Specific gravity (SG) is the ratio of a gemstone's mass to the mass of an equal volume of water. It is a fixed physical constant — like refractive index, it does not change with color, cut, lighting condition, or most treatments. (Surface coatings such as wax, resin, or oil create a small downward bias; this is noted below.) Combined with RI, SG identifies most gemstones to species level without any further testing.

The practical value becomes clear with species that share overlapping RI ranges. A red spinel and a ruby can produce similar refractometer readings in the 1.76x range. Specific gravity immediately separates them: ruby (corundum) runs 3.97–4.05, while spinel runs 3.57–3.63 — a gap of about 0.35. This is well within what a simple hydrostatic weighing setup can resolve, even with a consumer jeweler's scale.

Specific gravity is the measurement that needs no optics: weigh the stone in air, weigh it suspended in water, divide. In Loupewise's own engine, a single SG reading narrows the 132 entries in our reference table — 69 distinct species — to a median of 24 candidates and settles 4 outright, on a jeweler's scale and a glass of water. It also works on rough and on stones with no polished facet, exactly where a refractometer gives up. These figures simulate a perfect reading; a real hydrostatic run carries the error sources this page teaches you to control.

The Hydrostatic Weighing Formula

Archimedes' principle states that a submerged object displaces its own volume of water. The weight lost in water equals the weight of water displaced. SG is simply the ratio:

SG = Weight in air Weight in air − Weight in water

Worked example:
Stone weighs 2.50 ct in air · Weighs 1.88 ct suspended in water
SG = 2.50 ÷ (2.50 − 1.88) = 2.50 ÷ 0.62 = 4.03
Consistent with ruby (3.97–4.05) — and with almandine garnet, which is why RI or the polariscope confirms the call.

The denominator (W_air − W_water) is the buoyancy force — the weight of water displaced by the stone. The smaller this number, the lighter the stone relative to water, meaning lower SG. A stone with SG exactly 1.00 would weigh zero in water.

What You Need

A precision balance — and how fine it must be depends on your stone. There is no single answer, because the error in a hydrostatic reading is δSG = SG² × d ÷ A: the balance's resolution d, divided by the stone's weight A, times the square of the density you are measuring. Halving the stone doubles the error. An inexpensive jeweler's scale reading to 0.01 ct gives about ±0.03 on a five-carat ruby and ±0.32 on a half-carat one — four times the whole width of the ruby range. The chart below shows what each resolution actually buys.

Distilled water, or water that has been boiled and cooled. Not for its density — hard tap water is denser by roughly 0.0003, which is worth about 0.001 on an SG-4 stone and matters to nobody. The reason is dissolved air. Fresh tap water carries a great deal of it, and it comes out of solution as bubbles on the stone and the cage, which is the largest single error on this page. Temperature is a red herring for the same reason: water at 21°C is 0.998 against 1.000, worth 0.008 on an SG-4 stone, and Anderson says outright that the correction can be disregarded.

Hydrostatic weighing bridge. A metal frame that spans the scale pan and supports a beaker of water above the pan. Commercial versions are available from gem supply houses (Kassoy, Stuller). A DIY version works: suspend the stone over the beaker using wire looped through the scale's hook or pan edge, or use a retort stand and wire to hold the stone in the water while the wire connects down to the scale.

Wire — stiff for the cage, fine for the suspension. These are two different jobs and they want opposite things. The coil that holds the stone should be stiff enough to keep its shape: Anderson specifies copper of about 1–1.5 mm. The wire that carries it down through the water surface should be as fine as will bear the load, because surface tension drags on it exactly where it cuts the surface, and that drag is a real source of error on small stones. One compromise wire of about 0.5 mm brass does both adequately. Use metal, not thread — the basket must not absorb water.

Forceps or stone holder. To place and position the stone without contaminating it with skin oils, which can trap a small air layer and reduce the apparent SG reading.

Which Rig You Have Decides Which Formula Is Right

Two bench setups are sold and both are correct. In the commoner one the gallows carrying the basket stands on the scale pan, and the beaker is held up separately so it never touches the scale. The stone is buoyed upward, the reading falls, and the number you read is the stone's weight in water — so SG = A ÷ (A − W), the formula this page uses throughout.

