The Lycurgus Cup

A late Roman drinking cup, 15.9 centimetres high and weighing 700 grams, cut so that a scene of the death of King Lycurgus stands almost free of the vessel wall, joined to it only by hidden bridges. It is the one complete surviving object made of Roman dichroic glass: lit from the front it is a dull jade green, lit from behind it turns translucent wine red. The colour change is caused by particles of a gold and silver alloy about seventy nanometres across, suspended in the glass at a few hundred parts per million.

The two states do not look like one object. They look like two, and that is the practical problem this record has to solve before it can show a reader anything: almost every photograph of the Lycurgus Cup in circulation is either the green cup or the red cup, and captions rarely say which. Every plate here says which lighting it shows.

Its documented history is short and almost entirely English. It surfaces in 1845 in the hands of a M. Dubois, is exhibited at the Society of Arts in London in 1850, and is with Baron Lionel Nathan de Rothschild by 1857. It stayed in the Rothschild family for about a century, was described in the literature as the Rothschild Lycurgus Cup, and was bought by the British Museum from Victor, 3rd Lord Rothschild in 1958. Where it was for the fourteen hundred years before 1845, and where it was found, are not recorded anywhere.

The cup is two colours and the glass is only one of them

Stand in front of the case in Room 41 with the gallery lights on the front of the cup and the Lycurgus Cup is an opaque, slightly muddy jade green, the colour of old bottle glass. Put a light behind it and the same vessel becomes translucent and wine red, so red that the thin passages between the vine stems glow like coals. Nothing has been added, removed or restored between the two states. The light has moved.

The cause is in the glass itself. It carries a colloidal suspension of gold and silver particles roughly seventy nanometres across, present in tiny quantity: the figures usually given, derived from analysis of the glass, are about 330 parts per million of silver and 40 parts per million of gold added to an otherwise conventional Roman glass batch. Particles at that scale scatter and absorb light selectively, so the light bounced back off the surface and the light pushed through the wall are different colours. The modern name for the mechanism is surface plasmon resonance.

The consequence for anyone using pictures of this object is immediate and severe. The green cup and the red cup photograph as two different vessels in two different materials, and a reader who has seen only one will not recognise the other. This is not a version problem of the ordinary kind, where two similar objects get confused. It is one object that produces two irreconcilable images of itself, and the discipline the case demands is exactly the discipline this publication applies to versions: say which is which, every time.

It is a cage cup, which is a different problem from a cameo vessel

The Lycurgus Cup belongs to a class of late Roman luxury glass called cage cups, or diatreta. In a cage cup the decoration does not sit on the wall of the vessel; it stands away from it. An outer shell of figures, or of an openwork mesh, is separated from the inner beaker by an air gap and connected to it only by short bridges, deliberately placed where the design will hide them. Hold a cage cup up and you can see daylight between the ornament and the cup.

That is a categorically harder thing to make than the cameo glass of the Portland Vase, which is a two-layer blank with the outer layer carved back. A cameo cutter removes an overlay. A cage-cup cutter has to remove the glass behind the ornament without breaking the ornament, working blind into an undercut, with no way to repair a slip. Both classes are cut with the same lapidary equipment, a rotating disc charged with abrasive slurry, and both were probably cut by men trained on hardstone. There the resemblance ends.

The two classes also do not overlap in time. Roman cameo glass is concentrated in the early first century. Cage cups are late: the dates in the literature run from about the middle of the third century to the middle of the fourth, with a small and contested group of much earlier vessels from graves in the Netherlands and Belgium that some researchers read as precursors.

Being the only one of something, stated precisely

The British Museum describes this cup as the only complete Roman object in dichroic glass and as the only figural example of a cage cup. Both claims are worth pinning down, because they are repeated everywhere in a looser form that is not true.

Dichroic Roman glass survives in other pieces, but in fragments. The Metropolitan Museum of Art holds a fragment of a cage cup, accession 26.60.95, catalogued as glass, dichroic, late third to fourth century, acquired with the Fletcher Fund in 1926, and it carries an ovolo frieze and part of a human figure in relief. So there was at least one other figural dichroic cage cup, and it did not survive whole.

