The Polariscope shows what a gem polariscope shows when one of our seven gemstones turns between two polarizing filters. You see if the stone is dark or light, and if it blinks or not. Through the conoscope, you see its interference figure and optic sign. The page calculates everything by Jones calculus across the visible spectrum, from the refractive indices and measured absorption of the stone. The same CIE colorimetry as in the Dichroscope turns the result into color. It is a teaching model of the instrument. It does not measure or identify any real stone.
The instrument
A polariscope is a light under a polarizing filter (the polarizer), a stage for the stone, and a second filter above it (the analyzer). Crossed, the two filters pass no light at all. The polarizer leaves the light vibrating one way, and the analyzer passes only the vibration at right angles to it.
A singly refractive stone passes the light on unchanged, so it stays dark in every position. A doubly refractive stone splits the light into two rays that vibrate at right angles, and the analyzer partly passes them. So the stone lights up, except at four positions per turn, where its vibration directions line up with the filters. With the filters parallel (the open position), the field is light instead.
A strain-free glass sphere (the conoscope) set on the stone gathers light from many directions at once. So the eyepiece shows an interference figure, in which each point is one direction through the stone. A first-order red plate (the compensator) slid in above the stone shows which of the two rays is the slower. That is the optic sign of the stone.
The model
- The set-up
- Daylight (CIE D65) passes an ideal polarizer and then vibrates up and down the view. Then come the stone; in the conoscope, the red plate if it is in; then the analyzer. The analyzer is across the view when crossed, and up and down when parallel. The white of the page is the open polariscope with no stone.
- Directions
- In the frame of the eyepiece, x is to the right, y up and z toward the eye. The optic axis, tilted θ from the line of sight and turned φ clockwise, is c = (sin θ sin φ, sin θ cos φ, cos θ). At φ = 0, its shadow lies along the polarizer. For light that goes along k, the o-ray (ordinary) vibrates along c × k, and the e-ray (extraordinary) along k × o, at an angle χ in the field.
- The indices
- At an angle A to the optic axis, the index of the e-ray is
ne(A) = (cos2A / no2 + sin2A / ne2)−½,
so over a path L the two rays differ by (ne(A) − no) L, a retardance γ = 2π (ne(A) − no) L / λ. Straight down the axis (A = 0), the difference is zero. The indices are those of the 3D Gemstone Simulator, the same numbers that the Refractometer shows. Their difference is the birefringence (case P4). - Absorption
- As in the Dichroscope, the o-ray takes αo and the e-ray takes αo cos2A + αe sin2A, from the shared spectral registry. So a ruby glows red, not white.
- Jones calculus
- At each of 81 wavelengths, 380 to 780 nm, the stone is an absorber (amplitudes
exp(−αL/2) along o and e) followed by a retarder
R = cos(Δ/2) I − i sin(Δ/2) (n · σ),
Δ = √(γ2 + κ2) and n = (γ cos 2χ, γ sin 2χ, κ) / Δ on the Pauli matrices. This is a plain linear retarder when κ = 0. The red plate is a linear retarder of 550 nm, with its slow direction at 45°, northeast to southwest. The light that reaches the eye is |A · Plate · R · D · P|2. Between crossed polarizers, without absorption or plate, this is the familiar I = sin22φ sin2(γ/2). - Quartz
- Amethyst is quartz. Quartz also turns the plane of vibration of the light as it runs along the optic axis (optical activity). It turns it by ρ(λ) = 7.19 λ2 / (λ2 − 0.09262832)2 degrees per mm (λ in µm), 21.8° per mm at 589 nm (Chandrasekhar 1957). Off the axis, we take ρ cos2A: the circular retardance is κ = 2ρ cos2A L. This is why a slab of quartz cut across its axis stays light, in color, between crossed polarizers. It is also why its figure is a bull’s-eye with no cross at the center.
High-order white
A gem a few millimeters thick puts its two rays many wavelengths apart. For example, a ruby across its axis, with 8 × 10−3 in birefringence over 5.8 mm, puts them about 46 µm apart, some 80 wavelengths of green light. Across a band only 5 nm wide, the phase then turns several times. If we sampled it at one wavelength per band, we would invent colors that are not there. The intensity is exactly a + b cos Δ + c sin Δ in the phase of the stone. So the model takes the exact average of the band, a + (b cos Δ + c sin Δ) sinc(w/2), where w is the spread of the phase across the band (and, in the conoscope, across one pixel).
