Reappraising the Homeric Moly: Mediterranean Geophytes as Antidotes to Antimuscarinic Intoxication in the Odyssey

1. Introduction

The Homeric poems are not only literature. Composed within a long oral tradition and fixed in writing around the eighth century BCE, they also functioned as repositories of social memory, encoding navigational, technical, ritual and medical knowledge in mnemonically stable narrative form [1,2,3]. Reading them as such is a legitimate historiographical operation provided it is disciplined: the interpreter must distinguish what the text asserts from what the modern reader wishes it to assert.

Few episodes have attracted more of that aspiring reading than the moly of Odyssey 10. Odysseus, having sent half his crew to the palace of Circe on the island of Aeaea, learns that they have been drugged and transformed into swine. On his way to rescue them he meets Hermes, who uproots a plant and hands it to him as a pharmakon esthlon, an “excellent drug”, describing it in five hexameters that constitute the whole of the botanical evidence available to us. The plant is black at the root, its flower resembles milk, the gods call it moly, and it is hard for mortal men to dig up (Od. 10.302–306). Identifying this plant has occupied commentators for more than twenty-five centuries, with results that do not encourage confidence. Stannard catalogued proposals spanning eleven genera in eight families; almost all rested on morphology alone, and almost all contradict one another [4]. The field has been, in short, a graveyard of confident attributions. The question changed character in 1983. Plaitakis and Duvoisin introduced a criterion nobody had applied before, a pharmacological one. If Circe’s drug produced an antimuscarinic delirium, then the antidote must have possessed anticholinesterase activity; and the one Mediterranean plant that then seemed to fit both the description and that pharmacology was the snowdrop, Galanthus nivalis L., the source of galantamine [5]. It was an elegant argument, testable in principle, and it has been repeated for forty years with steadily diminishing scrutiny, most often as a decorative preamble to reviews of galantamine in Alzheimer’s disease, where nobody had any particular reason to check it. In 2024 Molina-Venegas and Verano finally did check it and found several of its philological premises unsound and its biogeographical premise close to untenable [6]. We take their criticism as our starting point, and ask a narrower question, one that interests us as historians of medicine and pharmacology rather than as botanists: given what the text actually constrains, and given what is now known about the phytochemistry of Mediterranean geophytes and about the clinical management of antimuscarinic delirium, which candidate best accounts for the episode and what kind of knowledge would such a plant represent?

2. Materials and Methods

This is a narrative critical review. Ancient sources are cited by author, work, book and line according to the conventions of classical scholarship; the Greek text of the Odyssey follows Allen [7], Theophrastus follows Hort [8] and Dioscorides follows Wellmann [9]. Translations given here are literal renderings using the current online translation software. Pharmacological and phytochemical evidence was retrieved from PubMed, Scopus and Web of Science using combinations of the terms moly, Galanthus, Pancratium, Amaryllidaceae alkaloids, galantamine, acetylcholinesterase inhibition, antimuscarinic delirium, physostigmine, Mandragora and Hyoscyamus, without date restriction. Taxonomic nomenclature follows World Flora Online; phylogenetic relationships within Amaryllidoideae follow Meerow and Snijman [10]. Five constraints were extracted from the primary texts and each candidate scored against them. No quantitative weighting is claimed. The scoring is a way of keeping the argument honest, not a test that decides anything by itself.

3. The Homeric Evidence and the Constraints It Imposes

3.1. The Two Relevant Passages

The first passage describes the effect of the drug (Od. 10.237–240): Circe mixes pharmaka lugra into a kykeon of cheese, barley meal, honey and Pramnian wine, so that the men forget their homeland; she then strikes them with her wand and pens them, and they have the heads, voice, bristles and body of swine, but their mind remained unchanged as before (αὐτὰρ νοῦς ἦν ἔμπεδος ὡς τὸ πάρος περ). The second describes the remedy (Od. 10.302–306): Hermes draws the plant from the earth and shows its nature; it was black in the root, but the flower was like milk; the gods call it moly, and it is hard for mortal men to dig. The encounter takes place as Odysseus moves inland through a wooded bessa (Od. 10.275), a term of uncertain etymology usually rendered “glade”, “glen” or “grove” and elsewhere formulaically associated with mountain topography [6,11].