The classification behind the figural claim is Donald Harden and Jocelyn Toynbee's. Publishing the cup in Archaeologia in 1959, they divided the known cage cups into two classes: Group A, with figured decoration, with or without an inscription in the network, and Group B, with network only. A Group A therefore existed as a category. What the Lycurgus Cup is, exactly, is the only one of that group that survives as a complete vessel with its whole figural scene intact.

Say it that way and the object gets more interesting rather than less. The point is not that the Romans made one of these. It is that they made an unknown number, of which we have one.

A king strangled by a vine, and why the subject is unusual

The scene running round the cup is the punishment of Lycurgus, king of the Thracian Edonians, who attacked Dionysos and his followers. In the version cut here Lycurgus has been caught and is being pulled down and bound by vine shoots that grow around his legs and body. The nymph Ambrosia, whom he attacked, is the vine: transformed into it, she is shown at smaller scale at his feet, one arm raised. Around him are Dionysos himself with the thyrsus, Pan, a satyr swinging a rock, and a panther, whose face is lost.

Two features of the cutting matter more than the story. First, the figures are almost entirely undercut, standing free of the wall except at hidden points, so that a viewer looking down through the rim sees the reverse of the relief as a set of hollows in an otherwise smooth interior. Second, the modelling exploits the dichroism: the thin, high-standing limbs and drapery read reddish while the thick body of the cup behind them reads green, so the figures separate from the ground in colour as well as in relief. Whether the cutter knew he was doing that or discovered it is not recoverable.

The subject is a warning. Lycurgus is the impious king who refused the god and was destroyed by him, and the cup shows the moment of destruction, not of resistance. A vessel for wine decorated with what happens to a man who refuses wine is not a subtle programme, but it is a coherent one, and it is the ordinary late antique use of the myth. The reader should be told plainly that there is nothing hidden here: the scene means what a fourth-century Roman would have said it meant.

The Licinius reading, and the grade it deserves

A political reading has been in circulation for decades. It proposes that the choice of a scene showing a bound and defeated king alludes to the defeat in 324 of the co-emperor Licinius by Constantine, and to Licinius's death in 325 after a period under close guard. On that reading the cup is a piece of imperial triumphalism disguised as myth, and its date is fixed to within a year or two.

This record grades that as argued and not documented. There is no inscription on the cup, no ancient text linking the object to the event, and no findspot to place it in a context that would support or refute the identification. The scene is a standard myth with an independent iconographic tradition; the resemblance between a defeated Thracian king in art and a defeated eastern emperor in politics is a resemblance, and a resemblance is not a source.

What can be said without argument is that the Lycurgus subject is uncommon, and that the cup was made in the century in which the event happened. Those two facts are what the reading is built on. They are not nothing, and they are not proof.

The mounts are modern, and the sources do not agree what they are made of

The cup does not stand on Roman glass. It stands on a metal foot, and its rim is bound by a metal band of leaf ornament. Neither is ancient.

The British Museum's own record describes both as silver-gilt: a silver-gilt band of leaf ornament at the rim, and a silver-gilt foot with openwork vine leaves. The Commons record of the photographs used in this entry, written from the museum's gallery text, gives gilded silver added probably in the late 1700s. The Wikipedia article gives gilt-bronze added about 1800. The metal is visibly white under the gilding in the plates below, which is consistent with silver rather than bronze, but this record follows the holding institution and notes that at least three published descriptions of the same two mounts are in circulation and they do not agree.

The date, wherever it is precisely placed, carries an inference. A late eighteenth or early nineteenth-century remounting of a Roman glass vessel usually means the object had been in an ecclesiastical treasury and was dispersed, and the suggestion in the literature is that the cup left a church treasury during the upheavals of the French Revolution. That is a plausible route, not a documented one, and no treasury inventory naming this cup has been produced.

1845, 1850, 1857: how the cup entered the record

The first notice of the cup is French and casual. In 1845 a French writer records seeing it in the hands of a M. Dubois. Nothing else about that transaction is recorded: not what was paid, not where Dubois had it from, not his first name.