The result is high-order white tinted by the stone’s own color, as a real polariscope shows. Interference colors come only from small path differences, close to the optic axis or in the conoscope. Case P8 checks the bands against plain Jones matrices every 0.1 nm.
The path through the stone
The light takes one straight path, through the middle of the stone you choose, along the line of sight. We measure its length on the simulator’s model of that shape at its real size. When you look straight down the optic axis (the normal of the table, as in the simulator), the length is the depth of the stone. This is true when the culet sits under the middle of the table, as in a round, a square or a cube (5.03 mm for the T73, case P11).
A pear, heart or triangle, whose culet sits off the middle, gives a little less (3.45 mm against a depth of 4.11 for the pear). Tilted, the length is the thickness of the stone through its middle along the new line of sight. In the conoscope, a direction at an angle inside the stone runs a longer path, as through a slab of the same thickness.
The conoscope and the red plate
Each point of the figure, at a distance r from the center of the view, is a direction at sin ψ = r sin 40° in air, refracted into the stone. The 40° cone is a working choice, not a measured sphere. The center of the figure (the melatope) is the optic axis. When the stone tilts, the melatope moves off center. Past 14° (moissanite) to 25° (amethyst), it leaves the view.
The GPU (WebGL2) draws the figure by the same equations as above, line for line. Without WebGL2, the page calculates it on the CPU at a lower resolution. The two agree within two of 256 levels at every point we checked, across seven stones and settings.
With the red plate in, the page reads the sign as a gemologist does. Close to the center, the path difference of the stone is about 200 nm. There, the page compares the northeast and northwest quadrants. Where the slow ray of the stone lines up with the slow direction of the plate (northeast to southwest), the two add, and the color climbs to blue. Where they cross, the color falls to yellow.
Blue in the northeast means the slow ray is the e-ray, which vibrates out from the center: positive. Blue in the northwest means it is the o-ray: negative. The page reads this from the interference colors alone, without the stone’s own color. It does this because a deep color, such as the color of emerald, can all but hide them on screen (case P5).
The readings
The bold reading describes a full turn, calculated at every 5°. Between crossed polarizers, the stone “stays dark” when its lightest position stays under Y = 0.003 (the open polariscope is 1). It “blinks” when its darkest position is under half its lightest. A singly refractive stone stays dark at every tilt. A doubly refractive stone stays dark straight down its optic axis. That reading tells you to tilt the stone and look again.
Rarely, a doubly refractive stone also stays dark a few degrees off the axis, in a deeply colored stone (emerald tilted about 7°). There, its two rays come out a whole wavelength apart in the only colors that it lets through, and the page names this case as such. Quartz close to its axis turns the light more than it splits it, so it stays light and does not blink. These thresholds are our own choice.
Strain, illustrative
Many diamonds grew under stress, which leaves them weakly doubly refractive in patches (anomalous double refraction). Between crossed polarizers, a strained diamond shows patchy gray light instead of staying dark. With Strain on, the diamond carries two smooth random fields fixed to the stone. One is a small path difference γ, up to 140 nm. The other is the direction χ of the slower ray, which wanders through every angle. Each point is a thin linear retarder. So between crossed polarizers it passes sin22(χ + φ) sin2(πγ / λ) of the light, and when the polarizers are parallel, one minus that.
Each patch goes dark at its own turn, so the stone as a whole never does. Over a turn, its average light stays between 0.020 and 0.023 of the open polariscope. The same strain pointing one way would blink like a doubly refractive stone, from 0.002 to 0.35 (case P12). That is what the reading teaches: patchy light that never goes dark all at once is strain, not double refraction. The pattern, its size and its strength are invented to show the effect, not measured from any stone. The page labels it so.
What the model leaves out
- A real faceted stone mixes many paths, reflected inside it, so its light between crossed polarizers is patchy. Here the light takes one straight path, so the whole stone shows one color.
- Strain in any stone except diamond, and any measured strain at all. The diamond’s pattern is invented to show the effect.
- Aggregates, which stay light at every turn, and biaxial stones (two optic axes, a different figure). None of the seven is one.
- The change of the birefringence with wavelength (its dispersion), the Fresnel losses at the surfaces, and the walk-off of the e-ray. They change the light by a few percent, not what you see.
- Twinning in amethyst, which can turn its bull’s-eye into an Airy spiral. The figure here is of one untwinned crystal.