3.2. Five Constraints

C1 — A subterranean storage organ with dark outer coverings. Early Greek botany did not distinguish bulbs from roots, routinely describing the former as “round roots” (cf. Theophrastus, HP 6.6.8, 7.12.3); melan applied to a bulb with dark brown tunics is unremarkable, since colour terms in archaic Greek encode saturation and luminosity as much as hue [6].

C2 — A white flower. The comparison to milk is a comparative anchor rather than a colour word, and since Greek milk was caprine and ovine, hence markedly white, the constraint is firm.

C3 — Exceptional difficulty of extraction. This is the least exploited datum in the entire tradition. Theophrastus explicitly notes that the plant known to his informants was not hard to dig up, contrary to what Homer says — a remark that only makes sense if the difficulty was a recognised diagnostic feature of the Homeric plant and not a mere index of divine prerogative.

C4 — An inland, wooded, shaded site. Whether or not the geography of Aeaea is fictitious, the poem places the encounter away from the shore.

C5 — Pharmacodynamic capacity to reverse the narrated syndrome. The plant protects Odysseus before exposure and its administration is contemporaneous with the reversal of the crew’s condition; on the pharmacological reading, this requires constituents able to restore central cholinergic transmission against competitive muscarinic blockade. Unlike the preceding four, which are morphological or ecological, this constraint is pharmacological, and it is applied below as a discriminator rather than as a formality: it excludes Allium nigrum, inverts the verdict on Mandragora, and qualifies rather than satisfies the verdict on Peganum harmala.

4. The Clinical Phenotype of Circe’s Kykeon

The syndrome as narrated comprises amnesia for home, a florid zoomorphic delusion of bodily transformation, a change of voice to inarticulate grunting, and preserved inner awareness. Each element has a straightforward correlate in central antimuscarinic toxicity. Tropane alkaloids — hyoscyamine, atropine and scopolamine — produce anterograde amnesia, vivid and often zoomorphic hallucinations, disturbance of body schema, mumbled or garbled speech attributable in part to profound xerostomia, and carphologia; peripheral features include mydriasis, tachycardia, anhidrosis, urinary retention and ileus [12,13,14]. Scopolamine in particular is amnestic at doses well below those producing coma, and the Mediterranean Solanaceae available to an Archaic Greek household — Mandragora officinarum, Hyoscyamus niger, Atropa belladonna — were demonstrably in medical use for analgesia, sedation and anaesthesia into the Roman and Byzantine periods [13,15,16].

The detail of the noos empedos is diagnostically the most valuable line in the passage, and it is usually treated as mere pathos. It excludes the two obvious alternatives. Opioid intoxication from Papaver somniferum, long proposed for the Homeric nepenthes [17], produces sedation and miosis rather than agitated delusion with retained self-awareness. Ethanolic or sedative-hypnotic intoxication produces global obtundation. Only the antimuscarinic toxidrome characteristically yields a patient who is perceptually deranged, behaviourally disorganised and yet, at intervals, aware of being so — the clinical picture that a pre-scientific narrator, lacking a vocabulary for delirium, would most naturally render as a human mind trapped in an animal body.

A further philological detail supports the association. Dioscorides records that mandrake was also called circaea, the plant of Circe (De materia medica 4.75), and Greek knew henbane as hyoskyamos, “pig bean” — a coincidence with the swine of the myth that later commentators found irresistible and that should be treated with corresponding caution [6,18].

