Marius Comper RO

British Museum 1958,1202.1 · Roman, 4th century AD

Why the Lycurgus Cup turns from green to red

Lit from the front, this Roman glass cup is a dull jade green. Put a light inside or behind it and it glows red. The colour comes from specks of metal typically about 70 nanometres across. Move the lamp, change the metal, and watch the calculation decide the colour.

front-litlit from behind
Try
The cup, simplified, drawn from the calculation. Drag the lamp round it, or use the slider.
Light passing through the glass, by wavelength. Line: the calculation. Dots: Brill’s 1965 measurement of a chip of the cup. Shaded: the comparison window used here (495–516 nm).

A cup with two colours

The British Museum dates the cup to the 4th century AD; Rome and Alexandria have both been proposed as where it was made. It is a cage cup: the glass was cut and ground back so that the figures stand proud of the wall, some joined to it only by small bridges. They show the Thracian king Lycurgus, who attacked the god Dionysus, trapped by Ambrosia, a follower of the god turned into a vine. The cup has been in the British Museum since 1958.

In daylight from the front it looks opaque and greenish. With a light inside or behind it, it turns a deep red, with a touch of purple. The museum describes it as the only complete example of this kind of glass; other pieces survive as fragments.

Specks of metal

In 1965 Robert Brill of the Corning Museum of Glass reported his study of a chip that had come away from the cup’s broken foot. The glass held about 300 parts per million of silver and 40 of gold, by weight. He also measured, on a polished face about 2 square millimetres in area, how much light of each colour passed through and how much came back.

In 1990 David Barber and Ian Freestone looked at fragments of the cup under a transmission electron microscope. They found the metal as tiny particles, 50 to 100 nanometres across and typically about 70: an alloy of 66.2 % silver, 31.2 % gold and 2.6 % copper, counted by atoms.

The two ratios describe different things. In the glass as a whole there are about 14 silver atoms for each gold atom; in the particles, about two. The particles are rich in gold; Barber and Freestone link this to gold being easier to bring out of the glass as metal.

The calculation

Light drives the conduction electrons of a small metal particle to oscillate together; near a resonance the particle absorbs and scatters light especially strongly. In 1908 Gustav Mie wrote down exactly how a uniform sphere of any size, in a uniform surrounding, scatters and absorbs light. The page runs his solution, in the form Craig Bohren and Donald Huffman published in 1983, for a sphere of the measured alloy in glass of refractive index 1.5. The metal’s optical constants are the measurements of Peter Johnson and Robert Christy (1972); for the alloy, the three metals’ permittivities are averaged by their share of atoms, which is an approximation.

For particles 70 nm across, least light gets through at 506 nm, in the blue-green. Brill’s curve for the real glass has a broad dip with its bottom near 505 nm. The comparison window used here, 495 to 516 nm, brackets that flat bottom with a margin for reading the graph. Gold particles of any size from 50 to 100 nm, in glass of index 1.50 to 1.55, put the dip at 548 nm or redder: the dip rules gold out. Silver does not rule itself out so cleanly: silver particles about 84 nm across reach 494 nm, and in slightly denser glass they enter the window. Changed one at a time, the glass index (1.50 to 1.55), a spread of sizes from 50 to 100 nm, a different way of mixing the metals, or more copper move the alloy’s dip only between 506 and 516 nm; the most extreme combination tested reaches 520 nm.

CaseTransmission minimum
measured alloy, 70 nm, glass index 1.50506 nm
glass index 1.52510 nm
glass index 1.55516 nm
sizes spread over 60–80 nm506 nm
sizes spread over 50–100 nm514 nm
refractive indices mixed instead514 nm
no copper506 nm
10 % copper, same silver-to-gold ratio510 nm
the 50–100 nm spread in glass of index 1.55520 nm
pure gold, 50–100 nm, index 1.50–1.55≥ 548 nm
pure silver, 50–100 nm, index 1.50up to 494 nm

With the blue-green weakened, red light gets through most readily, with some violet adding a purple tint: the cup’s glow. What the specks scatter back towards the viewer is greenish: the colour seen in front light.

What the calculation does not explain

How dark the glass is depends on how much metal the light meets on its way through. That cannot be worked out from the analyses, so it is set so that the measured alloy lets 4.2 % through at the minimum, as Brill measured; the glass surfaces are taken to pass 92 %. The brightness of the front-lit colour is scaled to his measured reflectance, about 14 %. These settings also affect the hue a little. The agreement is mainly in where the dip falls: towards the red the calculated glass lets through more than Brill measured (about 58 % against 44 % at 615 nm).

Brill’s reflectance peaked around 540 to 550 nm. The light the calculated particles send back peaks at 514 nm. Real glass reflects from its surface as well, and light scattered inside it is filtered on the way out; neither is modelled here. Nor are the iron in the glass, which tints it, or particles that are not perfect spheres.

The copper figure varies between accounts. Barber and Freestone measured 2.6 % of the atoms; a 2007 summary says about 10 % copper without giving a basis. With 10 % copper and the same silver-to-gold ratio, the dip moves from 506 to 510 nm.

Accident or recipe

Nothing shows that the glassmakers knew what made the colour, and the Roman recipe is unknown. Brill’s laboratory reproduced the effect in melts of 1.2 kg. In most of his experimental glasses the colour appeared when the glass was reheated, between about 580 and 1000 °C, and almost completely bleached above about 1050 °C, when the metal redissolved. Composition, time, temperature and the furnace atmosphere all mattered; why the effect is so rare in Roman glass is still open.

The sky’s blue is also scattered light, but from air molecules, far smaller than a wavelength; their scattering grows smoothly stronger towards shorter wavelengths. Here the colour comes from a resonance in the metal, which weakens one range of wavelengths most.

Share your glass

The link keeps the lamp, the metal and the size you chose.

Sources

Every figure on this page is traced in the project’s source notes. The cup is a simplified drawing coloured by the calculation; no museum photograph is used.

  1. Robert H. Brill, “The chemistry of the Lycurgus Cup”, Comptes rendus du VIIe Congrès international du verre, Brussels, 1965, vol. II, paper 223 (Figs. 3–4, Table 1); Rakow Research Library scan, Corning Museum of Glass, CC BY-NC-SA 4.0. The transmission points on this page are redrawn from Fig. 3.
  2. David J. Barber and Ian C. Freestone, “An investigation of the origin of the colour of the Lycurgus Cup by analytical transmission electron microscopy”, Archaeometry 32 (1), 1990, 33–45, doi:10.1111/j.1475-4754.1990.tb01079.x.
  3. Ian Freestone, Nigel Meeks, Margaret Sax and Catherine Higgitt, “The Lycurgus Cup — A Roman nanotechnology”, Gold Bulletin 40 (4), 2007, 270–277, doi:10.1007/BF03215599.
  4. P. B. Johnson and R. W. Christy, “Optical constants of the noble metals”, Physical Review B 6, 1972, 4370–4379, doi:10.1103/PhysRevB.6.4370; data from refractiveindex.info (CC0).
  5. Craig F. Bohren and Donald R. Huffman, Absorption and Scattering of Light by Small Particles, Wiley, 1983 (the BHMIE algorithm); Gustav Mie, Annalen der Physik 330, 1908, 377–445.
  6. Chris Wyman, Peter-Pike Sloan and Peter Shirley, “Simple analytic approximations to the CIE XYZ color matching functions”, JCGT 2 (2), 2013, jcgt.org (the displayed colours).
  7. British Museum, 1958,1202.1 (object record).