In the other, the beaker stands on the pan and the basket hangs from an independent stand. Now the stone's buoyancy pushes down on the water, the reading rises, and the number you read is the loss of weight — so SG = A ÷ R, with no subtraction at all.

Applying one rig's formula to the other's reading does not announce itself. It returns a number between 1 and 2 for every gem denser than water twice over — the amber and opal band, which looks like a perfectly real result — and it inverts the order of a parcel, so the densest stone reports the lowest figure. One question tells the two apart: does the beaker of water sit on the scale pan?

Step-by-Step Measurement

1

Weigh the stone in air (W_air)

Clean the stone with a lint-free cloth to remove oils. Set it directly on the scale pan. Record the weight in carats or grams to the full precision of your scale. This is W_air. Do not round — you will subtract from this value, so rounding errors compound.

2

Set up the hydrostatic bridge

Position the bridge over your scale so the beaker of water sits above the scale pan and the scale can still register weight. The stone will hang from wire through the water and ultimately connect to the scale. Commercial bridges usually clamp to the scale body or the bench surface with the beaker elevated above the scale pan.

3

Suspend the stone in water

Loop wire around the stone's girdle, or use a small wire basket for round or irregular stones. Lower the stone into the beaker of distilled water, ensuring it is fully submerged and not touching the beaker walls or bottom. Any contact will transfer force to the beaker rather than the scale, throwing the reading off. Check that no air bubbles cling to the stone surface — they add buoyancy and lower the apparent SG. Brush gently with a fine brush if needed.

4

Tare the scale, then weigh in water (W_water)

With the bridge, beaker, and wire in place but the stone removed, press TARE (zero) on your scale. Now attach the stone and lower it fully into the water. The scale now reads only the apparent weight of the stone in water. Record this value as W_water. It will be less than W_air — that difference is the buoyancy.

5

Calculate SG

SG = W_air ÷ (W_air − W_water). If W_air = 2.50 ct and W_water = 1.88 ct, then SG = 2.50 ÷ 0.62 = 4.03. A result ending in more than two significant decimal places is false precision for most bench setups — report to two decimal places unless your scale resolution warrants more.

6

Cross-reference the result

Enter your SG value in Loupewise alongside your RI reading. The engine filters all 132 gem species against both values simultaneously. Alternatively, consult the SG table below. If your result falls at the boundary of two species' ranges, look at the optic character from your refractometer reading — isometric (SR) vs. non-isometric (DR) often resolves the ambiguity immediately.

Accuracy Tips

Use distilled water

Hard tap water (high mineral content) is measurably denser than distilled. Use distilled water at room temperature (~23°C) for consistent results.

Prove the rig on quartz first

Anderson's own check, and this page has never carried it: run 30–40 carats of clean quartz before you trust the rig on an unknown stone. Pure quartz has an invariable density of 2.651. If it does not come back at 2.651 the fault is in your water, your tare or your technique — and you have found it on a stone whose answer you already knew, instead of on the one you are trying to identify.

Run it three times

One run cannot tell a bubble from a genuinely low stone, because both read low. Three runs can: bubbles shift between weighings, so a bubble error disagrees with itself, while a real low reading repeats. Dry the stone before each air weighing. If your three runs do not agree, the disagreement is the finding — do not average it away.

Eliminate air bubbles

Air bubbles clinging to the stone or the cage add buoyancy and reduce the apparent SG. The sourced routine is three steps, not one: immerse the empty cage in the beaker before putting it on the balance; wet and squeeze the brush itself first, so it is not carrying air between its own hairs; then brush the cage under water until no bubbles show. Wet the stone by dipping and rubbing it with your fingers before it goes in, and it should arrive clean.

Tare out bridge and wire

Tare the scale with everything except the stone in place. Recalibrate after any shift in the bridge position or wire geometry.

Small stones: fight surface tension

Anderson's advisory: hydrostatic results are less accurate for small stones, and below about 3 carats water's surface tension drags on the suspension wire and breeds tenacious bubbles. His remedies, in order: the finest wire that bears the load, a single pinhead-sized drop of detergent in the water, and for best results a lower-surface-tension fluid — with the result multiplied by that fluid's density.

