians) than about who used charms.

3. Where each culture drew the line, and how it punished

Across the library the line between proper ritual and forbidden magic usually separates two settings for the same technique. Four tests recur almost everywhere.

1. Public or secret. Plato bans private shrines because private cult opens the way to secret propitiation (Pl. Leg. 909e–910d). Apuleius defines the criminal art as secret, nocturnal, solitary and muttered (Apul. Apol. 47). Constantine forbids a haruspex to cross a private threshold while permitting public rites (CTh 9.16.1).
2. For or against the community. The Mesopotamian exorcist "loosens" and the witch "binds", though both make figurines, and in Maqlû the bewitched man binds the witch's likeness in return (Maqlû VII 67). Constantine punishes magic against health or chastity but protects healing charms and spells against hail (CTh 9.16.3).
3. Ours or theirs. Deuteronomy's list of forbidden specialists is a list of "the nations" Israel is to replace, with a prophet "like me" as the alternative (Deut 18.9–15). The Norse laws make the journey to consult Sámi seers an unatonable crime (Borgarþ. kristinréttr I 16). In the Old French epics the charge of enchantment flows from Saracens to Christians, and the Christians' marvels are credited to God (Otinel 1319–1324).
4. Given or sold. Peter's answer to Simon, who offered money for the power of the Spirit, made the sale of sacred power a sin with Simon's name (Acts 8.9–24).

The clearest sign that technique does not decide the matter is that the same act appears on both sides of the line within a single culture. A Roman consul who devoted himself and the enemy army to the gods below performed the most honoured act of state religion (Livy 8.9). Lead tablets bearing Germanicus' name, found in the walls of his sickroom, were a capital crime (Tac. Ann. 2.69). Moses' bronze serpent cures by looking, on God's command (Num 21.9). Lucan defines his necromancer by what she omits.

Nec superos orat, nec cantu supplice numen / Auxiliare vocat, nec fibras illa litantes / Novit

She makes no prayer to the gods above, calls on no helping god with a humble chant, knows nothing of entrails that show a sacrifice accepted.

Roman epic lets state seers consult the dead as well, and they are pious because they pray. Martin of Tours makes the shade of a robber appear and confess, and his poet presents this as miracle because it comes by prayer (Paul. Petric. Mart. 2.196–221).

Punishment followed the line, and here the cultures really do differ. The Mesopotamian library holds no law against witchcraft. The witch is punished in ritual, when the exorcist makes a clay figure of her and pours kiln slag and ashes over its head (KAR 94 37'–40'). The Pyramid Texts know no human sorcerer to punish. Israel's laws prescribe death (Lev 20.27). Athens executed the Lemnian drug-woman Theoris "and all her family" ([Dem.] 25.79), and Plato's model code grades penalties by expertise, death for a physician or seer who harms, while a layman's penalty is left to the court (Pl. Leg. 933d–e). Rome had the Twelve Tables' clauses against charming away crops (Pliny NH 28.17), the Sullan law on poisoners (Cic. Clu. 148), treason trials for consulting astrologers about the emperor and repeated expulsions of magi, one of whom was thrown from the Tarpeian Rock (Tac. Ann. 2.32).

The Norse sagas remember killing without trial. Kotkell and Gríma are stoned, and their son's head is bagged so that his evil eye cannot reach the land before he is drowned (Laxdæla saga 37). Óláfr Tryggvason gets the sorcerers drunk and burns the hall over them (ÓT 62). The Christian laws that followed were punitive but careful about evidence. An accusation must come from three households, she may clear herself by the oaths of six women or by ordeal, and an accuser from a single household whose charge fails suffers the penalty he sought for her (Sverris kristinréttr; Borgarþ. kristinréttr I 16). The Church's own sanction was penance. Elvira refused communion even on the deathbed to anyone who killed by sorcery (Conc. Elvira can. 6), Basil gave the confessed sorcerer a murderer's term (Basil Ep. 217 can. 65), and Columbanus priced love-magic that harmed no one at a year on bread and water for a cleric (Columb. Paenit. B 6).

4. How it was supposed to work

Six theories of mechanism run through the library. Most cultures held several at once.

Delegated divine authority. In Mesopotamia and Egypt the power belongs to the gods, and the practitioner borrows it by identity or commission. The exorcist "belongs to Ea". The Egyptian king eats the gods' heka and holds it "in his belly" (PT 274 §403), and he may threaten a god who refuses him a place in the sky (PT 254 §277). The Hebrew Bible keeps the divine source and abolishes the technician. Effects follow from God's will, and the prophet replaces the diviner.

The power of names. Knowing a being's name gives standing over it in the Pyramid Texts (PT 301 §449). Two and a half thousand years later Origen argues that names belong to their bearers by nature and lose their force in translation (Orig. Cels. 1.24–25), and the Hermetic teacher tells a king to keep the Egyptian words untranslated (Corp. Herm. 16.1–2). The belief reaches the German herbalist who adjures verbena by the seventy-two names of God (Bartholomäus, verbena). Galen drew his line between medicine and sorcery at exactly this point. A hung substance might work, but "barbarous names" were the sorcerer's stock (Gal. SMT 10.2.21).

Sympathy and analogy. "As I melt this wax, so may Delphis melt" (Theoc. Id. 2.28–31) is the purest statement of a principle every culture used and some theorised. Plotinus turned it into cosmology.

καὶ γὰρ μηδενὸς μηχανωμένου ἄλλου πολλὰ ἕλκεται καὶ γοητεύεται: καὶ ἡ ἀληθινὴ μαγεία ἡ ἐν τῷ παντὶ φιλία καὶ τὸ νεῖκος αὖ. καὶ ὁ γόης ὁ πρῶτος καὶ φαρμακεὺς οὗτός ἐστιν

Many things are drawn and spellbound even when no one is working on them. The real magic is the love and the strife at work in the whole universe. That is the first sorcerer and the first maker of drugs.

Belief. Greek prose is where the library first proposes that spells work through the minds of those involved. Gorgias calls magic a deception of opinion. Plato says binding spells persuade the attacker that he can harm and the victim that he is harmed (Pl. Leg. 932e–933c), and he advises the lawgiver to forbid such spells without trying to argue people out of their fear.

Demons. Diotima places all divination, initiation, incantation and sorcery in the daimonic realm between gods and men (Pl. Symp. 202e–203a). Porphyry assigns harmful magic to evil daimones fed on smoke and blood (Porph. Abst. 2.40–42). Christian writers made this the single explanation for everything. Once magic was demonic, the hard question was which natural remedies were still allowed, and Augustine's test became the standard answer.

Aliud est enim dicere, Tritam istam herbam si biheris, venter non dolebit; et aliud est dicere, Istam herbam collo si suspenderis, venter non dolebit. Ibienim probatur contemperatio salubris, hic significatio superstitiosa damnatur.

To say "Grind this herb and drink it and your stomach will stop aching" is one thing. To say "Hang this herb round your neck and your stomach will stop aching" is another. The first is approved as a healthy mixture. The second is condemned as a superstitious sign.

Signs, stars and hidden properties. Mesopotamian scholars read the world as writing. Gods wrote on livers, births and the sky, and a scholar could even re-read a divine decree, so that Marduk's seventy years of desolation for Babylon became eleven (RINAP 4 104 ii 2–9). Isidore split astrology into a natural part, the calendar and eclipses, and a superstitious part, the birth horoscope (Isid. Etym. 3.27). The Norse texts locate power in knowledge itself. Fjǫlkynngi means "much-knowing", and Snorri says Óðinn's arts were taught in runes and songs (Yng. 7). Greek naturalists and their medieval heirs located it in the hidden virtues of plants, stones and animals, and the Old French epics give their cups an innate "vertu" that works only for the sinless (Huon de Bordeaux 3646–3679).

Laid out in time, the theories move. The oldest texts place the power in the speaker's divine warrant. Classical Greek writers push it into nature and into belief. Late antique philosophers let tokens and names act by themselves, "even when we are not thinking of them" (Iambl. Myst. 2.11). Christianity gives all of it to demons except what can pass as nature or as God's.

5. Techniques that recur and travel

Binding and figurines

Binding appears in every collection surveyed, and no technique changes hands more visibly. The Mesopotamian witch binds and the exorcist loosens, though in Maqlû the bewitched man binds the witch's likeness in return (Maqlû VII 67). The Furies dance a "binding song" around Orestes (Aesch. Eum. 306–396). Plato's "so-called bindings" come with wax images left at doors, crossroads and parents' tombs. In Latin the verb is defigere, to pin. Ovid wonders whether a witch has "pinned my name in red wax" and driven needles into the liver (Ov. Am. 3.7.29–30). Horace's witches set a wool doll over a wax one (Hor. S. 1.8), and Virgil's singer hardens a clay figure and melts a wax one in the same fire (Verg. Ecl. 8). Holinshed's witches at Forres roast a wax image of King Duff, who sweats as it melts (Holinshed, Duff), and Dante's women work with "herbs and images". The figurine travels intact across more than two thousand years. What changes is its owner. In Assur it was the licensed exorcist's weapon against the witch, though Maqlû accuses the witch of making figurines of her victim from nearly the same materials (Maqlû IV 36–38; see 'Techniques in the practitioners' own texts' below). In Rome and Scotland it was the witch's weapon.

Love magic

Love magic is the most female-coded technique in the counts and the best-documented in Greek and Latin poetry. It is almost absent from Mesopotamia (2 per cent of passages) and the Hebrew Bible (2 per cent), against 27 per cent in Latin poetry. The Christian penitentials describe women's love magic from the confessor's side, as with a wife who mixes her husband's semen into his food to make him love her more (BnF lat. 12270). In Greek tragedy and the Athenian courts the love charm repeatedly turns out to be a poison. Deianira's love-charm kills Heracles (Soph. Trach. 555–581), and Athens heard cases where a philtron was said to have killed (Antiph. 1.9). That is one reason φάρμακον and venenum never separated "drug" from "spell".

Drawing down the moon

Some techniques belong to one cultural sphere only. The claim that witches pull the moon from the sky is purely Greek and Roman, and its home is Thessaly. The Hippocratic author scoffs at men who say they can bring down the moon by magic and sacrifice (Hippoc. Morb. Sacr. 3–4). Aristophanes' Strepsiades plans to buy a Thessalian witch to pull the moon down so that his monthly interest never falls due (Ar. Nub. 749–756). Pliny sets the belief against astronomers who could now predict eclipses to the hour (Pliny NH 25.9–11). Babylon, the home of eclipse science, had no such belief in the library. There an eclipse was a sign to be read and, by the fourth century BCE, forecast by explicit rules (SpTU 4 162). The same event was a message from the gods in Mesopotamia and a witch's crime in Thessaly.

Necromancy

Consulting the dead appears in nearly every corpus, and each gives it a different moral weight. In the Odyssey it is a heroic rite of pit and blood (Hom. Od. 11). In Israel the woman of Endor succeeds, though the practice is banned (1 Sam 28). Lucan makes it the supreme horror (Luc. 6.667–830), and Dante borrows Lucan's witch to say that Erichtho once forced Virgil himself down to fetch a soul (Inf. 9.22–27). In the North, Óðinn preserves the severed head of Mímir with herbs and chanted spells so that it can answer him (Yng. 4). The Mesopotamian material treats the dead chiefly as a danger to be kept off. A queen's tomb is protected by an oath that turns a robber into a restless, thirsty ghost (RINAP 1 2003). An Uruk tablet added to the library since also gives the wording for a ghost to appear and make a decision (SpTU 2 20 r 22–24; see chapter II).

Wolves and beasts

Transformation into animals shows the same motif under several theories. Circe's drug turns men into pigs while they keep their minds (Hom. Od. 10.233–243). Herodotus reports, without believing it, that the Neuri become wolves for a few days each year (Hdt. 4.105). Petronius' freedman tells a werewolf story among the tombs (Petron. Sat. 62). Augustine heard of Italian innkeeper-women who turned travellers into pack animals with drugged cheese. He would not allow that any body changed and explained the transformation as an image seen by others while the drugged man slept (Aug. Civ. 18.18). The Norse texts have their own explanation in the hamr, a shape that can leave the body. When a Norwegian translator rendered a French werewolf lay he used native words, vargulfr and "changing one's hamr" (Strengleikar 5).

Healing with a story

The historiola, a short sacred story recited so that its outcome will repeat in the patient, is the signature of the medieval German charms.

Krist unde Johan giengen zuoder Jordan. dô sprach Krist stant Jordan. biz ich unde Johan über dich gegân. alsô Jordan dô stuont. sô stant dû .N.. illius bluot. hoc dicatur ter. et singulis vicibus fiat nodus in crine hominis.

Christ and John walked to the Jordan. Christ said, "Stand still, Jordan, until John and I have crossed you." As the Jordan stood then, so may you stand, blood of N. [Latin:] Say this three times, and each time tie a knot in the patient's hair.

The structure is old. Pyramid Text offerings are made effective by declaring each one to be the Eye of Horus. A Babylonian birth incantation tells of the moon-god's cow in labour (OIC 22 131 no. 20). A Latin fever charm in a twelfth-century manuscript cites Christ's healing of Peter's mother-in-law (BL Egerton 821). The oldest Greek blood-stopping charm in the library is a single word in Homer. Autolycus' sons bound Odysseus' wound and checked the dark blood "with an incantation", its words unrecorded (Hom. Od. 19.457). In the Strasbourg version Jordan is a person shot by one Genzan, and whether that is a pre-Christian survival is disputed (Straßburger Blutsegen).

Amulets

Amulets are the most ordinary magic in the library and the one most often argued over by physicians. Galen tested a peony root against a child's epilepsy. The child had no fits while wearing it, fell ill when it slipped off, and recovered when it was replaced (Gal. SMT 6.3.10). An Old Norse herbal tells what looks like the same case under the name of Dioscorides, though its boy is "eight winters" old where Galen's had worn the root for eight months (AM 655 XXX 4to). The Council of Laodicea forbade clergy to make phylacteries (Conc. Laodicea can. 36), yet Gregory the Great sent the Lombard prince "phylacteria" containing a relic-cross and a Gospel text (Greg. Reg. 14.12). The word survived the Church's condemnation, and so did the practice once its contents were changed.

6. Lines of transmission

Several threads can be followed from text to text inside the library.

From Persian priest to Christian demon-worshipper. The magus begins as Herodotus' Persian priest and Plato's tutor of princes. He becomes the Greek charlatan in the late fifth century BCE. In Pliny's history he practises a single world-wide art descended from Zoroaster, an art that grew out of medicine and then took over religion and astrology (Pliny NH 30.3–11). Matthew's magi read a royal star and honour the child (Matt 2.2), while Acts' Simon "practised magic" in Samaria (Acts 8.9). Pliny's opening sentence reached Isidore around 625 almost word for word. Pliny calls magic "the most fraudulent of the arts" and says it plurimum in toto terrarum orbe plurimisque saeculis valuit, "held the most power across the whole world for the most ages" (Pliny NH 30.1). Isidore keeps the clause and changes the cause.

Itaque hæc vaniias magicarum artium n ex tradi. tione angelorum malorum in toto terrarum orbe plurimis saeculis valuit

So this vanity of the magic arts, passed down from the evil angels, held power across the whole world for many ages.

A few lines later Isidore makes the magus a synonym for the maleficus, who kills by the force of a spell without any poison (Isid. Etym. 8.9.9). Pliny's fraud has become the demons' work. Through the Etymologies that taxonomy passed into the medieval schoolroom.