Five years later it was in London and on public view. The British Museum's curatorial file records it exhibited at the Society of Arts in 1850, catalogued as number 626 in the Catalogue of Works of Ancient and Mediaeval Art exhibited at the Society's house, and illustrated in the Illustrated London News of 13 April 1850. By 1857 the art historian Gustav Friedrich Waagen had seen it in the collection of Baron Lionel Nathan de Rothschild, and in 1862 it was lent to the great special exhibition at South Kensington.

What that sequence gives is a firm terminus: the cup was above ground, in western Europe, and known to specialists from the 1840s. What it does not give is any information at all about the preceding fourteen centuries. There is no findspot for the Lycurgus Cup. Nobody knows what soil it came out of, or whether it came out of soil at all rather than passing hand to hand above ground from antiquity.

The Rothschild century, and the name in the literature

Baron Lionel Nathan de Rothschild, who died in 1879, is the first securely documented owner. The cup then remained in Rothschild ownership until 1958, when the British Museum bought it from Victor Rothschild, 3rd Baron Rothschild. The museum's acquisition note names those two men and no one in between; which members of the family held it across the intervening eighty years is not stated in the sources consulted here, and this record does not guess.

The length of that tenure is why the scholarly literature does not call the object the Lycurgus Cup at all. Harden and Toynbee's foundational paper of 1959 is titled The Rothschild Lycurgus Cup; so is Harden's own addenda and corrigenda in the Journal of Glass Studies in 1963, and so is the Chirnside and Proffitt analytical paper in the same volume. A researcher searching only for the popular name will miss the primary literature on the object, which is a small, avoidable and very common error.

There is also a rediscovery in this chain that is usually left out. The cup was in a Rothschild house, unstudied, until 1950; the Bonhams catalogue note for a related cage cup records that cage cups were first discussed as a group by Harden and Toynbee in 1959 only after the Lycurgus Cup was rediscovered in a home of Lord Rothschild in 1950. The modern study of an entire class of Roman glass therefore begins with somebody noticing what was on a shelf.

The purchase price in 1958 is given as 20,000 pounds, with 2,000 pounds contributed by the National Art Collections Fund, now the Art Fund. Those figures come from a tertiary source and this record flags them for replacement with the museum's own acquisition papers. Note also what the figure is and is not: a negotiated private purchase from a family trust, not a hammer price. The Lycurgus Cup has never been sold at public auction in any source consulted for this record.

The foot came off in 1958 and went back on in 1973

The condition line in the British Museum's record is one sentence long and worth reading slowly: the foot was removed from the cup by the Research Laboratory in 1958 and reunited in 1973, and was given its own registration number, 1958,1202.1.a, in 1958.

That is not a conservation accident. It is what happens when a museum acquires an object it intends to analyse. Taking the foot off gave the laboratory access to the base of the glass, and separating the modern mount from the ancient vessel in the register is the correct cataloguing decision, because they are two objects of different dates that happen to be screwed together. Fifteen years passed before the two were put back.

The analytical programme that followed is the reason this cup is famous outside archaeology. Chirnside and Proffitt published an analytical investigation in 1963, Robert Brill reported on the chemistry of the cup at the international glass congress in Brussels in 1965, and Barber and Freestone identified the colloidal gold and silver particles by analytical electron microscopy in Archaeometry in 1990. Freestone, Meeks, Sax and Higgitt published the summary account under the title The Lycurgus Cup - A Roman Nanotechnology in the Gold Bulletin in 2007. Two further conservation treatments are recorded, on 3 November 1999 and 9 March 2012.

The nanotechnology claim, stated exactly

The phrase Roman nanotechnology is the title of a paper, and it has been carried into headlines in a form the paper does not support. What the analytical work establishes is that the colour behaviour is caused by metal particles at the nanometre scale, and that those particles are an alloy of gold and silver. What it does not establish is that any Roman knew that, or could do it again on demand.

The concentrations involved are minute and the effect is exquisitely sensitive to them. The usual reading is that the Roman glassmakers did not target these quantities but arrived at them by dilution, most plausibly by reusing or diluting a batch that carried much higher concentrations of precious metal, and that the result was neither predictable nor repeatable. The survival of exactly one complete dichroic vessel, against a corpus of over fifty cage cups of which the rest are not dichroic, is itself evidence for that reading: a workshop with a reliable recipe would have produced more.