Watch for surface treatments

Wax, resin, oil fracture filling, and polymer impregnation all lower the apparent SG. The direction is always downward: a fracture-filled emerald reads below its true value.

Tare with the cage immersed

Zero the scale with the bridge, wire and cage already sitting in the water, not dry — a cage tared dry still gets its own buoyancy subtracted from the stone's, and the SG reads low. Skipping the cage weight entirely collapses the arithmetic to nonsense.

Dry stone in air, porous stones with care

A wet stone weighed in air carries water weight, and the SG reads low. The opposite trap: porous material — turquoise, opal — soaks up water during immersion, and the SG reads high. Work briskly; a hydrostatic run is not a place for a long pause.

No touching the beaker

The stone and cage must hang free of the beaker. Resting on the bottom transfers part of the weight to the glass and the SG reads low.

Specific Gravity Reference — 20 Gem Species

These ranges reflect natural, untreated specimens. Synthetic counterparts typically fall within the same range unless noted. Surface coatings reduce measured SG; heavily included stones (e.g., emerald with extensive fracture filling) may read low.

Gem Species	SG Range	Crystal System	Common Confusion
Ruby (corundum)	3.97–4.05	Trigonal	Spinel (3.57–3.63), garnet
Sapphire (corundum)	3.95–4.03	Trigonal	Tanzanite (3.35), synthetic sapphire (same SG)
Spinel	3.57–3.63	Isometric	Ruby — SG gap of ~0.35 immediately separates
Tsavorite Garnet	3.59–3.65	Isometric	Demantoid (3.82–3.88), spinel
Pyrope Garnet	3.62–3.87	Isometric	Almandine — SG overlaps at high pyrope/low almandine
Almandine Garnet	3.93–4.30	Isometric	Ruby — RI (garnet SR, corundum DR) separates immediately
Rhodolite Garnet	3.74–3.94	Isometric	Pyrope/almandine blend — intermediate SG
Emerald (beryl)	2.67–2.78	Hexagonal	Treated emerald reads low (~2.60); green glass reads ~2.50
Aquamarine (beryl)	2.68–2.78	Hexagonal	Same species as emerald; RI and SG overlap
Tanzanite (zoisite)	3.35	Orthorhombic	Strong trichroism and DR are the confirming tests
Tourmaline	3.01–3.26	Trigonal	Wide SG range due to compositional variation; RI required
Amethyst / Quartz	2.65	Trigonal	Consistent SG regardless of color; very low for a gem
Topaz	3.49–3.57	Orthorhombic	Overlaps spinel SG; RI and biaxial optic character confirm
Alexandrite (chrysoberyl)	3.70–3.78	Orthorhombic	Color change is visual; SG + RI are the physical confirmation
Diamond	3.52	Isometric	RI 2.42 (off-scale); thermal conductivity probe standard test
Moissanite	3.22	Hexagonal	SG vs. diamond (3.52) clearly separates; DR vs. diamond SR
Cubic Zirconia (CZ)	5.60–6.00	Isometric (synthetic)	SG far above all natural gems — unmistakable by weight alone
Zircon	4.60–4.80 (low-type: 3.90–4.10)	Tetragonal	Wide range due to metamictization; low-type zircon reads ~3.90
Peridot	3.27–3.48	Orthorhombic	Strong birefringence (back-facet doubling) is the visual tell
Jadeite	3.24–3.43	Monoclinic	Nephrite: 2.90–3.03; jadeite SG higher; RI confirms species

Heavy liquids: Methylene iodide (SG ~3.32), bromoform (SG ~2.89), and Clerici solution (SG up to 4.25) are historically used to bracket SG by float/sink behavior. They are toxic, regulated, and require disposal as hazardous waste. Hydrostatic weighing gives the same result safely and is the standard bench method for a reason. Avoid heavy liquids unless you have a fume hood and proper disposal infrastructure.

When SG Alone Resolves the Identification

In four situations, specific gravity measurement alone closes the identification without needing the refractometer:

Diamond vs. Moissanite

Diamond SG 3.52, moissanite SG 3.22. The gap is 0.30 — three times the ±0.10 tolerance Loupewise itself matches against. Thermal conductivity testers are the standard, but SG provides a secondary confirming measurement.