Egypt's reputation. Where Egypt's own texts speak of heka, they are mostly royal and funerary: of the 216 lines of the TLA's Egyptian texts that contain heka, 179 are in funerary books, royal tombs, temple rites for Pharaoh, a court tale, royal teachings and inscriptions or a royal letter, though its medical books also protect private patients against male and female magicians (pChester Beatty 8). Egypt in other people's texts is the school of magic. The thread runs from the experts Esarhaddon deported, to Pharaoh's ḥarṭummîm, to the Septuagint, where Alexandrian Jews renamed them sophists, φαρμακοί and enchanters (Exod 7.11 LXX). It continues through Lucian's Pancrates and Celsus' claim that Jesus learned his powers in Egypt (Orig. Cels. 1.28) to Jerome's Gaza youth who studies at Memphis for a year before burying a love spell (Jer. V. Hilar. 21). One pharaonic feature can be followed the whole way. Threats against a god, like the king's conditional threats in PT 254, reappear in Iamblichus around 300 CE, who defends the operator's threats to "stop the sacred bark" or scatter Osiris' limbs as aimed at lesser powers (Iambl. Myst. 6.5).

Hecate and Medea. In Hesiod Hecate is a universal benefactress with no magic at all (Hes. Theog. 416–420). By 431 BCE she is Medea's chosen helper at the hearth. In Apollonius she teaches Medea her drugs (Ap. Rhod. 3.528–533). In Ovid she receives Medea's great invocation, in which rivers run back to their sources (Ov. Met. 7.192–219). Around 1200 Herbort von Fritzlar's Medea sends water back to its source and studies the black book still taught at Toledo, and then Herbort remarks that people only imagined it (Herbort, Liet von Troye 545ff.). Golding's 1567 English turns Ovid's gods of the groves into "Elves of Hilles, of Brookes, of Woods" (Golding, Met. 7), and Shakespeare gives those elves, and Medea's claim to have opened graves, to Prospero (Tmp. 5.1.33–50). Macbeth's Hecate, "mistress of your charms", completes the line (Mac. 3.5). Shakespeare's learned magician claims his powers in the words of the archetypal foreign witch.

One verse of Exodus. The Hebrew law, which the survey read in its English version, names a single woman, mĕkaššēpâ in the Hebrew text (Exod 22.17, Hebrew): "You shall not allow a sorceress to live" (Exod 22.18). The Greek translators made her plural and male and turned her craft into drugs, Φαρμακοὺς οὐ περιποιήσετε (Exod 22.18 LXX). Jerome's Latin reads Maleficos non patieris vivere, "evil-doers" (Vulgate Exod 22.18). An early fifth-century verse paraphrase turns it into a duty to denounce, "it is a crime to hide a sorcerer" (Cypr. Gall. Exod. 927). By 1599 James VI was telling his heir that witchcraft headed the crimes a king must never pardon (Basilikon Doron 2). By the time witch-hunters quoted the verse, its target had changed sex and number and its offence had changed more than once.

7. What changed with Christianity and with law

Christianity inherited Israel's ban on foreign specialists, Greek suspicion of paid ritual and Roman criminal law. What it added was a single theory that applied to every culture at once. For Tertullian, Lactantius, Augustine and Isidore, Persian magi, Egyptian priests, Etruscan haruspices, Greek theurgists and Rome's own state cult were all one demonic system. Augustine refused the Platonists' distinction between low goetia and high theurgy and treated them as two names for one cult (Aug. Civ. 10.9). Earlier cultures had words for other people's harmful rites. Christian writers had a theory that made every non-Christian rite the same thing.

At the same time Christianity built a counter-practice that outsiders read as magic. Jesus' own opponents granted that his exorcisms worked and credited them to Beelzebul (Mark 3.22). The name of Jesus, the cross, holy oil and relic dust did the work of pagan amulets and adjurations. The German baptismal exorcism uses the same verbs as the charms, beswern and verblâsen (Baptismal exorcism sermon). The Church's own line was drawn by source and authority, which is the line Mesopotamia had drawn between the exorcist and the witch.

Law moved in three directions. First, the exemptions disappeared. Constantine in 321 protected healing charms and spells for the vintage, and the Visigothic interpretation of 506 punished "storm-senders" along with all other sorcerers (CTh 9.16.3 and interpretatio). Second, the offence widened from harm to association. The Twelve Tables and the Sullan law punished what a spell did. The church councils punished clients for bringing diviners into the house (Conc. Ancyra can. 24). Narbonne ordered the diviners themselves, "free or slave, male or female", to be flogged and sold (Conc. Narbonne can. 14). Magnús's Norwegian law of the late 1260s listed heathen beliefs as offences, among them faith in land-spirits (Magnúss kristinréttr 1–2). Third, torture entered the record, together with clerical unease about it. Gregory of Tours refused to credit part of the confessions Fredegund tortured out of Parisian women (Greg. Tur. Hist. 6.35), and the Council of Mérida in 666 stopped sick priests from torturing church slaves they blamed for their illness (Conc. Mérida can. 15).

A thirteenth-century Icelander saw the change from the far side. Of a woman skilled in magic, the author of Fóstbræðra saga says that because Christianity was then "young and immature" many people thought such skill an accomplishment (Fóstbræðra saga 9). He was describing within his own culture the shift the whole library records, in which a skill became a crime.

The cultures compared

What the comparison shows

Set side by side, the cultures in this library did much the same things and quarrelled over who was allowed to do them. The differences lie in three places. The first is the location of authority: a temple that licensed its own exorcists, a prophet who replaced the technician, a law court that tried poisoners, or a church that saw demons behind every rite but its own. The second is the gendering of blame, which in the Greek and Roman poets falls on women's drugs and in the North on men doing women's rites. The third is the record that survived. Mesopotamia and medieval Germany are known through their working manuals, which are male and confident. Greece and Rome are known mainly through their poets and critics, which are female-haunted and sceptical, though the curse tablets and magical papyri added since let their practitioners speak. Nearly every society in the library also kept for itself, under another name, the arts it condemned in others.

Chapter XII

Did any of it work?

Tested claim by claim against modern evidence, ancient magic worked wherever it carried a real drug, real care, real knowledge or a believing mind, and in no passage does the rite itself move matter, weather or the future.

How the claims were judged

The question has two easy answers, and both are wrong. One says the ancients knew secrets we have lost. The other says magic was all fraud practised on fools. The texts in this library support neither. Their authors were often careful observers of real events, and they were often the first to doubt what they reported. A fair answer has to go claim by claim.

That is what the plausibility panel did. Its members were AI agents (see How this report was made). The corpus surveyors flagged the passages in which a text claims that a rite, substance, object or person produced an effect that modern knowledge can test. The panels merged duplicates across cultures and judged 283 claims in seven groups (the twelve flagged in the Renaissance English texts never reached them): drugs and poisons, medicine, the mind, prediction and weather, chemistry and fire, devices and fraud, and the impossible. In each panel one agent gathered the evidence and drafted the verdicts, two sceptics challenged the drafts they thought too generous or too dismissive, and a reconciler answered the challenges, reopening the ancient passage in the library where it needed to, and gave each verdict a confidence of high, medium or low. The verdicts are these.

• Real physical effect: the effect happens much as the text describes, for the reason it gives or for an obvious material one.
• Real effect, different mechanism: something real happened, but the drug, the dressing or the natural course of the illness did the work the text credits to the rite.
• Psychological effect: the effect is real but passes through the mind of someone who believes in it.
• Knowledge presented as magic: a practitioner who understood a regularity of nature staged it as power.
• Trick or fraud: staging, apparatus or deception, usually exposed by an ancient source.
• Implausible: no known mechanism could produce the effect as claimed.
• Insufficient evidence: the text or the science cannot settle it.

So about four claims in ten rest on something that really happened. Only eighteen happen as described. The rest of the real cases are real by another route, through a substance, a believer or a piece of knowledge that the rite took credit for.

1. Real drugs and poisons

Greek had one word, pharmakon, for a drug, a poison, a dye and a spell, and the woman called a pharmakis could be a herbalist, a poisoner or a witch. Latin venenum was just as wide. That overlap explains why this is the group in which ancient magic most often worked. The substances were real, and the texts describe their effects with an accuracy that modern toxicology still recognises.

The clearest case has no magic in it at all. In 400 BCE the Greek mercenaries retreating with Xenophon reached the Black Sea coast near Trapezus and found beehives.

καὶ τῶν κηρίων ὅσοι ἔφαγον τῶν στρατιωτῶν πάντες ἄφρονές τε ἐγίγνοντο καὶ ἤμουν καὶ κάτω διεχώρει αὐτοῖς καὶ ὀρθὸς οὐδεὶς ἐδύνατο ἵστασθαι

Every soldier who ate the honeycomb lost his wits, vomited and had diarrhoea, and not one of them could stand up.

Xenophon adds that nobody died and that the men recovered within days. Every detail matches poisoning by grayanotoxins, which bees carry from rhododendron nectar into "mad honey". A review of the clinical literature (Jansen and colleagues, Cardiovascular Toxicology, 2012) describes dizziness, low blood pressure and heart block lasting one or two days, rarely fatal, with cases still clustered around the Black Sea, and it cites Xenophon as the earliest report.

Mandrake is the drug that sits closest to the boundary with magic, and the two great handbooks of the first century CE handle it with a pharmacologist's care. The army doctor Dioscorides gives a dose of mandrake wine.

χρώμενοι ἐπὶ τῶν ἀγρυπνούντων καὶ περιοδυνώντων κυάθῳ ἐνὶ καὶ ἐφʼ ὧν βούλονται ἀναισθησίαν τεμνομένων ἢ καιομένων ποιῆσαι.

They use one cupful for the sleepless and for people in great pain, and for those they want to make insensible while they are cut or cauterised.

Pliny gives the same dose, warns that more will kill, and says it is drunk before incisions "so that they are not felt" (Pliny NH 25.147–150). In the same passage he records the root-diggers' rite of drawing three circles round the plant with a sword and digging while facing west. The pharmacology is sound. Mandrake contains scopolamine and hyoscyamine, which sedate and wipe memory, with a narrow margin before delirium and death, and wine extracts them well: one experiment drew about two-thirds of the alkaloids out of nightshade-family material within half an hour (Fatur and colleagues, Plants, 2022). A historical survey in the BMJ (Carter 1996) traces these plants as the sedatives of pre-modern surgery. The panel added one correction. These alkaloids are poor painkillers, so "not felt" in practice meant a stupefied patient who would not remember; later surgeons added opium for the pain. The circles and the westward stance add nothing, though Pliny's rule of keeping the wind at one's back makes sense with a plant whose smell was believed to stupefy.

The poisons are described just as exactly. Nicander's verse handbook on antidotes, from the second century BCE, says that aconite, a plant he sets on the banks of the river of the dead, makes the gums and palate pucker and the victim see double (Nic. Alex. 12–28), which matches the numbness, tingling and heart rhythm disturbance of aconitine poisoning (Chan, Clinical Toxicology, 2009). His opium victim sinks into deep sleep with cold limbs (Nic. Alex. 433–438). Theophrastus records a compound of hemlock and poppy juice made by one Thrasyas of Mantinea, and he flags its promised "easy and painless" death as the maker's own claim (Theophr. HP 9.16.8). Hemlock's coniine paralyses the breathing muscles, as a toxicological study of Socrates' death sets out (Dayan, Postgraduate Medical Journal, 2009). An Icelandic medical fragment of about 1300, copying Dioscorides, recommends white hellebore in milk and honey to kill mice (AM 655 XXX 4to, 3v). The panel judged that pest control accurate on the plant's well-known toxicity, though it checked no modern study for it.

The word for a sorcerer could even name a fisherman. Oppian calls those who stun fish with a paste of clay and cyclamen root "fish-killing pharmaktai" and describes a plain technique (Opp. Hal. 4.647–663). Cyclamen tubers contain saponins (Harvala and Hylands, Planta Medica, 1978), a class of compounds used to poison fish around the world, so the account is credible, but no modern test of cyclamen as a fish poison was found.

Love potions that maddened or killed

The love potion shows the same pattern in a darker form. In an Athenian murder trial of the late fifth century BCE, the prosecutor says his stepmother had once been caught dosing his father and "did not deny it, but said she gave it not to kill him but as a love potion" (Antiph. 1.9). A story told in a work attributed to Aristotle has the Areopagus court acquit a woman whose love potion killed a man, because she meant no harm. Ovid warns that philtres "harm the mind and carry the force of madness". The texts almost never name an ingredient, and the panel treats the usual modern guesses (nightshades, Spanish fly, hellebore) as reasonable inference only. Cantharidin, the "Spanish fly" of later centuries, burns the mouth, damages the kidneys and causes seizures (Karras and colleagues, American Journal of Emergency Medicine, 1996). With drugs like these, amateur dosing closes the gap between flushed excitement and death. The madness and the deaths were real. The love never came from the drug.

The same logic explains why early Christian moral codes put drug-craft beside abortion. The Didache, a church manual of about 100 CE, runs "you shall not practise magic, you shall not use drugs, you shall not kill a child by abortion" in a single breath (Did. 2.2). Ancient abortifacients were real and dangerous. Pennyroyal oil, one of the best known, damages the liver and has killed modern users (Anderson and colleagues, Annals of Internal Medicine, 1996). How well such drugs worked is still argued: John Riddle's thesis that ancient contraceptives were often effective (1992) rests mainly on animal evidence and is contested.

Plausible, but out of reach

Some famous drugs fit real pharmacology but cannot be tested, because the plant is unnamed or the text is poetry. Homer's Helen puts an Egyptian drug in the wine so that a man "would not shed a tear all day" even if his son were killed before his eyes (Hom. Od. 4.220–226). Opium is the obvious candidate, and opium alkaloids have been detected in a Late Bronze Age Cypriot juglet, though only one vessel and at low levels (Smith and colleagues, Analyst, 2018). Circe's drugged barley drink makes Odysseus' men forget their homeland (Hom. Od. 10.235–240), which suits an anticholinergic drug. The identification of Hermes' antidote, moly, with the snowdrop, whose galantamine reverses such poisoning (Plaitakis and Duvoisin, Clinical Neuropharmacology, 1983), is ingenious and speculative. The transformation into pigs is myth.

The Delphic oracle is the most argued case. Plutarch, himself a priest at Delphi, says the room where consultants sat was filled "neither often nor regularly" with a sweet smell and a breath, a pneuma (Plut. De def. or. 50), and tells of a Pythia who prophesied against bad omens, ran screaming from the shrine and died days later (51). Geologists found faults under the temple and light hydrocarbons in nearby springs and proposed ethylene as the intoxicant (de Boer and colleagues, Geology, 2001). A later gas survey found that the setting could not produce intoxicating levels of ethylene (Etiope and colleagues, Geology, 2006), and toxicologists called the hypothesis implausible (Foster and Lehoux, Clinical Toxicology, 2007). Trained trance under strong expectation explains the Pythia without any gas, and a small gas effect cannot be ruled out.

Where no substance could do what is claimed, the verdict is implausible, and often an ancient author says so first. Pliny reports that the Druids cut mistletoe with a golden sickle and believe it makes barren animals fertile and cures every poison, then adds "so great is the devotion of peoples, mostly in trivial matters" (Pliny NH 16.249–251). Mistletoe is more poison than cure. Rachel's mandrakes in Genesis (Gen 30.14) contain no compound known to aid conception, and the narrative itself credits each pregnancy to God.