- Real polarizers, which pass a little light when crossed and tint it slightly.
- Your screen. Most phone and laptop screens give out polarized light. So a real polariscope held against one shows the light of the screen, not the light of the stone.
Validation
These cases run on the page’s own data each time we build the site. The build stops if one fails.
| Case | What it checks | Result | Needs | Outcome |
|---|---|---|---|---|
| P1 | Singly refractive stones (diamond, cubic zirconia): no light between crossed polarizers at any turn or tilt | largest Y 0 | 0 | Pass |
| P2 | Uniaxial stones go dark at 0° and 90° and are lightest at 45° (across the axis and at 45° to it); quartz, whose optical activity leaves a trace of light off the axis, within a ten-thousandth | at 0° and 90°: 1.5e-32 of the lightest (quartz 7.3e-5); at 45°: 0 short of it | under 1e-9 (quartz 1e-4) and 1e-9 | Pass |
| P3 | Down the optic axis the birefringence is zero, so the stone stays dark (quartz excepted, P7) | largest index difference 0, largest Y 0 | 0 and 0 | Pass |
| P4 | Birefringence and optic sign as the simulator and the Refractometer give them | Diamond 0, Moissanite 0.043 (+), Cubic Zirconia 0, Ruby 0.008 (-), Sapphire 0.008 (-), Emerald 0.006 (-), Amethyst 0.009 (+) | all equal | Pass |
| P5 | The red plate reads every uniaxial stone's optic sign from the figure's colors (blue quadrants), not from the data | Moissanite + (b* NE -4, NW 38); Ruby - (b* NE 35, NW 0); Sapphire - (b* NE 35, NW 0); Emerald - (b* NE 35, NW 0); Amethyst + (b* NE -28, NW -17) | each as its indices say | Pass |
| P6 | High-order white: across the axis at 45° a stone shows its own color at half its light, no invented colors | chromaticity off by 0.002, light 0.501 at most | under 0.005 and 0.5 +/- 0.01 | Pass |
| P7 | Quartz (amethyst) turns the light along its axis: 21.7°/mm at 589 nm, so down the axis it stays light | 21.77°/mm; lightest Y 0.284, darkest 0.284 | about 21.7, and not dark | Pass |
| P8 | The 5 nm band averages against plain Jones matrices at every 0.1 nm | Ruby across the axis ΔE00 0.09; Moissanite 3° off the axis ΔE00 0.02; Emerald 20° off the axis ΔE00 0.13 | each under 1 | Pass |
| P9 | Nothing lost or made: for a colorless stone, crossed plus parallel is the open polariscope | largest miss 1.2e-15 | under 1e-9 | Pass |
| P10 | The conoscope down the axis: dark at the center, the figure the same turned 90° | center Y 0, largest difference 1.1e-16 | 0 and under 1e-9 | Pass |
| P11 | The path through the stone looking down the optic axis is the shape's depth (T73, hexahedron) | T73 5.029 mm (depth 5.03), hexahedron 5.774 mm (5.77) | within 0.02 mm | Pass |
| P12 | Strain (illustrative, diamond): patchy light that never goes dark all at once, because its slow ray points every way; the same strain pointing one way would blink | over a turn Y 0.020 to 0.023; one way 0.002 to 0.349; reads "strain" | darkest over half the lightest, lit; one way under a tenth | Pass |
Sources
We are not affiliated with any of these publishers.
- R.C. Jones, “A new calculus for the treatment of optical systems”, Journal of the Optical Society of America 31, 488 to 493, 1941.
- S. Chandrasekhar, “The optical rotatory dispersion of quartz”, Proceedings of the Indian Academy of Sciences, 1957; T.M. Lowry and W.R.C. Coode-Adams, Philosophical Transactions of the Royal Society, 1927.
- G.R. Rossman, Mineral Spectroscopy Server, California Institute of Technology, and E.V. Dubinsky, J. Stone-Sundberg and J.L. Emmett, Gems & Gemology 56(1), 2020: the absorption spectra, as on the Dichroscope page.
- CIE, “Colour-matching functions of CIE 1931 standard colorimetric observer” (CIE 018:2019, DOI 10.25039/CIE.DS.xvudnb9b) and “CIE standard illuminant D65” (DOI 10.25039/CIE.DS.hjfjmt59).
- IEC 61966-2-1:1999, the sRGB color space.