5. The Pharmacological Thesis and Its Clinical Warrant

The modern therapeutic analogue of moly is physostigmine, a tertiary carbamate and reversible acetylcholinesterase inhibitor that crosses the blood–brain barrier and competitively restores cholinergic neurotransmission at central muscarinic synapses. Case series and prospective comparisons show reversal of agitation in the great majority of patients and of delirium in most, with superiority over benzodiazepines, which control agitation poorly and delirium not at all [19,20,21,22]. Its limitations are equally instructive: a short duration of action relative to that of most anticholinergic agents, with delirium recurrence in a substantial proportion of complete responders, and dose-related cholinergic toxicity including seizures [20,21]. Longer-acting cholinesterase inhibitors, galantamine among them, have been used successfully in the same indication, and the reversal of the central anticholinergic syndrome specifically by galantamine was reported as early as 1977 [20,23].This is the clinical foundation on which Plaitakis and Duvoisin built [5], and it is sound; we have no quarrel with it. An orally administered Amaryllidaceae bulb containing galantamine and congeneric alkaloids would be expected, in principle, to attenuate central antimuscarinic delirium. Nor is the underlying pattern unfamiliar. The route by which galantamine entered modern pharmacopoeias — Mashkovsky’s observation of Caucasian and Ural folk use of Galanthus woronowii, leading to isolation of the alkaloid in the early 1950s and eventually to its registration for Alzheimer’s disease — is itself a documented case of folk knowledge running ahead of pharmacological description [24,25]. The general claim that the moly episode encodes some empirical knowledge of an anticholinesterase antidote is therefore not fanciful at all. The specific claim about Galanthus nivalis is another matter entirely.

6. Candidate Taxa Reassessed

6.1. Galanthus nivalis L.

Morphologically the fit is excellent: a bulb with dark brown outer tunics and a solitary pendulous milk-white perianth [26]. Pharmacologically it is excellent: galantamine is a selective, reversible, centrally penetrant AChE inhibitor with additional allosteric potentiation of nicotinic receptors, and the species yields a rich Amaryllidaceae alkaloid profile [27,28]. C1, C2 and C5 are therefore satisfied. The failure is ecological and biogeographical. Galanthus nivalis is a Euro-Siberian element whose Mediterranean occurrence is restricted to scattered humid refugia, largely montane and largely peripheral to the Aegean world in which the poem was composed and transmitted [29,30]. Plaitakis and Duvoisin turned this rarity into an argument in its favour, glossing bessa as “moist sheltered ground” and reading the plant’s inaccessibility as the referent of chalepon oryssein. Neither move survives inspection. The translation is tendentious; moist forest is atypical of the Mediterranean woodland biome; and, decisively, the text says the plant is hard to dig, not hard to find [6]. The distinction is not a quibble — it is the whole of C3. C3 therefore fails, since a small and shallowly seated bulb offers no resistance to extraction, and C4 is met only in habitats marginal within the species’ Mediterranean range. There is also a practical objection, which we have not seen raised: a remedy that must be fetched from a rare montane refugium makes a poor household antidote to what would have been a fairly common domestic poisoning.

6.2. Peganum harmala L.

Syrian rue fails C1 comprehensively: it is a woody, branching perennial with no tunicated bulb, and its ecology is steppic and ruderal rather than woodland, so C4 fails likewise. It should nonetheless be excluded for the right reasons. The claim that its β-carbolines would aggravate an antimuscarinic delirium, sometimes advanced in this debate, simply is not supported by the phytochemical literature: harmine and harmaline are themselves acetylcholinesterase inhibitors, with reported IC₅₀ values in the low micromolar range in Ellman assay — approximately 9 and 11 μM respectively, with harmane more potent still and selective over butyrylcholinesterase — and bioassay-guided fractionation of P. harmala seeds has yielded multiple potent AChE and BChE inhibitors [31,32,33]. Direct comparison of potency with galantamine requires data from a common assay and is not attempted here. Harmine crosses the blood–brain barrier and reverses scopolamine-induced memory impairment in rodents [34]. What makes the plant a poor antidote — that is, what qualifies rather than satisfies C5 — is not the absence of anticholinesterase activity but the pharmacological company it keeps: reversible MAO-A inhibition, serotonergic agonism, tremorgenic and convulsant activity at overlapping doses, and a narrow margin between the doses producing cholinesterase inhibition and those producing neurotoxicity. What the net central effect of a crude seed preparation would be in a delirious patient, nobody can say. The distinction is worth making, and not only for tidiness: it shows how a single-mechanism account of a plant’s pharmacology can produce a verdict that is confident, widely repeated, and wrong.