2. Real medicine wrapped in ritual

The second finding is the most consistent in the whole survey. Where a healing rite worked, the words were seldom the part that worked. Inside the ritual there is usually a practical step, sometimes a small one, that modern medicine still uses or can explain. Homer, the oldest Greek poet in the library, gives the pattern. When the young Odysseus is gored by a boar, his uncles treat him.

ὠτειλὴν δʼ Ὀδυσῆος ἀμύμονος ἀντιθέοιο δῆσαν ἐπισταμένως, ἐπαοιδῇ δʼ αἷμα κελαινὸν ἔσχεθον

They bound up the wound of noble, godlike Odysseus with skill, and with a charm they held back the dark blood.

The skilled binding comes first. The epaoidē, a sung charm, follows it, and may have calmed the patient and given the healers authority. Pindar's Asclepius likewise pairs gentle chants with drinks and the knife. The medieval German blood charms are different, and the difference matters. The Bamberg charm tells how Christ, wounded in the side, "took his thumb and pressed it shut", and then commands the blood to stand still as the Jordan stood (Bamberger Blutsegen). The pressure appears only inside the story. The healer is told to say the words and three paternosters. Modern first-aid guidance makes firm direct pressure the first step for bleeding (American Heart Association and Red Cross guidelines, 2020 and 2024), and Britain's health service advises pinching a bleeding nose for ten to fifteen minutes. A charm takes a minute or two. Where the German charms "worked", the bleeding was the kind that stops by itself, and the successes were the ones remembered.

Other charms carry their remedy inside them. In the Bavarian "Three Good Brothers" charm Christ tells three brothers to take olive oil and sheep's wool and lay them on the wound, and forbids them to charge for it (Münchener Wundsegen). A covered, oiled wound heals faster than one left to form a dry scab, a principle established in 1962 (Winter, Nature). Cato's charm for a dislocation, the oldest healing charm in Latin, begins with the promise "Luxum siquod est, hac cantione sanum fiet", any dislocation will be made sound by this chant. Two men hold a split green reed while the nonsense words are sung, and the reed is then bound to the limb (Cato Agr. 160). For a fracture that is a splint, and splinting is still the core of treatment. For a dislocation nothing in the rite puts the joint back, so the cure the charm names most is the one it cannot deliver.

Sulphur ran through Greek purification and real hygiene alike. After killing the suitors Odysseus calls for "sulphur, remedy of evils" to fumigate his hall (Hom. Od. 22.481). Burning sulphur gives sulphur dioxide, which kills insects and microbes, and ships were still being gassed with it in the early twentieth century (Engelmann and Lynteris, Sulphuric Utopias, 2020). In the Greeks' terms this was katharmos, a cleansing of pollution, and it cleansed real dirt along with the religious kind. A Late Babylonian commentary copied for an exorcist, an āšipu (the specialist who treated illness as an attack by gods, demons or witches), glosses the ingredients of remedies for a scalp that "the lice gnaw": sulphur, naphtha and aromatic oils, with the promise that "the old hair falls out and the new hair comes out" (SpTU 1 51 r 2–8). Sulphur ointment is still a recognised, if slow, treatment for scabies mites, a close analogue for lice (Mila-Kierzenkowska and colleagues, 2017). Here the exorcist was also the dermatologist, and in Mesopotamian terms there was no contradiction.

Some practical steps worked by accident. A Hippocratic treatise on women's diseases tells the doctor to hold foul-smelling substances (burnt horn, sulphur, asphalt) under the nose of a woman whose womb has "risen" and is choking her (Hippoc. Nat. Mul. 26). The theory is false. The fumes work like smelling salts, which provoke a reflex breath and rouse people from fainting. Columella, quoting the Graeco-Egyptian writer Bolus of Mendes, says a pustular sheep should be buried alive under the fold's threshold and the flock driven over it (Columella 7.5.17). Removing the sick animal is the core of modern control of sheep pox, as in Spain's eradication of an outbreak in 2022–23 (Viruses, 2024), though one animal is never enough and the walk over the grave adds nothing.

The healing sanctuaries fit the same pattern on a larger scale. Diodorus says Isis stands over the sick in their sleep and gives them remedies, and "those who obey her are healed beyond expectation" (Diod. 1.25.5). The obedience clause explains every failure in advance. Pilgrims at Epidaurus or Oropos got rest, baths, diet, ordinary remedies and sometimes hands-on treatment, and the sanctuaries carved only their successes on stone. A large analysis of trials comparing placebo with no treatment found no significant effect on objective outcomes and small effects on pain and other reported symptoms (Hróbjartsson and Gøtzsche, New England Journal of Medicine, 2001). Relief of pain and distress is therefore explicable. A blind man cured overnight is not.

Much of the rest "worked" by riding the natural course of illness. A German charm promises that a lump called an überbein will vanish after wood is laid on it on three mornings (Contra überbein). If the lump was a ganglion, which is a modern guess, about half disappear untreated, but over months or years (Dias and Buch, Journal of Hand Surgery, 2003). Malarial fevers come and go on the parasite's cycle, every third day for the quartan form (Collins and Jeffery, Clinical Microbiology Reviews, 2007), so any amulet applied before a fever-free day seemed to work. The rest is implausible. When Elisha revives the Shunammite's son (2 Kgs 4.34), some medical historians see an early account of resuscitation, but by the text's own timeline the child had been dead for hours, and in its own terms the story shows the power of Israel's God answering a prophet, with no technique involved. One case is fraud by an ancient witness's account. Gregory of Tours says the wonder-worker Desiderius had servants violently stretch the limbs of paralytics he could not heal "by divine gift", and that many died (Greg. Tur. Hist. 9.6); Gregory was hostile and partly relying on hearsay.

3. The mind: placebo, nocebo and curses that work because the victim knows

The third group contains the most interesting "successes", because they are real effects produced by belief. Plato saw the mechanism clearly. In the Laws he separates poisoning by natural means from harm done by spells and binding curses, and he places the second kind's power in the minds of both parties.

ἄλλη δὲ ἣ μαγγανείαις τέ τισιν καὶ ἐπῳδαῖς καὶ καταδέσεσι λεγομέναις πείθει τοὺς μὲν τολμῶντας βλάπτειν αὐτούς, ὡς δύνανται τὸ τοιοῦτον, τοὺς δʼ ὡς παντὸς μᾶλλον ὑπὸ τούτων δυναμένων γοητεύειν βλάπτονται.

The other kind works by tricks, charms and what are called bindings. It persuades those who dare to do harm that they can do it, and persuades the victims that they really are being harmed by people with the power to bewitch.

Plato goes on to describe the terror caused by wax figures found at doorways and crossroads. Lead curse tablets (defixiones, "fixings") were a real and widespread practice from the fifth century BCE to the fifth century CE (Gager, Curse Tablets and Binding Spells, 1992); the oldest in the library, from Selinus, are dated by I.Sicily to the late sixth or early fifth century BCE (chapter IV). The Babylonian anti-witchcraft series Maqlû ("Burning") lists the witch's methods, taking hair and scooping up the dust of footprints (KAR 94 7'–11'), and answers them by burning her effigy. No physical link joins a figurine to a body. A curse can still harm through what doctors now call the nocebo effect, the harm done by negative expectation, but only once the victim learns of it. The physiologist Walter Cannon collected reports of people who sickened and died after learning they had been cursed, and one who recovered almost at once when the curse was lifted (Cannon, American Anthropologist, 1942); a commentary sixty years on described the stress pathways that could make this happen (Sternberg, American Journal of Public Health, 2002). Cannon himself warned that poisoning must be excluded first. Since most curse tablets were buried in secret, they could only "work" when someone found or rumoured them, and then the counter-rite could cure by the same route.

Tacitus reports exactly this situation in 19 CE, when Germanicus, the heir to the empire, fell ill at Antioch.

saevam vim morbi augebat persuasio veneni a Pisone accepti; et reperiebantur solo ac parietibus erutae humanorum corporum reliquiae, carmina et devotiones et nomen Germanici plumbeis tabulis insculptum

The belief that Piso had poisoned him made the savage force of the illness worse. And there were dug out of the floor and walls the remains of human bodies, spells and curses, and Germanicus' name cut into lead tablets.

Tacitus states the nocebo idea in his own words. What killed Germanicus, disease or poison, Tacitus leaves open and the panel cannot settle, so that verdict is insufficient evidence; the deposits themselves match excavated curse tablets and were probably real.

The placebo effect, the benefit of positive expectation, accounts for the matching cures. In Plato's Charmides Socrates offers a Thracian cure for headache, a leaf with a charm, and says "without the charm the leaf is of no use" (Pl. Chrm. 155e). Headache is among the complaints most responsive to expectation, and the more elaborate the ritual around an inert treatment, the larger the response: in migraine trials about 22% responded to placebo pills, 38% to sham acupuncture and 58% to sham surgery (Meissner and colleagues, JAMA Internal Medicine, 2013). Galen tested a wolf-dung amulet for colic on his own patients and "marvelled to see most of them clearly helped", insisting that what he hung on them was "a substance, not barbarous names, as some of the sorcerers do" (Gal. SMT 10.2.21). His trial had no control group, and colic pain comes and goes by itself. His instinct, though, was a scientist's.

The Norse sagas tell the same story in their own terms. In Egils saga a girl named Helga lies long ill and sleepless. Egill has her moved to clean bedding, finds a whalebone carved with runes under her bed, scrapes the runes into the fire, carves new ones, and she recovers. A rejected suitor had tried to carve love-runes and got them wrong, and Egill delivers the moral: "Skalat maðr rúnar rísta, nema ráða vel kunni", no one should carve runes who cannot read them well (Egils saga 72). Removing a feared object, airing the bedding and an authoritative counter-rite are what a nocebo illness would respond to. Impotence blamed on sorcery follows the pattern. Ovid wonders whether a Thessalian drug or a name pricked into red wax has bound him (Ov. Am. 3.7.27–36), a Babylonian commentary on potency rites blames a husband's failure on kišpū, sorcery (SpTU 2 39 r 4'–8'), and in Petronius' novel Encolpius responds to an old woman's thread and spittle only when her hands get to work, then fails again with the woman he wants (Petron. Sat. 131). Placebo responses in modern trials for erectile dysfunction are substantial (Stridh and colleagues, JAMA Network Open, 2020).

Belief could also sort the guilty from the innocent. The "water of bitterness" in Numbers, holy water with dust and washed-off curses that a suspected wife must drink (Num 5.23), is chemically inert. A guilty believer facing it has reason to confess beforehand, and an innocent one has nothing to fear. An economic study of medieval ordeals makes this screening argument for trial by fire and water (Leeson, Journal of Law and Economics, 2012); the panel applies it to Numbers by analogy. The officiants could also tilt the result, and medieval writers knew it. Gottfried's Isolde carries the hot iron after a deliberately ambiguous oath, and the poet remarks that Christ is "wintschaffen als ein ermel", as wind-blown as a sleeve (Gottfried, Tristan 15739–15752).

Fear and expectation can change perception itself. Gregory the Great says that when the monks of Monte Cassino threw a dug-up bronze idol into their kitchen the whole room seemed to burst into flame, but the fire was "in the brothers' eyes" and not in Benedict's (Greg. Dial. 2.10). Gregory's own account is of a collective illusion, which he blames on a demon; a modern review of mass sociogenic illness finds such outbreaks throughout history, often in religious communities (Bartholomew and Wessely, British Journal of Psychiatry, 2002). Pausanias, who consulted the oracle of Trophonius himself, describes a night plunge in a cold river, a descent into a dark pit, a sudden pull through a narrow hole, and a consultant carried out "possessed by terror" until "laughter returns to him" (Paus. 9.39.13). Even fifteen minutes of near-total sensory deprivation produces perceptual disturbances in ordinary volunteers (Mason and Brady, Journal of Nervous and Mental Disease, 2009). The Norwegian King's Mirror of the thirteenth century explains why some men go geilt, wild-mad: when two armies raise the war-cry, timid and young men lose their wits from terror and run to the woods (Konungs skuggsjá ch. 11). That is a sound description of combat breakdown (Jones and Wessely, Shell Shock to PTSD, 2005), though the same passage adds that after twenty years they grow feathers.

The ancients also noticed who was immune. Celsus quotes an Egyptian musician who said magic has power over the uneducated and the immoral and none over philosophers (Orig. Cels. 6.41). For effects that run through belief that is broadly right, and modern work links an analytic thinking style to lower paranormal belief (Pennycook and colleagues, Cognition, 2012). Where the claim needs the rite to act on someone who does not know of it, or to compel a distant person, every verdict is implausible: Simaetha's spinning wheel and melting wax in Theocritus cannot draw her lover to her door (Theoc. Id. 2.10–16), and no ray leaves an envious eye, whatever Plutarch's theory of fiery effluence (Plut. Quaest. conv. 5.7.2).

4. Knowledge presented as power: eclipses and weather

Prediction worked when it rested on observation, and it was most impressive when the person who predicted kept the method to herself. Plutarch tells the story twice. Advising a young bride to study, he recalls a Thessalian woman.

περὶ Ἀγλαονίκης ἀκηκουῖα τῆς Ἡγήτορος τοῦ Θεσσαλοῦ θυγατρὸς ὅτι τῶν ἐκλειπτικῶν ἔμπειρος οὖσα πανσελήνων καὶ προειδυῖα τὸν χρόνον, ἐν ᾧ συμβαίνει τὴν σελήνην ὑπὸ τῆς σκιᾶς ἁλίσκεσθαι, παρεκρούετο καὶ συνέπειθε τὰς γυναῖκας

You will have heard of Aglaonice, daughter of Hegetor of Thessaly. She was expert in the full moons that are eclipsed and knew in advance the time when the moon is caught by the shadow, and so she deceived the women and persuaded them [that she was drawing the moon down].

Aglaonice's date is unknown and the story may be a rationalising tale attached to a name, but the knowledge was attainable. Assyrian court scholars were making crude eclipse forecasts by the seventh century BCE (Steele, Archive for History of Exact Sciences, 2000), and the Antikythera mechanism, a Hellenistic geared calculator, predicted eclipses from Babylonian cycles (Freeth and colleagues, Nature, 2006). Thessalian witches who "drew down the moon" were a stock figure of Greek and Roman poetry, and Romans banged bronze to rescue the eclipsed moon. That noise always succeeded, because every eclipse ends. Pliny saw the irony. The method of foretelling eclipses to the hour was long known, he says, "yet the old belief lasts in a great part of the crowd, that this is forced by poisons and herbs" (Pliny NH 25.10). Whether the poets' moon-drawing originally meant an eclipse at all is disputed.

An omen that predicts nothing can still change history, because people act on it. On the night of 27 to 28 August 413 BCE there was a total eclipse of the moon over Syracuse, lasting about three-quarters of an hour in modern catalogues (Espenak and Meeus). Nicias, commanding the Athenian army, was "somewhat too given to divination", and refused even to discuss withdrawal until they had waited "thrice nine days, as the seers prescribed" (Thuc. 7.50.4). The delay helped destroy the army. Before the battle of Pydna in 168 BCE the Roman officer Sulpicius Gallus did the opposite and announced the eclipse to the troops in advance, because it happens "naturali ordine statis temporibus", in a natural order at fixed times (Livy 44.37.5). Curtius says Alexander's Egyptian seers knew the cause of an eclipse before Gaugamela and told the soldiers it foretold Persian defeat anyway (Curt. 4.10.5).