The cup has since been cited in the physical-sciences literature as a precedent for plasmonic sensing, and modern papers on the optics of the glass continue to appear. That is a genuine afterlife for a Roman drinking vessel, and it should be described as what it is: modern researchers finding a well-preserved natural experiment, not ancient chemists anticipating a modern field.

Drinking cup, or lamp

The British Museum catalogues the object as a drinking-cup and it is beaker-shaped, which in the cage-cup corpus is the form associated with drinking. Several beaker-form cage cups carry inscriptions cut into the network that are drinking toasts, and those inscriptions settle the question for that form: the Trivulzio cup in Milan reads BIBE VIVAS MVLTIS ANNIS, the Cologne cup from Koln-Braunsfeld carries the equivalent in Greek, and a cup from Pljevlja in Montenegro reads VIVAS PANELLENI BONA. The Lycurgus Cup has no inscription.

Against that, the alternative reading of the class is that some cage cups were hanging lamps. The Corning cage cup carries a copper-alloy collar attached to the vessel in antiquity, with a lamp-hanger of hooks and loops, which is physical evidence of suspension for at least one bowl-form example. Christian Steckner published on cage cups as light vessels in 1999.

For this cup specifically the lamp reading has an argument the others do not, and it comes out of the physics. The dichroism only performs when the light is behind the glass. A green cup on a table is an inert green cup; the red only appears when the source is inside or beyond the wall. Whitehouse asked in 1989 whether two ancient descriptions of dichroic glass survive, and the question of whether a Roman ever saw this vessel red is not idle. If it was ever filled with a lamp flame or held against one, its owner saw the effect. If it was only ever filled with wine and looked at across a table, he may not have.

This record does not resolve it. It notes that the shape argues for drinking, that the class contains at least one documented suspended example, and that the optical property of this particular cup is invisible in ordinary reflected light.

Cut, or cast: the argument this object cannot settle

The orthodox account of cage-cup manufacture was set out by Fritz Fremersdorf in 1956 and has stood since: a thick-walled blank was made, and the cage, the struts and the figures were then cut and ground out of the solid, an enormously slow process with no recovery from an error. David Whitehouse's posthumous corpus of 2015 argues that position throughout.

The minority position is Rosemarie Lierke's. She has held since the 1990s that these vessels were not cut from a solid blank at all but formed in moulds, and she published specifically on the making of the diatreta and on the Lycurgus Cup itself in 2013. Reviewing Whitehouse's corpus for the Journal of Roman Archaeology in 2020, Sylvia Fuenfschilling set out the features on original vessels that in her view cutting does not explain: bubbles that have not been cut open but are horizontally elongated; the wall of the inner cup running parallel to the inner wall of the mesh; a completely smooth inner surface to the mesh; struts with rounded ends that do not reach the outer cage; a spiral structure visible inside the struts under a microscope; the uneven ledge below the rim; and the sheer quantity of glass that the cutting hypothesis requires to be thrown away.

Her methodological point is the one that travels. Modern replicas of cage cups have been cut successfully by George Scott and by Josef Welzel, and that is regularly produced as proof of the cutting hypothesis. Fuenfschilling's answer is that a successful replica shows a method was possible, not that it was used, and that modern tools, modern glass and modern habits of thought are not the ancient ones. Exactly the same objection applies to the experimental reconstruction of the Portland Vase.

Two of the plates in this record show the inside of the Lycurgus Cup through the rim. The interior wall is smooth and the figures read from within as shallow hollows. That observation is evidence in the argument; it is not a verdict, and this record does not pretend to give one.

Why the pictures in this record disagree with each other

Every photograph below is of this cup, in the British Museum, taken in Room 41 on the same day by the same photographer, and released under CC0. Put side by side they look like a green cup, a red cup, and a brown cup with green figures. All three are the same object.

The green plates are lit from in front; the red plate is lit from behind; the intermediate plates are lit obliquely, so that the thick body reads green and brown while the thin undercut passages transmit and glow red. That gradation is the most useful thing a reader can be shown about this object, because it demonstrates that the two famous states are the ends of a continuum and not two settings of a switch.

Anyone cataloguing this cup, or buying a photograph of it, should record the lighting. A file captioned only Lycurgus Cup is an incomplete record of what the picture shows.