6.3. Allium nigrum L.

Broadleaf garlic has a strong claim to be the plant that Theophrastus and Dioscorides called moly — a bulbous geophyte with broad leaves resembling those of the sea onion Drimia maritima, used, Theophrastus reports, against spells and sorcery [4,6]. That is a claim about the Classical and Hellenistic phytonym, not about the Homeric plant, and the two must be kept apart. Theophrastus himself distinguishes them, noting that his moly is not difficult to dig up, contrary to Homer: the species therefore fails C3 on the explicit testimony of the earliest botanical source. It fails C5 as well, since Allium organosulfur chemistry offers no centrally acting anticholinesterase of comparable relevance. It is worth adding that Plaitakis and Duvoisin’s rejection of the Allium identification rested on reading Dioscorides’ leukoion as the modern genus Leucojum; but Dioscorides expressly states that leukoion flowers may be white, yellow or purple (De materia medica 3.123), which excludes an all-white genus and points instead to Brassicaceae such as Matthiola and Erysimum [6]. That particular argument of theirs should be withdrawn. The conclusion it was meant to support survives anyway, on the independent grounds of C3 and C5 — which is a reminder that a bad argument for a true conclusion is still a bad argument.

6.4. Mandragora Officinarum and the Solanaceae

Mandrake is a pharmacological inversion of the required profile: it fails C5 not by falling short of it but by lying on the opposite side of the equation, since its tropane alkaloids are the competitive muscarinic antagonists that produce the syndrome, and administering it as an antidote would deepen the intoxication. Its morphology also fails C1 and C2. Its correct place in this analysis is as the likeliest identifiable constituent of the pharmaka lugra, alongside Hyoscyamus — a reading reinforced by the Dioscoridean synonym circaea [13,15,35].

6.5. Pancratium spp.

The sea daffodils have received far less attention than they merit, and we think this is the central omission of the whole literature. Pancratium maritimum is abundant and continuously distributed along Mediterranean and Black Sea coastal dunes; it bears large white flowers and a deeply buried bulb with dark outer scales, satisfying C1 and C2; and its leaves closely resemble those of Drimia maritima and Allium nigrum, satisfying the comparison Theophrastus transmits [6,36]. Crucially, it satisfies C3 by a mechanism no other candidate offers: the contractile roots of Pancratium can draw the bulb to depths reported up to about 160 cm, making extraction genuinely arduous [37]. Late Enlightenment horticultural description already noted the large dark-skinned bulb with thick fibres striking deep into the ground [6]. As to C5, the genus is embedded in the Mediterranean-centred clade of Amaryllidoideae (Galantheae, Narcisseae, Pancratieae) in which AChE-inhibiting activity is reported from seven of nine genera, and a potent acetylcholinesterase inhibitor has been isolated from Pancratium illyricum L. [10,38,39].

The remaining objection is habitat: P. maritimum is a psammophile of the shore, whereas the encounter is inland, so C4 is not met by the commonest species of the genus. Here P. illyricum, a Tyrrhenian endemic of Corsica, Sardinia, Capraia and Elba, is instructive: it occurs from sea level to the montane belt including shaded woodland, its range lies within territories reached by Mycenaean trade, and both Theophrastus (HP 9.15.1) and Apollonius Rhodius (Argonautica 4.660) locate Circe in precisely this Tyrrhenian region [6,40]. Sea daffodils are moreover attested in pre-Homeric Aegean iconography, depicted on the frescoes of the House of the Ladies at Akrotiri and on an inlaid Mycenaean bronze blade [41,42] — evidence of familiarity, not of medicinal use, but familiarity is the precondition of empirical pharmacognosy.