Babylonian omen science shows the same split within a single sentence. An omen has two halves, a protasis ("if Adad thunders out of his season") and an apodosis (what will follow). The protasis is often good observation. A commentary defines unseasonal thunder as thunder in the fourth to sixth months, high summer, when thunderstorms really are rare on the Mesopotamian plain (ACh Adad 25 o 1–3), though no climatological study was checked. The apodoses, plague, floods and the fate of kings, have no causal link at all. One rule that looks astronomical also fails. An Uruk commentary says that if Venus "enters the Moon" on the day the moon disappears, an eclipse will follow in 100 days (SpTU 4 162 o 15–17); by the panel's reckoning from standard eclipse geometry, which it did not check against a source, that interval would usually point away from the next eclipse season. Yet the scholars' habit of recording sky, weather and market prices side by side left series that economic historians now use (Slotsky, The Bourse of Babylon, 1997). The method failed and the record-keeping succeeded.

No ritual changes the weather. Herodotus says that after a three-day storm wrecked Xerxes' fleet the Persian Magi made sacrifices and sang to the wind, and then adds the decisive alternative.

τέλος δὲ ἔντομά τε ποιεῦντες καὶ καταείδοντες γόησι οἱ Μάγοι τῷ ἀνέμῳ, πρός τε τούτοισι καὶ τῇ Θέτι καὶ τῇσι Νηρηίσι θύοντες, ἔπαυσαν τετάρτῃ ἡμέρῃ, ἢ ἄλλως κως αὐτὸς ἐθέλων ἐκόπασε.

In the end the Magi, cutting victims and singing spells to the wind, and sacrificing besides to Thetis and the Nereids, stopped it on the fourth day, or else it somehow dropped of its own accord.

Storms end, and a rite performed late in one will be followed by calm. Pausanias says he has seen men turn back hail with sacrifices and charms (Paus. 2.34.3), and a Greek handbook of weather signs collected the cloud caps and haloes that warn of rain, so a practitioner could time a rite to the signs. The rain that saved Marcus Aurelius' army in about 172 CE was claimed for an Egyptian magus, for Christian prayer and for the emperor's virtue (Kovács, Marcus Aurelius' Rain Miracle, 2009).

One weather legend may hide a real public health measure. Empedocles was called the "wind-stopper". Diogenes Laertius, quoting a lost author, says that a plague struck Selinus "because of the stench from the river nearby", killing people and causing difficult births, until Empedocles brought two other rivers into it at his own expense and sweetened the water (D.L. 8.70). If the plague was malaria, which is a modern inference, flushing stagnant water works: a systematic review found environmental management of land and water cut malaria risk by about 88% (Keiser and colleagues, Lancet Infectious Diseases, 2005), and malaria in pregnancy causes fetal loss and stillbirth (Desai and colleagues, same journal, 2007). The sources are late and Empedocles attracted legend, so confidence is low.

Medical prediction could be valid too. A German medical handbook of the thirteenth century gives a test for the wounded: give burnet in water, and "if he is to recover he digests the herb; if he is to die you will find the herb in the wound" (Bartholomäus, wound prognosis). Drink can only appear in a wound if the gut is pierced, and the Hippocratic Aphorisms already say that a severed small bowel "does not grow together" (Hippoc. Aph. 6.24). Before abdominal surgery such wounds were usually fatal, though the panel could not confirm the historical mortality figures at source. The test is one-sided: a limb wound can never show the herb, so it always predicts recovery. Augustine, who believed demons foretell storms and plagues, explains their skill as experience, like a sailor's or a doctor's, and adds that "a physician who foresees what someone ignorant of his art does not foresee is not on that account to be thought divine" (Aug. Div. daem. 5.9).

Knowledge could also rewrite a divine decree. Esarhaddon of Assyria says Marduk had written seventy years as the term of Babylon's desolation, then relented, "turned it upside down" and ordered resettlement after eleven (RINAP 4 104 ii 2–9). In cuneiform numerals 70 is a sign for 60 followed by a sign for 10; reversed, the same two signs read 10 and 1. The scholars' command of their script let the king rebuild ahead of schedule with the gods' blessing.

5. Chemistry and fire

Fire is where magic and craft meet most openly, and the rule here is simple: the more a text describes a material procedure, the more likely the effect is real. An Assyrian copy of a Babylonian glassmaking text begins with ritual. The kiln must be founded on a propitious day, the kūbu deities (spirits associated with stillborn infants) installed, and no outsider or impure person admitted (K 2520+ o 1–5). Then comes the recipe.

šum-ma {na₄}ZA.GIN₃.DUR₅ a-na e-pe-ši-ka 10 MA.NA {[na₄]}⸢IM⸣.MA.NA / 15 MA.NA NE {u₂}NAGA 1 2/3 MA.NA U₂ BABBAR a-he-e ta-[mar-raq] 1{niš} HI.HI ... ⸢IZI⸣ ṭa-ab-ta la qa-tir-ta ta-šar-rap

If you want to make blue glass: ten minas of sand, fifteen minas of the ash of the naga plant, one and two-thirds minas of "white plant"; you grind them separately and mix them together ... you burn a good, smokeless fire.

Sand is silica and plant ash is soda, the two ingredients of ancient glass, and a smokeless fire avoids the reducing conditions that spoil colour. The standard edition includes a chemist's reading of these recipes as workable instructions (Oppenheim, Brill and colleagues, Glass and Glassmaking in Ancient Mesopotamia, 1970). The rites had no chemical effect, but the ban on strangers worked as trade secrecy and the purity rules kept the workshop clean. In Babylonian terms the craft needed divine favour as much as skill.

Petroleum produced the most dramatic fire wonders. Plutarch describes how the people of Babylon laid a trail of naphtha along a street for Alexander and lit one end, and the whole street blazed at once; he adds that some, "rescuing the myth for the truth", say this was the drug with which Medea anointed the crown and robe of tragedy (Plut. Alex. 35.5). Light petroleum gives off vapour that ignites from a nearby flame (NOAA chemical datasheets), so the demonstration is real. Euripides' crown, which bursts into flame on the princess's head with no fire near (Eur. Med. 1186–1201), cannot happen, and Plutarch's sources themselves change the story so that the flame comes from outside. Adrienne Mayor's argument that such myths reflect real incendiary chemistry (Archaeology, 1997) is a modern hypothesis. Second Maccabees tells how a "thick water" called Nephthar, hidden in a pit, blazed up on the altar of the restored Temple (2 Macc 1.22). Reading Nephthar as naphtha is a modern identification, widely accepted; the story reads like the founding legend of an oil seep made sacred.

Quicklime gave the Roman Bacchic cult its most famous wonder, as a former initiate described it to the consul in 186 BCE.

matronas Baccharum habitu crinibus sparsis cum ardentibus facibus decurrere ad Tiberim, demissasque in aquam faces, quia vivum sulpur cum calce insit, integra flamma efferre.

Matrons dressed as Bacchants, their hair loose, run down to the Tiber with blazing torches, plunge the torches into the water and bring them out with the flame unharmed, because there is live sulphur mixed with lime in them.

Quicklime meeting water can release enough heat to ignite nearby fuel (manufacturer's safety data sheet), and sulphur keeps burning. The witness gives the mechanism and does not say anyone was fooled by it; her next sentence, about men "carried off by the gods" who were in fact tied to machines and dragged into caves, describes a deliberate trick. The Refutation of All Heresies, an early third-century Christian work usually attributed to Hippolytus, exposes the same chemistry used for deception: pitch made to look as if it boils by stirring in vinegar and natron, which give off carbon dioxide while the pitch stays cool, and a pyre lit by a "libation" poured on quicklime hidden under the ash ([Hippol.] Ref. 4.33.2). The same author explains how a sealed question written in vitriol was made legible by fuming it with oak-gall (Ref. 4.28.4), which is the reaction that makes iron-gall ink, the standard ink of medieval Europe (Kolar and Strlič, Iron Gall Inks, 2006).

Some fire wonders were misperceived by outsiders. Pausanias watched a Persian magus in Lydia chant from a book over an altar of ash until the wood "must catch without fire" (Paus. 5.27.6). Strabo describes the Cappadocian magi guarding an undying fire under a heap of ash (Strabo 15.3.15). A fire banked under ash explains the sight with no fraud on anyone's part. Firewalking, finally, is physically real. Iamblichus took it as the greatest proof of divine possession that "many, when fire is brought to them, are not burned" (Iambl. Myst. 3.4), and Strabo says the priestesses at Kastabala walked barefoot over embers (Strabo 12.2.7). Embers hold little heat and conduct it poorly, so a brisk crossing usually leaves no burn (Leikind and McCarthy, Skeptical Inquirer, 1985). The god is not needed; the physics is enough.

6. Devices and fraud, briefly

The tricks are the subject of the next chapter, so a few cases will do here. Hero of Alexandria, in the first century CE, published the plans for "a little temple whose doors open by themselves when a sacrifice is made, and close again when it is put out" (Hero, Pneum. 1.38). The altar fire heats air in a hidden vessel, which pushes water into a bucket whose weight turns the doors (Woodcroft's translation, 1851; Tybjerg, Studies in History and Philosophy of Science, 2003). The physics is correct, and the author is an engineer, not a priest. The exposers name their tricks as precisely: Daniel scatters ash on the temple floor and shows the king the priests' footprints (Bel 19); Hippolytus describes a crane's windpipe or ten nested bronze tubes for whispering a god's answer into a child medium's ear ([Hippol.] Ref. 4.28.9); Lucian shows Alexander of Abonuteichos with a linen serpent's head whose mouth opened on horsehairs (Lucian, Alex. 12). Rufinus says a tiny window in the Serapeum at Alexandria was aligned so that a sunbeam struck the god's lips on the festival day, and that an iron image of the Sun rose toward a lodestone in the ceiling and then fell back (Rufin. HE 11.23). The falling matters. Augustine's story of an iron statue held still in mid-air between magnets (Aug. Civ. 21.6) is ruled out by Earnshaw's theorem of 1842, which forbids stable suspension under fixed magnets; the later extension of the theorem to iron bodies (Braunbek 1939) is standardly cited but was not read by the panel.

Not every marvel was staged. The broken colossus of Memnon at Thebes really did sound at dawn; Juvenal mocks its "magic strings" (Juv. 15.5), Strabo notes that the upper half fell in an earthquake (Strabo 17.1.46), and Roman visitors carved their testimony on its legs (Rosenmeyer, The Language of Ruins, 2018; Bowersock, 1985). The usual modern explanation, heating or drying of the cracked stone in the morning, is a standard account that the panel could not check at source. Some feats were real skills misread. Theophrastus explains how drug-sellers ate hellebore in public by habituation, and how a shepherd who ate a whole bundle daily shamed one of them, for "the same thing is a drug to some and not a drug to others" (Theophr. HP 9.17.2). Celsus says the street performers who put a finger in a snake's mouth have first stupefied it with drugs (Celsus Med. 5.27.3).

7. Impossible claims, and what may lie behind them

The largest single verdict is implausible: no one turns into a wolf, raises the dead, flies or kills with a glance. But very few of these claims are pure invention. Most describe a real experience or a real object that was misread, and the same misreadings recur in cultures with no contact, which suggests shared human experience behind them.

Sleep paralysis is the clearest. Ynglinga saga, Snorri's history of the legendary Swedish kings, tells how King Vanlandi's abandoned Finnish wife hired a sorceress to work seiðr, the Norse magic of chanted trance, against him.

Drifa keypti a Hulþ ſeið cono at hon ſcylde ſiða vanlanda til fiɴlandz. eða deyþa hann at ꜹðrom coſte. ... callaði hann oc ſagðe at mara traþ hann.

Drífa paid Huld the seiðr-woman to draw Vanlandi back to Finland by sorcery, or else to kill him ... he cried out and said that a mara was trampling him.

The mara, the crushing night-being that survives in the English word "nightmare", matches the experience of waking unable to move, with pressure on the chest and a sensed hostile presence. The folklorist David Hufford met the experience in Newfoundland, where it had a name, then found it among Americans who had never heard of such a tradition, and estimated that about 15% of people have it (The Terror That Comes in the Night, 1982); a pooled figure of about 7.6% from a later review was not checked at source. Sleep paralysis itself is harmless. A real death in sleep, from a stroke or a heart attack, would then be explained after the fact by the wife's hired magic.

Werewolves were doubted by their best reporters. Herodotus says the Scythians swear that the Neuri become wolves for a few days each year, and that he does not believe it (Hdt. 4.105). Pliny is blunter.

homines in lupos verti rursusque restitui sibi falsum esse confidenter existimare debemus aut credere omnia quae fabulosa tot saeculis conperimus.

We must confidently hold it false that men are turned into wolves and restored to themselves, or else believe everything that so many centuries have taught us is fable.

Yet a soldier in Petronius' novel still "pissed in a circle round his clothes and suddenly turned into a wolf" (Petron. Sat. 62). Modern psychiatry knows the delusion of having become an animal: twelve hospital cases in one study (Keck and colleagues, Psychological Medicine, 1988) and forty-three published cases of wolf or dog delusions in a review (Frontiers in Psychiatry, 2021), mostly with severe psychosis. Whether any ancient werewolf story began with such a patient is a hypothesis; the stories themselves are folklore.

Epilepsy runs through the whole library. The Hippocratic author of On the Sacred Disease, writing about 420–400 BCE, says those who first called it sacred were people like "the magoi, purifiers, begging priests and charlatans of our own day" (Hippoc. Morb. Sacr. 1), and that all diseases are "divine and human" alike (18). The boy in Mark's gospel who is thrown down, foams, grinds his teeth and goes rigid (Mark 9.18) is a clinical picture any neurologist would recognise. When Apuleius was tried for magic in North Africa around 158 CE, the prosecution claimed he had made a slave boy fall by a charm. He answered with a diagnosis.

ceterum Thallus, quem nominastis, medico potius quam mago indiget; est enim miser morbo comitiali ita confectus, ut ter an quater die saepe numero sine ullis cantaminibus corruat

As for Thallus, whom you named, he needs a doctor more than a magus. The poor boy is so worn down by epilepsy that he often falls three or four times a day without any incantations at all.

Epilepsy also explains why charms against it seemed to work. A German charm said over a fallen patient ends with the Latin instruction "do this three times, and soon you will see the sick man rise up healed" (Epilepsy charm 'Donerdutigo'). He would rise anyway. In video recordings of 579 seizures, generalised convulsions almost never lasted beyond eleven minutes (Jenssen and colleagues, Epilepsia, 2006). Galen's honest test of a peony-root amulet on one child, fits stopping while it was worn and returning when it fell off (Gal. SMT 6.3.10), could not exclude coincidence in an illness whose frequency fluctuates; placebo responses in modern epilepsy trials come largely from such fluctuation and from regression to the mean (Goldenholz and Goldenholz, Epilepsy Research, 2016).

Apparent death lies behind some resurrections. Philostratus says that when Apollonius of Tyana revived a dead bride at Rome, it was unclear whether he found "a spark of life" her attendants had missed, since her face was steaming in the drizzle (Philostr. VA 4.45); the history of premature burial shows such mistakes were real if rare (Bondeson, Buried Alive, 2001). Misread nature lies behind others. Virgil's recipe for bees born from a bullock's carcass (Verg. G. 4.295–314) describes the drone fly, a convincing bee mimic that breeds in rotting matter, as an entomologist proposed in 1893 (Osten-Sacken). Pliny reports from Cappadocia that when menstruating women walk naked round a field the caterpillars and beetles "drop off" (Pliny NH 28.77); many insects drop from plants when a large animal comes near, and aphids have been shown to flee the breath of a grazing mammal (Gish and colleagues, Current Biology, 2010), though the panel confirmed only that paper's title and gives the explanation low confidence. In the same passage Pliny says the same exposure drives off hail, which is impossible. The Roman strix, a night bird that scratched infants and drank their blood (Ov. Fast. 6.131–168), gave a cause to infant wasting and sudden infant death, and the screech owl shared its name. The Babylonian omen series on malformed births records real congenital conditions with great care (SpTU 2 38, 21–22; Leichty's edition, 1970); only their meaning as signs of kingship is unfounded. Even a cult can rest on a misreading: Justin Martyr's statue on the Tiber island inscribed to "Simon the holy god" (Just. 1 Apol. 26) is usually explained by the altar found there in 1574, dedicated to the Sabine god Semo Sancus (CIL VI 567).