Decisive

Cubic Zirconia vs. Any Natural Gem

CZ's SG of 5.60–6.00 is far above every gem it is ever sold as — diamond and its look-alikes all sit below 4. A CZ of the same face-up size as a diamond weighs noticeably more. Even without a scale, experienced dealers detect CZ by heft. On a scale, it is unmistakable.

Decisive

Jadeite vs. Nephrite

Jadeite SG 3.24–3.43, nephrite SG 2.90–3.03 — the ranges do not overlap, so SG alone separates them. Confirm with RI (jadeite 1.654–1.688, nephrite 1.600–1.627) and texture: jadeite is granular under magnification, nephrite fibrous.

SG separates — RI confirms

Synthetic Corundum vs. Natural

Synthetic corundum is the same mineral as natural, with the same SG range — the measurement cannot tell them apart. SG cannot distinguish synthetic from natural corundum. This requires inclusion observation (microscopy) or advanced spectroscopy.

SG cannot separate these

Where This Comes From

The technique and its error directions are from published gemological sources, re-verified in 2026 against the originals; the elimination figures and the ±0.10 tolerance are computed from Loupewise's own database and engine.

B. W. Anderson, Gem Testing, 6th ed., Heywood, 1958, pp. 54–65 — the formula, the worked-pair method, the small-stone surface-tension advisory and its remedies, and the temperature behaviour of immersion fluids.
Dr Juliette Hibou FGA, “Understanding Specific Gravity”, Gem-A Gem Hub — the formula in modern notation and the beaker-contact rule.
Gemology Project wiki (Hydrostatic Balance) and C. Lewton-Brain (Ganoksin) — rig construction and detergent dosing.
Loupewise engine measurements — median candidates, species solved, and the matching tolerance, all guarded by tests so they move when the data does. Error directions (reads low / reads high) are derived from SG = A/(A−W) and stated with their mechanism.

Frequently Asked Questions

My stone is in a setting — can I still measure SG?

No. A mounted stone cannot be accurately measured by the hydrostatic method. The setting metal adds mass in air but also displaces water, creating errors in both W_air and W_water. The resulting SG will be meaningless. For stones in settings, use other diagnostics: refractometer (if a flat facet is accessible), fluorescence, spectroscopy, or absorption characteristics. Unmounting is the only reliable path to an accurate SG measurement.

How accurate does my SG need to be?

Loupewise’s engine matches an SG reading against each species’ range with a tolerance of ±0.10 — so landing within a tenth is enough for the candidate list to be right. Many separations need far less: ruby (3.97–4.05) and spinel (3.57–3.63) sit 0.34 apart. Species pairs whose ranges genuinely overlap need RI as the second axis, whatever your precision.

Can I use a kitchen scale?

No — and not by a small margin. A typical kitchen scale reads to 1 gram (5 carats). On a ten-carat stone that gives an SG of 4.00 ± 8, which carries no information at all; to land within ±0.05 on a dense stone you would need roughly 1,600 carats, or 320 grams of gemstone. The threshold that matters is not the point where the buoyancy reading becomes larger than one scale division — it is the point where it becomes many times larger. Use a jeweler's digital scale accurate to at least 0.01g (0.05 ct). These are inexpensive, and the arithmetic favours bigger stones: the buoyancy difference you are measuring shrinks with the stone, which is why Anderson advises extra measures below about 3 carats.

What if my reading does not match any gem in the table?

An SG significantly outside all expected ranges suggests one of four situations: (1) An imitation or simulant — glass spans a wide range depending on formulation; synthetic materials can be almost any density. (2) A composite stone — doublets and triplets combine materials of different densities; the bulk SG reflects a blend. (3) A coated stone — resin or polymer impregnation lowers apparent SG, sometimes substantially (treated emerald, polymer-filled turquoise). (4) Measurement error — verify that the stone was fully submerged, not touching the beaker, and free of air bubbles, and that the wire and bridge were properly tared.

How to Measure Specific Gravity of Gemstones