7. Discussion

After our Narrative review and all researches behind we draw three conclusions as follow. The first is negative: the identification of moly with Galanthus nivalis should no longer be repeated as settled. It rests on a defensible pharmacological intuition supported by two philological arguments that do not survive examination and by a biogeographical premise that is close to inverted, since the species is rare and peripheral precisely in the region where the tradition was formed. The second is methodological. The debate has been structured throughout by the assumption that a single Linnaean binomial lies behind the phytonym. That assumption is anachronistic. Species-level recognition as a systematic practice postdates Theophrastus; earlier folk classification is organised by use and by gross resemblance, and ethnobotany documents many phytonyms that denote complexes of related taxa used interchangeably for shared properties [43,44]. Molina-Venegas and Verano’s proposal that moly names such a complex, a set of Mediterranean Amaryllidoideae with dark-tunicated bulbs, white flowers and anticholinesterase activity, converging under one name through centuries of oral transmission, accounts for the vagueness of the description better than any single-species hypothesis [6]. On this reading the question “which plant was moly?” is malformed, and the productive question is which clade the tradition tracked and which taxon most shaped its canonical image. The answer to the first is the Mediterranean clade of Amaryllidoideae; the answer to the second, on present evidence, is Pancratium. Two features of that clade make it an unusually good candidate for such a complex. It is morphologically distinctive in a way that supports reliable recognition in the field without systematic training: white-flowered geophytes with dark-tunicated bulbs are not numerous in the Mediterranean flora, and the combination is salient. And it is functionally homogeneous, anticholinesterase activity being reported across seven of its nine genera, so that a gatherer selecting on appearance alone would obtain the relevant pharmacology with high probability. A folk taxon defined by these two properties would be robust to the substitution of one member for another across regions and seasons — precisely the condition under which a complex, rather than a species, becomes the efficient unit of oral transmission. The third conclusion concerns what kind of knowledge is being claimed. It would be an overreading to attribute to Archaic Greece any concept of cholinergic neurotransmission, of receptor antagonism, or of enzymatic reversal. What can reasonably be attributed is an empirical association, stable enough across generations to be encoded in verse, between a recognisable class of bulbous plants and the resolution of a recognisable syndrome caused by another recognisable class of plants. That is a considerable epistemic achievement, and it is of the same type as the Caucasian folk use of snowdrop against paralysis that led Mashkovsky to galantamine [24]. The value of the moly episode for the history of pharmacology lies in this structural parallel, not in any claim to Bronze Age pharmacodynamics.

Two cautions temper the argument. The polysemy of pharmakon, poison, remedy and antidote at once, warns against imposing modern categorical boundaries on the vocabulary. And retrospective diagnosis from literary texts is a genre with a poor record: the Odyssey is not a case report, its author had no interest in nosological precision, and the correspondence between the narrated syndrome and the antimuscarinic toxidrome, however striking, is compatible with coincidence.

8. Limitations

This is a narrative review, not a systematic one, and the scoring in Table 1 is a heuristic device rather than an instrument. We present no laboratory work. Nobody, so far as we know, has characterised the anticholinesterase potency of crude bulb preparations of these taxa at doses and by routes plausible for Archaic Greek practice — and alkaloid content in the Amaryllidaceae varies enough with organ, season and provenance that such a characterisation would not be trivial to interpret even if it existed. The archaeobotanical record for the Aegean Bronze Age gives us nothing direct on medicinal use of any of the plants discussed here. And Aeaea is in all probability an invented island, so whatever we infer from its geography should be held loosely; we have tried to hold it loosely, but the reader should judge. Two lines of work would move the question forward. The first is laboratory work: comparative quantification of AChE inhibition in aqueous and ethanolic bulb extracts of Pancratium maritimum, P. illyricum, Galanthus nivalis and Leucojum aestivum, prepared as closely as possible to ancient practice and assayed under common conditions so that potencies can actually be compared. The second is archaeological: a systematic re-reading of Mycenaean and Minoan floral iconography, and of the Linear B tablets, for terms assignable to Amaryllidoideae geophytes. Neither is beyond reach.

9. Conclusions

The pharmacological reading of the moly episode, that the Odyssey preserves the oldest surviving literary trace of an anticholinesterase used against antimuscarinic poisoning, remains the most economical explanation of the passage. What requires revision is its canonical form. Galanthus nivalis is pharmacologically apt and biogeographically implausible. Peganum harmala is morphologically excluded, though not, as is usually said, pharmacologically antagonistic. Allium nigrum answers to the moly of the botanists rather than of the poet. Mandragora belongs on the other side of the equation altogether. That leaves the Mediterranean Amaryllidoideae as a clade, and Pancratium within it, satisfying every constraint the texts impose, including the difficulty of extraction, which four centuries of commentary have treated as poetic ornament and which Theophrastus, who was there first, treated as a diagnostic character. Our conclusion is therefore less a new identification than the abandonment of a wrong question. Moly was probably never one plant. It was the name a preliterate pharmacopoeia gave to a group of them, and the group is one we can still recognise.