Sometimes the accused supplied the natural explanation himself. Pliny tells of a freedman, Furius Cresimus, prosecuted for luring his neighbours' crops into his small field by sorcery. He brought his well-fed slaves, his iron tools and his oxen into court and said "Veneficia mea, Quirites, haec sunt", these, citizens, are my sorceries (Pliny NH 18.41–43). And where the outcome was uncertain, magic and its rivals could both claim success. Gregory of Tours says his household secretly called a diviner to a boy with bubonic plague, who "muttered incantations, cast lots, hung ligatures round his neck", and the boy died, while another boy given dust from a saint's tomb lived (Greg. Tur. Virt. Iul. 46a). Untreated bubonic plague kills 30 to 60% of its victims (World Health Organization fact sheet), so survivors appear under any treatment, and each side can count its own.

The verdicts at a glance

Why magic persisted when it did not work

If most of it did not work, why did it last for nearly four thousand years? The ancient texts answer that question themselves, often better than they answer any other. The first reason is that people remember hits and forget misses. Cicero tells how Diagoras, the notorious atheist of Melos, was shown the painted votive tablets in a temple of the sea gods.

nonne animadvertis ex tot tabulis pictis, quam multi votis vim tempestatis effugerint in portumque salvi pervenerint, ita fit inquit, illi enim nusquam picti sunt qui naufragia fecerunt in marique perierunt.

"Don't you see from all these painted tablets how many people escaped the force of the storm by their vows and came safe into harbour?" "Yes," he said, "because the ones who were wrecked and drowned at sea are painted nowhere."

That is survivorship bias, stated more than two thousand years before statisticians named it. Cicero applies the same argument to dreams, asking who, "shooting all day long, does not sometimes hit the mark?" (Cic. Div. 2.121), and to horoscopes, recalling how the Chaldaean astrologers promised Pompey, Crassus and Caesar that each would die old, at home and in glory (2.99); all three died violently. Eusebius compares the rare fulfilled oracle to two people who happen to draw the same lot out of ten thousand (Eus. PE 4.2.2). Thucydides watched memory rewrite an oracle as it happened: during the plague Athenians argued whether an old verse had foretold loimos, plague, or limos, famine, and "for the moment, naturally, plague won" (Thuc. 2.54.3). Aristotle noted that when seers speak ambiguously their hearers nod along (Arist. Rh. 3.5). A double-blind test in 1985 found that astrologers matched birth charts to personality profiles no better than chance (Carlson, Nature); later reanalyses dispute the design, but the result stands as the mainstream view.

The second reason is that the professional always had an escape clause. Philostratus, defending Apollonius against the charge of sorcery, describes how clients of magicians reason.

ἢν μὲν γὰρ παθόντων τι τῶν παιδικῶν πρὸς τοὺς ἐρῶντας ἢ δώροις ὑπαχθέντων προβαίνῃ τὰ ἐρωτικά, ὑμνεῖται ἡ τέχνη ὡς ἱκανὴ πάντα, εἰ δʼ ἀποτυγχάνοι ἡ πεῖρα, ἐς τὸ ἐλλειφθὲν ἡ ἀναφορά

If the beloved warms to the lover, or is won over by gifts, and the affair goes forward, the art is praised as able to do anything; if the attempt fails, the blame goes to something that was left out.

A missed ingredient, a broken taboo, an impure participant or a disobedient patient absorbed every failure. Herodotus' Egyptian king could be cured only by the urine of a woman who had known no man but her husband (Hdt. 2.111), so each failure convicted a wife instead of the remedy. Isis healed those who obeyed her. The German wound charm forbade payment, which protected the charm's reputation more than the patient.

The third reason is that the body and the weather usually recover on their own. Tertullian, explaining how pagan gods seemed to heal, says the demons "first injure, then prescribe remedies ... after which they stop injuring, and are believed to have cured" (Tert. Apol. 22.11). Remove the demons and that is an exact description of regression to the mean, the tendency of an illness at its worst to improve whatever is done, which still misleads doctors and patients (Morton and Torgerson, BMJ, 2003). Fits end, fevers break, storms blow out and every eclipse passes. A rite performed at the worst moment is followed by improvement almost every time.

The fourth reason is that magic did real social work, even where its physical claims failed. Ordeals sorted the guilty from the innocent among people who believed in them. A diviner who named thieves, like the slave woman in Gregory of Tours who "at once declared where the thief had gone" (Greg. Tur. Hist. 7.44), could draw on local gossip and put pressure on a suspect to return the goods. Assyrian oath curses came true because Assyrian armies enforced them; Esarhaddon reports that the oath's punishment fell on a rebel governor who fled to Elam and was killed there (RINAP 4 1 ii 55–57). The glassmakers' taboos kept their secrets. Rites at a sickbed gave frightened families something to do, and a curse tablet gave a losing litigant or a jilted lover a way to act on anger. These functions do not depend on the rite working physically, which is why disproof never ended them.

The ancient world also knew how to test magic, and sometimes did. Pliny argued that if the Magi's herbs could dry up rivers or rout armies, Rome would have used them, and that Nero, who had every resource and the strongest desire to command the gods, abandoned the art: "a vast and indubitable proof that the art is false" (Pliny NH 30.14–18). Pliny himself allowed it "certain shadows of truth", and placed them in the poisoner's art. That is close to this chapter's conclusion. Where ancient magic worked, it worked through drugs and poisons, through the dressings, splints and sulphur that rode inside the rite, through knowledge of the sky kept by a few, through skilled hands and clever machines, and above all through the minds of the people who believed in it. No passage in this library shows a word, a gesture or an image acting on matter, on the weather or on the future. The people who wrote these texts were neither fools nor magicians. Many of them noticed what worked, many of them doubted what did not, and the best of them left us the arguments we still use to tell the two apart.

Chapter XIII

The tricks, and the ancients who exposed them

From the Hippocratic author to Shakespeare, writers in this library took wonders apart, naming the hidden door, the speaking-tube, the drugged snake and the vague oracle, while most of them went on believing in powers of their own.

καὶ ἐγέλασεν Δανιὴλ, καὶ ἐκράτησεν τὸν βασιλέα τοῦ μὴ εἰσελθεῖν αὐτὸν ἔσω, καὶ εἶπ εν Ἴδε δὴ τὸ ἔδαφος καὶ γνῶθι τίνος τὰ ἴχνη ταῦτα.

Daniel laughed, held the king back from going in, and said, "Look at the floor, and know whose footprints these are."

The god Bel of Babylon ate twelve measures of flour, forty sheep and six jars of wine every day, or so his priests said. In the Greek addition to the Book of Daniel, Daniel asks the king to have ash sifted over the temple floor with no one but the king watching, and the doors are sealed. In the morning the seals are unbroken and the food is gone, and the ash shows the bare footprints of the priests, their wives and their children, who came in through a secret door under the table (Bel 1–22). Daniel then kills the temple's living serpent-god with cakes of pitch, fat and hair, using no prayer or spell at all (Bel 23–27). The story is a Hellenistic Jewish polemic and records no real Babylonian inquiry, yet its method is forensic, and it is the oldest complete exposure of a staged miracle in the library.

This chapter collects such exposures. The survey records 82 passages that expose trickery and 70 that mock magic, and many more give a natural cause for what others called a wonder. The panel that judged whether ancient magic could have worked (chapter XII) noticed that in roughly half the claims of staging and fraud, a source in this library names the mechanism itself. The ancients were their own first debunkers. Three cautions apply from the start. Almost every exposure is aimed at someone else: physicians at itinerant healers, Jews at Babylonians, Lucian at a rival cult, Christians at pagan temples and at other Christians. Many of the episodes come from comedy, satire, romance or saga, so "fraud" there describes how a story works and is no record of a real deception. And the debunkers rarely doubted everything.

Physicians and historians: classical Greece

The first sustained attack on magic in Greek comes from a medical treatise of the late fifth century BCE, On the Sacred Disease, about epilepsy. Its author, whose name is lost, says the disease is no more divine than any other, and he knows exactly who made it sacred.

ἐμοὶ δὲ δοκέουσιν οἱ πρῶτοι τοῦτο τὸ νόσημα ἀφιερώσαντες τοιοῦτοι εἶναι ἄνθρωποι οἷοι καὶ νῦν εἰσι μάγοι τε καὶ καθάρται καὶ ἀγύρται καὶ ἀλαζόνες, ὁκόσοι δὴ προσποιέονται σφόδρα θεοσεβέες εἶναι καὶ πλέον τι εἰδέναι.

I think the men who first made this disease sacred were like the magi, purifiers, begging priests and charlatans of our own day, who pretend to be very pious and to know more than other people.

The four words in his list name four kinds of freelance religious expert: the magos (originally a Persian priest, by now a dealer in private rites), the kathartēs or purifier, the agyrtēs or begging priest, and the alazōn, the boaster. The author's charge is structural. Their purifications, incantations and food taboos cover ignorance, and the arrangement protects them whatever happens, since a recovery is credited to them and a death to the gods. He even records their diagnostic system. A patient who bleats like a goat has been struck by the Mother of the Gods, one who neighs by Poseidon, and night terrors are "attacks of Hecate and assaults of heroes" (Hippoc. Morb. Sacr. 4). He thus preserves the richest description of Greek exorcistic healing in the corpus in the course of trying to abolish it.

Herodotus, writing at about the same time, works by aside. The Getae of Thrace worship Salmoxis, but the Greeks of the Black Sea told him that Salmoxis was a former slave of Pythagoras who built an underground chamber, vanished into it for three years and reappeared to prove his teaching on immortality (Hdt. 4.94–96). Herodotus says he neither disbelieves the story nor believes it too much, and then remarks that Salmoxis probably lived long before Pythagoras, which quietly ruins it. He records that the oracle-collector Onomacritus was "caught in the act" inserting a verse of his own into the oracles of Musaeus (Hdt. 7.6.3). When the Persian magi chant and sacrifice to a storm that has wrecked Xerxes' fleet, it stops on the fourth day, "or else it somehow dropped of its own accord" (Hdt. 7.191.2). Of the Neuri, who turn into wolves for a few days each year, he writes simply that those who say so do not persuade him (Hdt. 4.105).

Comedy went further. In Aristophanes' Wealth (388 BCE) the slave Carion spends a night in the healing sanctuary of Asclepius, where the sick slept in the dark waiting for the god. He looks up and sees the priest sweep the cakes and figs off the holy table, check every altar for leftovers, and "consecrate" the lot into a sack (Ar. Plut. 676–681). The joke is about priestly greed, and priests were in fact entitled to offerings, so the scene proves no staged epiphany. It shows that an Athenian audience could laugh at the idea that the god's night visits had human hands.

The philosophers supplied the theory. Plato says that shadow-painting, which fools the eye with perspective and shading, "falls nothing short of sorcery", and he sets beside it the stage conjuror's craft, thaumatopoiia, "wonder-making" (Pl. Resp. 602d). In the Republic his brother Adeimantus describes the begging priests who sell rich men cheap curses and claim to persuade the gods to serve them (Pl. Resp. 364b–c). Aristotle explains how prophecy protects itself. Seers speak ambiguously and their hearers nod along, and oracle-mongers never say when, because a guess at odd-or-even is likelier to land than a guess at an exact number (Arist. Rh. 3.5). His successor Theophrastus tells of a market drug-seller who amazed crowds by eating one or two roots of hellebore, a violent purge. A shepherd discredited him by eating a whole bundle and saying he did so every day. Another performer, Theophrastus notes, had secretly taken an antidote (Theophr. HP 9.17). The panel accepts this as a real effect with an ordinary cause, namely tolerance, a hidden countermeasure and showmanship.

Arguing from the record: Cicero and Pliny

Rome's two great sceptical texts on magic and divination argue less from apparatus than from evidence. Cicero wrote the second book of On Divination in 44 BCE, in the months after Caesar's murder, and he used the murder.

quam multa ego Pompeio, quam multa Crasso, quam multa huic ipsi Caesari, a Chaldaeis dicta memini, neminem eorum nisi senectute, nisi domi, nisi cum claritate esse moriturum!

How many things I remember the Chaldaeans telling Pompey, and Crassus, and Caesar himself: that none of them would die except in old age, at home, and in glory!

"Chaldaeans" was the Roman name for astrologers, after the Babylonian scholars who were supposed to have invented the art. All three men died violently. Cicero's other arguments read like early statistics. Were all the men who fell at Cannae born under one star (Cic. Div. 2.97)? Who shoots all day without sometimes hitting the mark, and why should we marvel that among a lifetime of nightly dreams a few come true (Cic. Div. 2.121)? The famous lots of Fortuna at Praeneste are no better than dice, and "the whole business was invented by deceit, for profit, for superstition or through error" (Cic. Div. 2.85).

A century later Pliny the Elder opened the thirtieth book of his Natural History with a verdict and a theory.

fraudulentissima artium plurimum in toto terrarum orbe plurimisque saeculis valuit. [...] natam primum e medicina nemo dubitabit ac specie salutari inrepsisse velut altiorem sanctioremque medicinam

The most fraudulent of the arts has had the greatest power over the whole world for the most ages. [...] No one will doubt that it was born first from medicine and crept in under the look of healing, as a loftier and holier medicine.

Magic, for Pliny, had absorbed the three arts that most grip the human mind: medicine, religion and astrology (NH 30.1–2). His proof that it was empty was Nero. No one had more wealth, power or desire to command the gods, Nero was initiated by the Armenian magus-king Tiridates, and Nero gave the art up. Pliny calls this "a vast and indubitable proof that the art is false" and concludes that whatever works in it belongs to poisoning, not magic (NH 30.14–18). In the same passage he recalls the scholar Apion, who claimed to have called up Homer's shade to ask about his birthplace and "did not dare to say what answer he got". The panel calls Nero's experience a fair natural experiment by ancient standards.

Pliny reasons the same way elsewhere. If herbs could dry up rivers or open locks, Scipio would have opened the gates of Carthage with a herb instead of battering them for years (NH 26.18–20). If a salamander could put out fires, fire-prone Rome would have tried it by now (NH 29.74–76). The bloodstone really works as a mirror for watching eclipses, and the magi's claim that it makes its wearer invisible is "the clearest example of their shamelessness" (NH 37.165). Of werewolves he writes that no lie is so shameless that it lacks a witness (NH 8.80–84). He also records the cost of belief. A Roman knight was put to death by Claudius for carrying a Druids' "snake's egg" into a lawsuit, and Pliny notes the magi's "cunning at hiding their frauds" in prescribing that it be gathered at a certain moon (NH 29.52–54).

Roman historians added chemistry and stagecraft. In Livy's account of the Bacchanalia scandal of 186 BCE, the informer Hispala tells the consul that matrons plunge burning torches into the Tiber and lift them out still alight "because live sulphur mixed with lime is in them", and that men said to be "carried off by the gods" are really tied to a machine and whisked into hidden caves (Livy 39.13.12–13). Livy reports a frightened informer's testimony in a moral panic, so the details may be inflated, but quicklime does heat when wet. Quintus Curtius shows the reverse case. Before Gaugamela, Alexander's Egyptian seers knew why the moon was eclipsed and deliberately told the troops a reassuring omen instead (Curt. 4.10.5). Plutarch tells a young bride to laugh at women who claim to draw down the moon, since the Thessalian Aglaonice simply knew when eclipses were due (Plut. Conj. praec. 48). The physician Celsus adds that the street performers who put a finger in a snake's mouth have first stupefied the snake with drugs (Celsus Med. 5.27).

The wonder-worker's workshop: the second and third centuries

The richest exposures in the library come from the Greek-speaking Roman empire of the second and third centuries CE, when oracles, healers and itinerant holy men competed for clients. Lucian of Samosata's Alexander, or the False Prophet, written after 180 CE, is a hostile biography of Alexander of Abonuteichos, who founded a successful oracle of a serpent-god called Glykon on the Black Sea coast. Lucian explains every effect. Alexander faked prophetic frenzy by chewing soapwort root to make his mouth foam (Lucian, Alex. 12). He planted a goose egg holding a newborn snake, its lid sealed "with white wax and white lead", in the flooded foundations of the new temple and dug it up in public (Alex. 14). A few days later the "grown" god was a large tame snake bought in Pella, shown in a dim room with its real head under Alexander's arm and a linen human face in its place, while the crowd was hurried in at one door and out at the other. For oracles that spoke aloud, Alexander joined cranes' windpipes through the artificial head and had an accomplice speak into them from outside (Alex. 26). Questions came to him sealed, and he read them anyway.

βελόνην πυρώσας τὸ ὑπὸ τὴν σφραγῖδα μέρος τοῦ κηροῦ διατήκων ἐξῄρει καὶ μετὰ τὴν ἀνάγνωσιν τῇ βελόνῃ αὖθις ἐπιχλιάνας τὸν κηρόν

Heating a needle, he melted and lifted out the wax beneath the seal, and after reading the question he warmed the wax again with the needle.

Lucian gives two more methods, a paste for casting a copy of the seal and quicklime in bookbinder's glue, and says he records them so that his reader can refute such things (Alex. 19–21). The work is addressed to a friend named Celsus. Whether he is the same Celsus who wrote against the Christians is disputed. The panel judged every one of these techniques physically workable and filed the Glykon oracle under trick and fraud.

Lucian's Philopseudes, "The Lover of Lies", turns the same scepticism on the believers. At a gouty friend's sickbed, distinguished philosophers trade cures: a shrew's tooth tied on in a lion skin, a Chaldaean's incantation that drew the venom out of a snake-bitten slave. The sceptic Tychiades asks how something tied to the outside of the body can act on a disease inside it (Lucian, Philops. 7–8). When a former sceptic describes a Hyperborean magus who raised the dead and sent a clay Cupid to fetch a married woman, Tychiades replies that the woman could have been had for twenty drachmas (Philops. 13–15). The story of the Egyptian priest Pancrates, whose enchanted pestle carries water and cannot be stopped, is the ancestor of the Sorcerer's Apprentice (Philops. 34–36). The dialogue is satire, and its point is that learned men will swallow anything.

The pagan philosopher Celsus, quoted by the Christian Origen, used the same market-place figures to belittle Jesus' miracles.

πρὸς τὰ ὑπὸ τῶν μαθόντων ἀπὸ Αἰγυπτίων ἐπιτελούμενα, ἐν μέσαις ἀγοραῖς ὀλίγων ὀβολῶν ἀποδομένων τὰ σεμνὰ μαθήματα καὶ δαίμονας ἀπὸ ἀνθρώπων ἐξελαυνόντων καὶ νόσους ἀποφυσώντων

...compared with the feats of those taught by the Egyptians, who sell their holy lore for a few obols in the middle of the market-place, drive demons out of people and blow diseases away...

The same performers, Celsus goes on, call up the souls of heroes, display banquets "that do not exist" and make lifeless things move, though only to the imagination (Orig. Cels. 1.68). Origen remarks that Celsus here almost admits that magic exists.

Physicians tested. Galen was shown an incantation that killed scorpions, in which the reciter spat on the scorpion each of the three times he spoke it. Galen then saw a scorpion die from the saliva alone, "without the incantation" (Gal. SMT 10.2.16). Leaving out the words is a genuine control, and it removes them as the cause. The panel declined to accept that saliva killed the scorpion, since the result fits Galen's own theory of sharp, hungry saliva a little too neatly, and marked the claim "insufficient evidence". The sceptic Sextus Empiricus mentions in passing that sorcerers smear lamp-wicks with verdigris or cuttlefish ink to make the people in a room look copper-coloured or black (Sext. Emp. PH 1.46).

Hippolytus' handbook of fakes

The fullest catalogue of conjuring in antiquity sits, unexpectedly, in a Christian work against heresy. The fourth book of the Refutation of All Heresies, attributed to Hippolytus of Rome (after 222 CE; the authorship is disputed), describes magicians' methods one by one. A client writes his question, which is supposedly burned so that the smoke carries it to the demons. The magician switches papers, and if the writing was made invisible he brings it back with copper vitriol ([Hippol.] Ref. 4.28). A boy medium lies face down in the dark while a tube made from the windpipe of a crane, a stork or a swan, or from ten nested bronze pipes, carries a hidden voice into his ear (Ref. 4.28.7–10). A lamb "cuts its own throat" because its neck has been smeared with an itching drug and a sword left beside it (Ref. 4.30). Pitch that seems to boil around the magician's hands has been stirred into bubbles by vinegar and natron over a gentle heat (Ref. 4.33). For divination in a bowl of water, gods appear through its glass bottom.

ὕπεστι δὲ οἶκος λεληθώς, εἰς ὂν συμπορευόμενοι οἱ συμπαικται σχήματα ὧν ἂν βούληται δεινύναι ὁ μάγος θεῶν καὶ δαιμόνων ἐνδυσάμενοι ἐμφαίνουσιν

Underneath is a hidden room. The accomplices go into it together, dress up as whichever gods and demons the magician wishes to show, and make them appear.

Hecate flying through the air as fire is a kite wrapped in burning tow, which sometimes sets the house alight (Ref. 4.35–36). The moon drawn down into the room is a mirror in the ceiling, a basin of water on the floor and a dim lamp between them, and the stars are fish scales glued overhead (Ref. 4.37–39). The talking skull is made of ox caul, wax and gypsum, with a crane's throat attached so that "the accomplice utters whatever he wishes", and it "vanishes" when the wax melts in the brazier (Ref. 4.41).

Hippolytus says he publishes these secrets to protect young men, who will be on their guard once they know (Ref. 4.34, 4.42). He also has a target. The heresiarchs, he says, copied these arts. The Gnostic teacher Marcus made the communion cup turn red during a long prayer by slipping a colouring drug into it (Ref. 6.39–40). Simon Magus is likened to the Libyan Apsethus, who taught a cage of parrots to say "Apsethus is a god" and set them free (Ref. 6.8). The panel accepted the apparatus as acoustically and chemically workable and noted that no standard modern study of these chapters could be checked.

Why people believed

Several writers of the period asked why the tricks succeeded. Philostratus, in his life of the holy man Apollonius, observes that when a lover wins the beloved, the sorcerer's art is hymned as all-powerful, and when the attempt fails, the blame falls on some sacrifice or fumigation left out (Philostr. VA 7.39). A text from the school of the philosopher Alexander of Aphrodisias says that sorcerers approach people in crisis, who all but pray for the prediction to be true, and make shrewd guesses from experience of what usually follows, which is a fair description of what modern conjurors call cold reading (Alex. Aphr. Mantissa 24). Eusebius notes that the rare oracle that comes true is carved on stone and proclaimed everywhere, while the countless failures are never mentioned (Eus. PE 4.2). Strabo, who stood by the broken colossus of Memnon at Thebes with the Roman governor and heard its famous sound at the first hour of the day, considered that someone posted around the base might have made it on purpose. He would not decide, "for because the cause is unclear, anything is more credible than that the sound comes from stones set like this" (Strabo 17.1.46). Here the panel sided with the stone: the sound was a natural noise from the damaged statue, and every ancient witness ties it to sunrise. Pausanias, told that sealed cauldrons at Elis filled overnight with wine, admits he missed the festival and says that anyone who believes it might as well believe the Ethiopians' tales (Paus. 6.26).

Machines of the kind these writers suspected did exist. The engineer Hero of Alexandria describes a little temple whose doors open by themselves when the altar fire is lit, driven by heated air pushing water into a hanging bucket (Hero, Pneum. 1.38). Hero writes as an engineer and never says that priests used such devices to deceive. The link to temple fraud is a modern inference.

Christian exposers and the pagan temples

After Christianity became the religion of the emperors, its historians reported what was found inside pagan temples. Rufinus of Aquileia, writing in 402 about the destruction of the Serapeum at Alexandria in 391, describes a small window aligned so that on a fixed day a sunbeam lit the lips of Serapis as the image of the Sun was carried in, and an iron Sun-image that rose toward a lodestone in the ceiling (Rufin. HE 11.23). A priest of Saturn named Tyrannus told noblemen that the god wanted their wives to spend the night in the temple, then crept through tunnels into the hollow statue, spoke as the god and seduced them, until one woman recognised his voice (Rufin. HE 11.25). At Canopus, he says, a priest defeated the Chaldaeans' fire-god with a porous water jar whose holes were stopped with wax, so that the wax melted, the jar "sweated" and the fire went out (Rufin. HE 11.26). A Latin church history of the sixth century tells how the Persian magi hid a man under the sacred fire to make it speak against a Christian bishop, and the bishop told the king that "the fire does not speak" (Hist. trip. = Socr. HE 7.8). Eusebius reports that the priests of a new oracular statue of Zeus at Antioch confessed under torture that the whole thing was a fraud contrived by Theotecnus, the official who had set it up (Eus. HE 9.11). A confession obtained by torture under a new regime proves little either way.

Augustine, the most careful of these writers, put temple marvels down to human skill working on the properties God had given to things.

ut in quodam templo lapidibus magnetibus in solo et camera proportione magnitudinis positis simulacrum ferreum aeris illius medio inter utrumque lapidem ignorantibus, quid sursum esset ac deorsum, quasi numinis potestate penderet

...so that in a certain temple, with magnets of matched size set in the floor and the ceiling, an iron statue hung in mid-air between the two stones, as if by the power of a god, for those who did not know what was above and below.

The panel notes that static magnetic suspension of this kind is physically impossible, and Rufinus himself says the iron Sun fell back, so the hanging statue was either a trick of another sort or a story improved in the retelling. Augustine also ran his own test of astrology. His friend Firminus was born, as two households had carefully recorded, at the same instant as a slave woman's child in the house of his father's friend, and the two boys, one free and one a slave, lived utterly different lives (Aug. Conf. 7.6.8–10).

Gregory of Tours, writing in sixth-century Gaul, reports impostors among Christians. Desiderius claimed that messengers ran between him and the apostles and "healed" the paralysed by having servants wrench their limbs, killing many. Another wandering holy man was searched by the bishop of Paris, who found a large bag of roots, moles' teeth, mouse bones and bear's claws and fat (Greg. Tur. Hist. 9.6). A woodcutter of Bourges, after two years of madness, proclaimed himself Christ, healed by touch and gathered more than three thousand followers, including priests (Greg. Tur. Hist. 10.25). Gregory credits him with "diabolical arts". The panel judged Desiderius a fraud and the Bourges healer a psychological case: touch-healing before a believing crowd works on complaints that respond to expectation, and predictions of illness in a time of plague will often come true. Isidore of Seville, a generation later, derived the Latin word for a conjuring trick, praestigium, from its dazzling (praestingere) of the eyes (Isid. Etym. 8.9.33).

Eye-turnings in the North

Old Norse had its own word for illusion, sjónhverfing, literally a "turning of the sight". Snorri Sturluson, the Icelandic Christian who wrote the Prose Edda around 1220, set the whole of his mythology inside one: Gylfi's visit to the Æsir is an illusion, and within it the giant Útgarða-Loki tells Þórr that every contest he lost was a sjónhverfing (Gylf. 47; see chapter VIII). In his history of the Swedish kings Snorri says Óðinn and Gylfi contended "in tricks and illusions" (Yng. 5), and the Norse translator of Gregory's Dialogues used the same word for the demonic fire that monks saw and Benedict did not (Greg. Dial. II.10, Old Norse). In Norse usage an illusion could be a trick, a demon's work or the gods' craft, and the word did not decide between them.

Ordeals, alchemists and the English stage

Medieval writers turned the charge of trickery on their own institutions. When Isolde carries the hot iron after a carefully ambiguous oath and is unburned, Gottfried von Straßburg comments that Christ turns like a sleeve in the wind and that her deceit and her "poisoned oath" saved her (Gottfried, Tristan 15739–15752). The panel classes the ordeal as a fraud open to the officiants, who controlled the heat of the iron and the manner of lowering the accused into the water. The Elizabethan William Harrison believed the ordeals had been abolished because people defeated them "by sundrie sorcerous and artificiall practises" (Harrison, Description of England II.9). Dante places the alchemists among the falsifiers in hell. Griffolino of Arezzo joked to a gullible patron that he could fly and was burned when he could not teach the art, though Minos damns him for alchemy (Inf. 29.109–120). Capocchio calls himself "a good ape of nature", a skilled imitator who never turned base metal into gold (Inf. 29.136–139).

Shakespeare put the exposure of a false miracle on stage in his early history Henry VI, Part 2. At St Albans a man named Simpcox claims that the saint has cured his lifelong blindness, and the pious king rejoices. Duke Humphrey asks him the colours of the courtiers' clothes, and Simpcox names them all.

If thou hadst been born blind, thou mightst as well have known all our names as thus to name the several colours we do wear. Sight may distinguish of colours, but suddenly to nominate them all, it is impossible.

If you had been born blind, you could as easily know all our names as name the colours we wear. Sight can tell colours apart, but no one could name them all the moment he first saw them.

Simpcox, who also claims to be lame, then jumps over a stool to escape a whipping (2H6 2.1). The story is older than the play; Thomas More told it. In the same play the priest Hume, who arranges the Duchess of Gloucester's séance, reveals that her enemies have paid him to lure her into it (2H6 1.2.87–107). Shakespeare returned to sham exorcism in The Comedy of Errors, Twelfth Night and King Lear, whose devils' names scholars trace to Samuel Harsnett's 1603 exposure of Catholic exorcism, and the English Faust Book of about 1592 treats conjuring tricks such as the straw horse and the Frankfurt jugglers' severed heads as real devilry (see chapter X).

The limits of the sceptics

Almost none of these writers doubted everything, and the library shows the edges of their doubt clearly. Theophrastus, who called amulets "sillier and less credible", accepted that certain plants act on the mind (Theophr. HP 9.19). Cicero ends his demolition of divination by insisting that "religion is not destroyed when superstition is destroyed", and he kept the ancestral rites (Cic. Div. 2.148). Pliny, scourge of the magi, asks whether words and charms have any power and admits that "life believes it at every hour without noticing" (NH 28.10–13). He vouches for the power of spittle and calls one of its effects "a marvel, but easy to test" (NH 28.35–38). Galen, who stripped the words from the scorpion charm, ran an on-off trial of a peony-root amulet on an epileptic child and believed it worked, explaining it by particles breathed in from the root (Gal. SMT 6.3.10). Hippolytus' recipe for walking on coals includes salamander, a fire-proof creature of legend (Ref. 4.32–33).

The Christian exposers took the machinery out of the temples and left the demons in. Augustine, having explained the magnet statue, at once concedes that demons do help the magi, and he gave a whole treatise to how demons know the future through their airy bodies and long experience (Aug. Div. daem. 3.7). Minucius Felix defined magic as illusion, making "what is not seem to be", and said the magi do it through demons (Min. Fel. Oct. 26.10). Tertullian could call magicians' wonders street-jugglers' tricks in the same sentence in which he credited them with the help of demons (Tert. Apol. 23.1). To these writers, exposing a pagan priest's speaking-tube and believing in a demon behind the pagan god were parts of one argument.

One ancient informant even offered a theory of who could be fooled. An Egyptian musician named Dionysius told Celsus that magic has power over the uneducated and the corrupt but none over those who have practised philosophy (Orig. Cels. 6.41). Origen answered that philosophers had fallen under the spell of Apollonius. The panel thought Dionysius largely right for effects that work through belief, since a person who rejects the frame of a curse is less likely to feel it. For the tricks in this chapter he was wrong. Lucian's educated believers and Augustine's startled visitors to the temple were fooled by mirrors, tubes and hidden machinery, which work on anyone who cannot see above and below.

What the panel concluded

The table sets the principal exposed wonders beside the verdicts of the panel on staging and suggestion. Ovid's Medea belongs here too, since she proves her rejuvenating art by turning an old ram into a lamb and then boils "plain water and herbs with no power in them" for the old king Pelias (Ov. Met. 7.297–349). A staged demonstration followed by a worthless product is the shape of the confidence trick in every age.

Of the 49 claims the panel judged in its group on devices and fraud, none showed an effect produced by the mechanism its practitioners claimed. Where something real happened, the cause was engineering, optics, chemistry, an animal's behaviour, the body's tolerance for a drug, or a performer's knowledge of people. The ancient exposers found most of these causes themselves, using the methods still used today. Daniel spread ash on the floor, Galen repeated the charm without the words, Cicero counted the dead at Cannae, and Duke Humphrey asked a blind man to name colours. Their blind spot was usually the wonder that belonged to their own side.

Chapter XIV

The numbers: the vocabulary of magic across time

Counting every word for magic in the library shows which terms each language used and how their meanings moved, and it also shows how strongly the shape of the library itself governs any number drawn from it.

What was counted

The other chapters read passages one at a time. This one reads the whole library at once. Every original-language edition in the library on 19 October 2026 was scanned, one per work and language, with translations left out (the collections added since are not counted; see the note at the end of this chapter): 1,822 Greek and 1,879 Latin editions, 226 Old Norse, 406 Middle High German, 33 Old French chansons de geste, Dante, 77 English Renaissance texts, 1,196 Akkadian tablets from the Oracc projects, and the Pyramid Texts of five Egyptian kings and a queen.

For each language a family of magic words was built and then cleaned by an AI agent, since a search for μάγ- also finds μάγειρος, "cook", and Latin mag- finds magis, "more". Each work was given an approximate date, and the hits became a rate per million words for each language and era. Most languages also have a yardstick family for divination (Greek μάντις, "seer", and its relatives; Latin mathematicus and astrologus; Norse spá, "prophecy"; Akkadian bārû, the diviner).

The figure at the head of this chapter plots these rates, and the table below gives the main ones.

Which words each language used

Egyptian has one word that matters here, ḥkꜣ (heka). It occurs 25 times in the Pyramid Texts, the spells carved in the burial chambers of Old Kingdom kings. Heka is a power that gods and the dead king possess, store and even eat. In the "Cannibal Hymn" the gods' bellies are "full of heka" (PT 273, §397). Nothing in the word is shameful.

Akkadian, the language of Assyria and Babylonia, can only be counted through the English glosses Oracc attaches to each word. The commonest term is šiptu, "incantation", written with the sign EN₂ at the head of a spell, as in a bilingual tablet whose first line opens en₂ and is glossed "incantation, spell" (OECT 11 26, obv. 1). The professional who recites such spells is the āšipu, rendered "exorcist" or "incantation priest". An inscription of Esarhaddon, king of Assyria in the seventh century BCE, names the exorcist among the cult personnel and experts he installed before the gods of Babylon, beside purification priests, lamentation priests and musicians; the line is partly restored (RINAP 4 53, rev. 1). In the library as counted here, the hostile counterpart, kišpū, "witchcraft", is rare; the anti-witchcraft series added since make it common (see the note at the end of this chapter). All twelve of its tokens here sit in late scholarly commentaries, for example a broken line glossed "that man, in sorcery, brought to tears" (SpTU 2 39, rev. 6').

Greek is the richest. Across 27.9 million dated words the main families are μάγος and μαγεία (930 tokens), γόης and γοητεία, "sorcerer" and "sorcery" with a strong flavour of fraud (838), ἐπῳδή, the charm sung over someone (449), and βασκανία, the evil eye (318). φάρμακον, drug, poison or spell, is kept out of the strict total for reasons given below.

Latin has magus, magia and magicus (1,975 tokens in 31 million words), maleficium and maleficus (515), veneficium, which covers both poisoning and sorcery (400), incantatio (304), hariolus, a soothsayer (183), and praestigiae, conjuring tricks and illusions (148).

Old Norse uses fjǫlkynngi, literally "much-knowing", as its most general word for magic (45), seiðr for a staged rite of seeing and harming (40), and galdr for spoken or sung spells (18). Middle High German has zouber, "sorcery" (81), and goukel, "jugglery" or deceptive magic (45). Old French epic has almost no word for magical practice and no sorcier at all. Its marvels are fairy work, with fée 56 times against 17 for enchant-. In Renaissance English, "witch" (147) is far commoner than "magic" (32).

How the rates move

Greek: three peaks, each made by a few authors

Greek magic vocabulary runs at about 146 per million words in archaic and classical texts, sinks to 43–82 between the fourth and first centuries BCE, climbs to 287 in the fourth century CE, collapses to 33 in the sixth to eighth centuries, and rebounds to 181 in the Byzantine chronicles. Each movement has a few names behind it. Herodotus alone supplies 70 of the 150 earliest tokens, mostly when describing the Persian Magi. The fourth-century CE peak belongs to Eusebius (224 tokens), Libanius (121), Epiphanius (89) and Iamblichus (63), in an age of Christian–pagan polemic and of theurgy, the Neoplatonists' ritual for raising the soul to the gods. The early medieval trough exists because the commentators on Aristotle, Philoponus and Simplicius, contribute over two million words and almost no magic. About half of the Byzantine rebound is Zonaras' chronicle (114 of 224 tokens), and much of the rest is a medieval epitome of the pseudo-Clementine romance, in which Simon Magus is the apostle Peter's adversary.

The word γοητεία shows the polemical peak most clearly. It runs at 5–17 per million before Christ and at 60–72 in the third and fourth centuries CE. Origen's Against Celsus alone contains 84 tokens. Answering the pagan Celsus, who had claimed that Jesus learned his wonder-working powers in Egypt (Orig. Cels. 1.28), Origen asks whether powers really "come about through sorcery from wicked daemons" while none at all is worked by the divine and blessed nature (Orig. Cels. 2.51). The charge of goēteia was a weapon both sides used.

In most Greek eras divination words outnumber magic words, by four to one in the earliest texts and by seven to one in the first century BCE. The exceptions are the fourth century CE and the Byzantine chronicles, where magic words are slightly ahead. Akkadian reverses this. Incantation outnumbers diviner by about four to one, because the tablets in the library are largely incantation texts.

Latin: from courtroom and encyclopedia to pulpit

Latin begins high with Plautus (200 per million), largely through his comic soothsayers and poisoners, falls in the age of Cicero and Livy, and peaks at 384 in the second century CE. That peak is essentially one man. Apuleius, tried for magic and the author of a novel whose hero is turned into an ass when he tries a witch's ointment (Apul. Met. 3.21), supplies 157 of the 212 tokens. The first century CE owes its rate to Pliny the Elder (200 tokens) and to the declamation schools, where poisoning and sorcery were stock courtroom themes. veneficium peaks then at 64 per million and has fallen to 4 by the sixth to eighth centuries. Late antique Latin, nearly 25 million words from the fourth to the eighth century and most of it from Migne's Patrologia Latina, holds steady at 100–122, carried by Jerome in the fourth century and Augustine in the fifth.

How the meanings moved

For the main families every occurrence was sorted by other words in the same passage (Persia, star and Bethlehem, Simon, Pharaoh, poison, remedy). Many library passages are long paragraphs, so the results are indicators, checked against passages read in full.

μάγος: from a Median tribe to a Christian villain

In the earliest Greek the Magi are a people. Herodotus lists them as one of the six tribes of the Medes: ἔστι δὲ Μήδων τοσάδε γένεα … Βούδιοι Μάγοι, "these are the tribes of the Medes … the Budii, the Magi" (Hdt. 1.101). Of the 84 tokens before 400 BCE, 61 are in Persian or Median settings. The insult is already present in tragedy.

ὑφεὶς μάγον τοιόνδε μηχανορράφον, δόλιον ἀγύρτην, ὅστις ἐν τοῖς κέρδεσιν μόνον δέδορκε, τὴν τέχνην δʼ ἔφυ τυφλός.

He has secretly set on me this magus, this stitcher of schemes, a crafty beggar-priest who has eyes only for profit and was born blind in his art.

Oedipus is abusing the prophet Teiresias. A Hippocratic doctor uses the same cluster of words for the men who first called epilepsy a sacred disease, men "such as the magi, purifiers, beggar-priests and charlatans are now" (Hippoc. Morb. Sacr. 1). By the second to fourth centuries CE, sorcery in general accounts for more than half of all tokens. Christian Greek adds the Nativity Magi, a small share, and Simon, the Samaritan who was "practising magic" (μαγεύων) before he met the apostles (Acts 8.9), grows from 2 tokens in the first century to 32 in the fourth and 42 in the Byzantine chronicles. In medieval Greek the magus par excellence is a heretic.

magus: priest, sorcerer, wise man from the East

Latin keeps the Persian sense alive as a learned defence. At his trial Apuleius argues that Persarum lingua magus est qui nostra sacerdos, "in the Persian language a magus is what we call a priest" (Apul. Apol. 25). His audience used the word otherwise. In the first and second centuries CE generic sorcery is the dominant sense, four to six times as common as the Persian. Christian Latin then splits the word. By the fifth century the Nativity Magi (273 tokens) outnumber generic sorcerers (249), and the biblical court magicians of Pharaoh and of Daniel's Babylon, together with Simon Magus, add about 150 more.

maleficium: from crime to sorcery

The Latin word with the sharpest shift is maleficium, "evil-doing". In Cicero's defence of Sextus Roscius it occurs 35 times, every one meaning a crime.

scelestum, di immortales! ac nefarium facinus atque eius modi quo uno maleficio scelera omnia complexa esse videantur!

Immortal gods, a wicked and unholy deed, of the kind in which a single evil act seems to take in every crime there is!

The crime is parricide. In the first century BCE only 27 percent of maleficus tokens have any magic word nearby. In the fourth and fifth centuries the figure is 68–72 percent. The late authors say outright that the change has happened. Augustine reports that the defenders of the art some call magian, others by the uglier name goetian or the more honourable theurgian, try to split it into a praiseworthy kind and a condemned kind, whose practitioners are those quos et maleficos uulgus appellat, "whom the common people also call malefici". Augustine himself holds both kinds bound to demons (Aug. Civ. 10.9). Two centuries later Isidore makes the two words equivalent.

Magi sunt qui vulgo malefici ob facinorum magnitudinem nuncupantur. IIi et elementa concutiunt, turbant mentes hominum, ac sine ullo veneni haustu violentia tantum carminis interimunt.

Magi are those commonly called malefici because of the enormity of their crimes. They shake the elements, disturb men's minds, and kill by the sheer force of a spell, without any draught of poison.

As maleficus became more magical, it became rarer, falling from 48 per million in the first century BCE to 7 in the fifth century. It narrowed onto sorcery and left ordinary talk about crime.

carmen: the song that is also a spell

Latin carmen means song or poem, and also a ritual formula. The spell sense is as old as Roman law. Pliny quotes the Twelve Tables against anyone qui malum carmen incantassit, "who has chanted an evil song" (Plin. NH 28.17–18), and Virgil's enchantress sings that Carmina vel caelo possunt deducere Lunam, "songs can even draw the Moon down from the sky" (Verg. Ecl. 8.69). Still, in the first centuries BCE and CE only 12–14 percent of tokens sit near magic words, because most are poems. In the fifth to eighth centuries the share is 28–34 percent, as in Isidore's violentia tantum carminis above.

φάρμακον: the drug that would not stay one thing

Greek φάρμακον cannot be counted as magic without first being sorted. Homer's Odyssey already uses it for arrow poison (Od. 1.261), Helen's drug (Od. 4.220), Circe's potions (Od. 10.236) and Hermes' protective herb (Od. 10.302). Of all 9,169 Greek pharmak- tokens, 72 percent are in medical writers, and second-century medical writing, overwhelmingly Galen, accounts for about 5,750. In medicine φαρμακεία means a purge. Plato, writing on the law, separates bodily poisoning from harm done "by sorceries, incantations and so-called bindings" (Pl. Leg. 11.933a). The Greek Bible pushes the word toward sorcery. In the Septuagint, Pharaoh's experts are φαρμακοί and ἐπαοιδοί, "drug-men" and "chanters" (Exod. 7.11 LXX), and Paul lists φαρμακία between idolatry and enmity among the works of the flesh (Gal. 5.20). Christian Greek inherits these biblical senses.

Persuasion as enchantment

The charm-word ἐπῳδή was turned into a metaphor early. Gorgias, defending Helen, says that inspired speech works like an incantation, and that "two arts of sorcery and magic have been found, which are errors of the soul and deceptions of opinion" (Gorg. Hel. 10). The language of trickery is a constant minor strand in every Greek period, and in Latin the "conjuring trick" word praestigiae peaks with Tertullian in the third century, at 32 per million.

The North and medieval Germany: who names what

Old Norse literature calls seiðr unmanly. Ynglinga saga opens Heimskringla, which is traditionally attributed to Snorri Sturluson (c. 1230); the library's text is a seventeenth-century copy. The saga says that Óðinn practised the art himself and that it carried so much shame that men thought it unfit to practise.

En þessi fiolkyɴge er framit er fylgir ſva micil ergi at eigi þótte karl mǫnnom viþ at fara. oc var gyðionom kend ſ iþrót.

But this magic, when it is performed, brings with it so much unmanliness that men thought it unfit to practise, and the art was taught to the priestesses.

The word ergi means effeminacy with a sexual charge. The counts complicate the statement. Many of the practitioners the texts actually name are men, the seiðmenn whom one king keeps at court among his entertainers (Yng. 23) and another, Óláfr Tryggvason, burns in a hall (ÓT 62). In the sagas and the Norwegian Christian laws both seiðr and galdr are illicit heathen practice (Magnúss Lagabœtis kristinréttr 1–2).

Middle High German shows the gap between self-description and condemnation most sharply. The 30 short charm texts in the corpus, against bleeding, worms or eye disease, use segen, "blessing", 19 times and zouber never. The word zouber appears where the act is condemned or narrated, in sermons, chronicles, the Troy romance and confession formulas.

in zoubere. in goukele. in heilsôde. in getrügede.

In sorcery, in jugglery, in omen-seeking, in deceit.

The person who recited a blood charm called it a blessing. The priest hearing confession called it sorcery. A count of zouber measures the condemnation of practices that called themselves by another name.

The English evidence carries its own warning. A naive count found 63 "witch" tokens in Macbeth, and 59 of them were speaker labels such as "First Witch". Once those were removed, the leading source was Holinshed's chronicle, where combatants in a trial by battle swear that they delt with no witchcraft, nor art magike (Holinshed, Richard II).

What the numbers cannot show

• The corpora are different kinds of text. Akkadian and Egyptian top the table because the library holds ritual texts for them: bilingual incantation tablets (about 107 of the 136 "incantation" glosses) and the Pyramid Texts, which are spells from beginning to end. Greek and Latin are elite literature, philosophy, medicine and church writing. Old French is 33 epics, Italian is Dante alone, and English is chiefly Holinshed and Shakespeare. Comparisons across languages therefore compare genres, and say nothing about how magical a culture was.
• A few authors make each era. Herodotus, Plato, Apuleius, Pliny, Origen, Eusebius, Jerome, Augustine, Zonaras and Holinshed each create or erase a peak.
• The dates are assigned. Each author was given an approximate floruit. Undated Patrologia items were placed by volume order, Norse texts by manuscript date, and Middle High German and Old French as single blocks. Latin after the eighth century is almost absent, because the library's Patrologia ends with Bede. Early-modern Latin filed there under ancient names was counted separately.
• Matching is by word form. Forms not on the lists are missed. Some false friends remain, such as the name Mago in Diodorus, and βάσκανος often means merely "envious".
• Some readings are uncertain. At Herodotus 7.191, where the Magi chant to the wind (Hdt. 7.191.2), the text is disputed, and the library's texts of Isidore and the Septuagint carry scanning errors, visible in "IIi" in the quotation above.

Three results survive these limits. Words for magic often began as neutral or ethnic labels (a Median tribe, an evil deed, a song) and acquired the sense of illicit sorcery through use, often by opponents. Greek and Middle High German speak of divination far more often than of spellcraft, while Latin, English and the Akkadian tablets go the other way, though each language's divination yardstick is a different and partial list. And practitioners tended to call what they did priesthood, healing or blessing, which left the word "magic" largely to those who condemned it.

Conclusion

What the texts tell us about magic

Over nearly four thousand years of text in one library, the techniques of magic changed little, while the right to use them, the theory of how they worked and the people blamed for them changed again and again; where the techniques worked, they worked through drugs, care, knowledge and belief.

A long history in five movements

The oldest texts in the library treat what we would call magic as a power belonging to gods and kings. In the Pyramid Texts, carved inside royal tombs from about 2350 BCE, heka is something the dead king eats, holds and is.

č̣(ṭ) mṭw ꜣwꞽ p.t śṭꜣ tꜣ ṭp-ꜥw(.wꞽ) Ppy pn / Ppy pw ḥkꜣ.w / Ppy pw ẖr(.ꞽ) ḥkꜣ

Words to be spoken. The sky stretches out (the verb is uncertain), the earth quakes before this Pepi. Pepi is heka. Pepi is the one who holds heka.

In Assyria and Babylonia the same kind of power belonged to salaried, hereditary temple scholars. The exorcist claimed nothing for himself. He belonged to Ea and worked with Ea's hand (incantation against the oath, GAAL 2 pl. 24–25, 15–22). His enemy was the witch, whom he fought by burning her effigy, and the library holds no law against her. In both cultures the practitioners wrote the record themselves, and none of it doubts them.

The second movement, between about 800 and 350 BCE, named the category and banned it. The Hebrew Bible kept the divine source of power and abolished the technician, listing the diviner, the sorcerer and the medium as practices of "the nations" and offering a prophet in their place (Deut 18.9–14). The Greeks took a title from Persia. Herodotus' magus is a priest without whom a sacrifice is not lawful (Hdt. 1.132.3), yet within a generation a Hippocratic doctor was lumping magoi with purifiers and charlatans (Hippoc. Morb. Sacr. 1). The general word for magic entered the library as an insult for a rival.

The third movement, from about 200 BCE to 400 CE, belongs to law, books and philosophy. Rome punished harmful charms, poisoning and astrology about the emperor. It also honoured a consul who devoted himself and the enemy army to the gods below (Livy 8.9), while lead tablets bearing Germanicus' name were a capital crime. Pliny wrote the first history of magic as one art spread across the world, "the most fraudulent of the arts" (Pliny NH 30.1). Greek writers of the same centuries sold magic, tested it and exposed it, and the Neoplatonists turned part of it into theurgy, a philosophy of rites.

The fourth movement is Christian. From the gospels to Bede, Christian writers took it for granted that magic worked and traced its power to demons. Augustine's test decided what was still allowed. A herb drunk for a stomach ache is medicine, and the same herb hung round the neck is a sign agreed with demons (Aug. Doctr. chr. 2.29.45). Isidore copied Pliny's sentence about magic's world-wide power and changed its cause from fraud to "the evil angels" (Isid. Etym. 8.9.3). Meanwhile the Church built a counter-practice of holy names, relics and blessed objects that outsiders could hardly tell from what it condemned.

The fifth movement runs from the North to the English stage. Snorri gives seiðr to Óðinn and explains that it passed to priestesses because it shamed the men who practised it (Yng. 7). German monks copied blood charms that retell Christ stopping the Jordan, and Dante put the sorcerers of Italy in Hell among the frauds. In 1599 James VI told his heir that witchcraft headed the crimes a king must never pardon (Basilikon Doron 2). A decade later Shakespeare gave Prospero a speech built from Medea's prayer in Ovid, by way of Golding's English, and then had him give the art up.

But this rough magic / I here abjure … I'll break my staff, / Bury it certain fathoms in the earth, / And deeper than did ever plummet sound / I'll drown my book.

But this crude magic I here renounce … I will break my staff, bury it many fathoms in the earth, and drown my book deeper than any sounding-line has ever reached.

The history ends there because, when this report was written, the library did; the English books added since carry the argument on to Glanvill's Saducismus Triumphatus of 1681 (Saducismus Triumphatus, title page; chapter X).

What stays the same

The techniques hardly change. Binding, figurines, love charms, consulting the dead, healing by a sacred story, amulets, the power of names and the logic of "as this wax melts, so may he melt" recur in culture after culture, often in words that can be traced from one text to the next. The Assyrian exorcist smothered a clay figure of the witch in kiln slag and ashes, and more than two thousand years later Holinshed's witches roasted a wax figure of King Duff. The figurine survived intact. Its owner changed.

Each culture had words for its own powerful rites, such as heka, āšipūtu, carmen, galdr and segen, and other words for other people's, such as kišpū, goēteia, veneficium and witch. None but Egypt has a neutral word for magic as a whole. The category exists as a name for other people's rituals. The line between proper rite and forbidden magic was drawn by the same four tests almost everywhere. A rite was suspect if it was secret, harmful to the community, foreign or sold. The same act could fall on either side within one culture, and nearly every society kept for itself, under another name, the arts it condemned in others.

Doubt is constant too, wherever practitioners did not write the record. Cicero's Diagoras counted the drowned sailors missing from the votive tablets, Hippolytus explained a sealed-question oracle with vitriol and oak-gall, and Galen tested an amulet by taking it off. Almost none of them doubted everything. Each sceptic left some power standing, usually one that belonged to his own side.

What differs

Authority sat with a temple that licensed its own exorcists in Mesopotamia, with a prophet who replaced the technician in Israel, with law courts in Athens and Rome, and with a Church that saw demons behind every rite but its own. The power was a substance eaten by kings in Egypt, a divine commission read in written signs in Mesopotamia, God's will in Israel, nature, belief and the daimones in Greece, exact words and hidden virtues in Rome, demons for the Christians, and knowledge, chant and correctly carved runes in the North.

Blame fell differently. Greek and Roman poets imagined magic as women's drugs, and Pliny reports the common belief that it was "the one science in which women excel" (Pliny NH 25.10). The Norse texts worried about men doing a woman's rite. In the Mesopotamian texts the survey read, the only female practitioner is the witch; the anti-witchcraft rituals added since also have a pregnant woman perform parts of her own protective rite (CMAwR 5.5.2:11–13). Punishment ranged from a burned effigy in Assur to death in Israel's law, execution in Athens, stoning and burning in the sagas, penance in the early Church and, in James's England, a crime beyond pardon.

Much of the apparent difference is a difference of genre. Of the Mesopotamian passages surveyed, 54 per cent are practical instruction, and of the medieval German ones 46 per cent are charms and recipes, while Latin poetry is 40 per cent fiction. Women are 63 per cent of named practitioners in literary fiction and 6 per cent in practical instruction. A culture known from its own manuals looks male and confident, and one known from its poets looks lurid and female.

Did any of it work?

The plausibility panel judged 283 claims that modern knowledge can test.

About four claims in ten rest on something that really happened, but only eighteen happen as described, and the clearest of them, such as the honey that felled Xenophon's army (Xen. Anab. 4.8.20), carry no magic at all. The other real cases are real by another route. Odysseus' uncles bound his wound with skill before they sang over it (Hom. Od. 19.456–458). Fits end, bleeding stops, fevers come and go and storms blow out, so a rite performed at the worst moment was nearly always followed by improvement. Curses harmed people who learned of them, and charms soothed people who trusted them. A few practitioners knew a regularity of nature and staged it as power, as Aglaonice is said to have done with the eclipsed moon (Plut. Conj. praec. 48).

The frauds were real, and in the panel's group on devices and fraud a source in the library names the mechanism for about half the claims. The impossible claims are seldom pure invention, since many rest on a real experience misread, such as sleep paralysis taken for a night-hag or epilepsy for possession. In no passage of the library does a word, a gesture or an image act on matter, on the weather, on the future or on a person who does not know of it.

The texts also explain why magic lasted when most of it failed. People remembered hits and forgot misses, as Diagoras said of the votive tablets of the saved (Cic. ND 3.89). Every failure could be blamed on a step left out. Bodies and weather recover on their own. And rites did social work, sorting the guilty among believers, giving the frightened something to do and keeping craft secrets, none of which required the rite to work physically.

What remains open

Twenty-eight claims could not be settled. The Delphic vapour is the best known. Plutarch describes a sweet-smelling breath in the consulting room (Plut. De def. or. 50), geologists proposed ethylene, and later surveys found the site could not produce intoxicating levels, though a small effect is not ruled out. Helen's grief-killing drug fits opium and cannot be tested without a named plant. How well ancient abortifacients worked is still argued, and what killed Germanicus the sources leave open.

Larger gaps came from the library itself. When this report was written it held no Greek magical papyri, curse tablets or magical gems, Egypt's own language appeared only in the Pyramid Texts, the Mesopotamian anti-witchcraft series survived only in commentaries, and the Hebrew Bible was read through English and Greek. Practitioners spoke for themselves in only a few corpora, and women who used charms were seen almost entirely from outside. The library has since gained the papyri, the tablets, the Coptic books, the Book of the Dead and the Egyptian medical papyri, Maqlû with its companion rituals and the Hebrew Bible. Read against them, the history looks much the same: the practitioners keep the hostile words for others (chapter XI), and their claims work, where they work, through a substance, through the natural course of an illness, or through a mind that knew of the rite (chapter XII). They also show women at work, laying down a supplication against a wrong (PGM XL 1–12) or performing a rite for their own pregnancy (CMAwR 5.5.2:11–13). Magical gems, the great Greek handbooks and the Coffin Texts are still missing. Some readings stay disputed as well, among them whether moon-drawing first meant an eclipse and whether Prospero's renunciation is Shakespeare's farewell.

The most basic question stays open too. In these texts magic names a line that each culture drew between its own powerful rites and other people's, and the line kept moving. Across the whole library people used a small set of techniques, argued over who might use them, and noticed, often in the same passage, which parts worked. Those parts were the drug, the dressing, the observation and the believing mind.

Method

How this report was made, and what was checked

AI agents did the research and the writing, and scripts checked the block quotations and the references against the library. No specialist has reviewed the report, so read its translations and interpretations as a well-sourced first draft that anyone can check.

Who did what

The agents were Claude models, made by Anthropic, working read-only on the Classica Codex library. The checks were scripts that look passages up in that library.

The October 2026 update

The report was written on 19 October 2026. In the days that followed, the library gained the practitioners' own texts, which the report had said it lacked: 841 curse tablets, 151 Greek magical papyri, 353 Coptic magical texts, 2,332 Egyptian texts from the Thesaurus Linguae Aegyptiae with the Book of the Dead among them, and Maqlû with the other anti-witchcraft rituals. Then came the Hebrew Bible in Hebrew, the Pseudepigrapha, five tractates of the Talmud, the Quran and twenty-one early English books on spirits and witches, James VI's Daemonologie, Scot's Discoverie and Marlowe's Doctor Faustus among them.

Five AI agents first read the first group of texts against the report, five more tried to break what they found, and a last one read the result. Five AI drafting agents then wrote the additions chapter by chapter, from those notes, from surveys made for a companion study of demons and angels, and from passages they opened in the library. The same kind of script checked every new citation against the library and every new block quotation against the passage it cites, and refused the page if any failed; it also listed quoted phrases in the prose that it could not find, for a reader to check. Five AI reviewers then read the drafts as hostile specialists, a skeptic tested each objection, and the objections that stood were corrected. Statements the new collections made false were rewritten. A last AI agent then read the updated report through and corrected the places where it disagreed with itself. The word counts in Chapter XIV and the 283 verdicts are left as they were made; the sixty further verdicts on claims from the new texts are reported separately in Chapter XII. No specialist has reviewed the additions either.

What the checks establish, and what they do not

The scripts establish two things: every block quotation occurs in the library's edition at the passage cited, and every reference on this page names a passage that exists. They do not show that a reference names the right line for the sentence citing it, or that the passage says what that sentence says. Words quoted inside the running text were read by AI agents but not compared by script.

Nor do the checks establish that the English translations are right. The agents made them; AI fact-checkers compared some with the originals, but nobody has checked them against a specialist's reading. The checks do not show that a passage means what the report says it means, that its date and context are as stated, or that the library's edition is the best text: several are transcriptions of a single manuscript or digitised older editions.

In Chapter XII and the verdicts, statements about modern science and scholarship come from the panel's web searches; where the panel could not find a reference, the data marks it, and Chapter XII says so for the verdicts it discusses. What the other chapters say about modern scholarship is the writing agents' own, and nothing has checked it against the works they name. The verdicts are judgments by AI models weighing the evidence, not experiments, and the 283 count judgments: a few claims were judged in more than one panel.

No historian, classicist, Assyriologist, Egyptologist or medievalist has reviewed the report. What it offers instead is that every claim can be checked. Each citation copies the passage's identifier in Classica Codex, the free library the texts were read in, and the data below gives every surveyed passage and every verdict with its reasoning.

The data

• passages.csv (1.7 MB) and passages.json: the 1,343 surveyed passages, with excerpt, translation, summary, practice types, practitioner, the author's stance and the result of the library check. Some passages were found by two or three surveys, and two long library passages each hold several entries from one survey, so 40 library locations appear in more than one row.
• verdicts.csv (0.6 MB) and verdicts.json: the 283 verdicts, with claim, confidence, sources, the panel's reasoning and the modern studies it relied on, linked to the surveyed passages that share a source's reference.

Both are shared under CC BY-SA 4.0. Each JSON file describes its fields; in the CSV files, list columns hold JSON arrays. The 57 excerpts that quote the Perseus Art & Archaeology catalogue, and 1 from a CATMuS subset whose licence is not confirmed, are left blank.