Rocks & minerals
Which mineral is this?
Minerals are identified by properties, not appearance. Quartz comes in a dozen colours and every one of them scratches glass; hematite is silver, red or black and every one of them leaves a red-brown streak. Rub the specimen on an unglazed tile, note the lustre, try a scratch, and look at how it breaks. Those four answers name most minerals you will ever pick up.
Narrow it down
Showing all 36 minerals
-
Quartz
SiO₂
Glassy, scratches glass and steel easily, breaks with curved shell-like faces and never into flat cleavage planes. Six-sided prisms ending in a point when it has room to grow.
In the record Mohs 7 · SiO2 · trigonal · vitreous lustre check on Macrostrat
Quartz identification in detail -
Orthoclase feldspar
KAlSi₃O₈
Pink, cream or white blocky crystals that break along two flat directions meeting at right angles, giving stepped rectangular faces that flash in the light.
-
Plagioclase feldspar
(Na,Ca)AlSi₃O₈
White to grey feldspar showing fine parallel striations — like the finest possible pinstripe — on one cleavage face. Those twinning lines are diagnostic and orthoclase never has them.
-
Calcite
CaCO₃
Fizzes strongly in vinegar, scratches with a copper coin, and breaks into leaning rhombs rather than cubes. Clear crystals show double refraction — text under them appears twice.
In the record Mohs 3 · CaCO3 · trigonal · vitreous lustre check on Macrostrat
-
Dolomite
CaMg(CO₃)₂
A calcite look-alike that will not fizz until powdered, is slightly harder, and often forms rhombs with curved, saddle-shaped faces — a texture calcite never produces.
In the record Mohs 3.5–4 · CaMg(CO3)2 · trigonal · vitreous lustre check on Macrostrat
-
Gypsum
CaSO₄·2H₂O
Soft enough to scratch with a fingernail, colourless to white, splitting into flexible but not elastic sheets. Does not fizz, does not taste of anything.
In the record Mohs 2 · CaSO4·2H2O · monoclinic · vitreous lustre check on Macrostrat
-
Halite
NaCl
Perfect cubes, dissolves in water, and tastes of salt. The taste test is standard practice and there is no substitute for it here.
In the record Mohs 2.5 · NaCl · isometric · vitreous lustre check on Macrostrat
-
Fluorite
CaF₂
Cubes in purple, green, blue or yellow, softer than glass, with four cleavage directions that knock the corners off into triangles. Most specimens glow blue-violet under UV — the word fluorescence is named after it.
In the record Mohs 4 · CaF2 · isometric · vitreous lustre check on Macrostrat
-
Muscovite mica
KAl₂(AlSi₃O₁₀)(OH)₂
Peels into thin, clear, elastic sheets that spring back when bent — the elasticity separates mica from every other sheet mineral. Silvery to colourless.
-
Biotite mica
K(Mg,Fe)₃(AlSi₃O₁₀)(OH)₂
The same elastic sheets as muscovite but black to dark brown, and the flakes are translucent brown when held to the light rather than opaque.
-
Talc
Mg₃Si₄O₁₀(OH)₂
Feels soapy or greasy, marks paper, and a fingernail carves it easily. Pale green, grey or white. It defines the bottom of the Mohs scale.
In the record Mohs 1 · Mg3Si4O10(OH)2 · triclinic · sub-vitreous lustre check on Macrostrat
-
Graphite
C
Steel-grey, greasy to the touch, marks paper and your fingers black, and is soft enough to smear with a fingernail. Feels lighter than it looks.
In the record Mohs 1–2 · C · hexagonal · sub-metallic lustre check on Macrostrat
-
Pyrite
FeS₂
Brass-yellow metallic cubes with fine parallel striations on each face, hard enough to scratch glass, leaving a greenish-black streak and smelling faintly of sulphur when struck.
In the record Mohs 6–6.5 · FeS2 · isometric · metallic lustre check on Macrostrat
-
Chalcopyrite
CuFeS₂
A deeper, more golden brass than pyrite, soft enough for a knife, and often tarnished into peacock blues and purples. Streak is greenish-black.
In the record Mohs 3.5–4 · CuFeS2 · tetragonal · metallic lustre check on Macrostrat
-
Galena
Handle with carePbS
Bright silver-grey cubes that break into perfect smaller cubes, astonishingly heavy for their size, and soft enough to mark with a coin.
In the record Mohs 2.5 · PbS · isometric · metallic lustre check on Macrostrat
-
Sphalerite
ZnS
Resinous brown to black with six cleavage directions giving a jumbled sparkle, a pale yellow-brown streak, and a rotten-egg smell when scratched.
In the record Mohs 3.5–4 · ZnS · isometric · adamantine lustre check on Macrostrat
-
Magnetite
Fe₃O₄
Black, heavy, scratches glass, black streak, and strongly magnetic — a fridge magnet on a string swings to it. Often as sharp octahedra.
In the record Mohs 5.5–6.5 · Fe2+Fe23+O4 · isometric · metallic lustre check on Macrostrat
-
Hematite
Fe₂O₃
Whatever colour the specimen is — silver, black, red or brown — the streak is always the same rusty red-brown. That single property identifies it against everything else.
In the record Mohs 5–6 · Fe2O3 · trigonal · metallic lustre check on Macrostrat
-
Goethite and limonite
FeO(OH)
Yellow-brown to almost black, often botryoidal like a cluster of grapes, with a distinctly yellow-brown streak that separates it from hematite's red.
-
Malachite
Cu₂CO₃(OH)₂
Bright green with concentric banding in light and dark green, silky when fibrous, and it fizzes in acid because it is a carbonate. Green streak.
In the record Mohs 3.5–4 · Cu2(CO3)(OH)2 · monoclinic · adamantine lustre check on Macrostrat
-
Azurite
Cu₃(CO₃)₂(OH)₂
Deep azure blue with a light blue streak, fizzing in acid. Frequently found grown together with malachite, slowly altering into it.
In the record Mohs 3.5–4 · Cu3(CO3)2(OH)2 · monoclinic · vitreous lustre check on Macrostrat
-
Olivine
(Mg,Fe)₂SiO₄
Sugary olive-green grains with a glassy shine, in dark igneous rock, breaking with curved fractures and no cleavage.
In the record Mohs 6.5–7 · (Mg,Fe2+)2SiO4 · orthorhombic · vitreous lustre check on Macrostrat
-
Garnet
(Mg,Fe,Ca)₃Al₂(SiO₄)₃
Deep red to red-brown balls with twelve rhombic faces, embedded in mica schist, harder than glass and with no cleavage at all.
In the record Mohs 6.5–7.5 · X3Z2(SiO4)3 · isometric · vitreous to resinous lustre check on Macrostrat
-
Hornblende
Ca₂(Mg,Fe,Al)₅(Al,Si)₈O₂₂(OH)₂
Black needle-like or bladed crystals with two cleavage directions meeting at about 60 and 120 degrees. That angle separates the amphiboles from the pyroxenes.
In the record Mohs 6 · (Ca,Na)2–3(Mg,Fe,Al)5(Al,Si)8O22(OH,F)2 check on Macrostrat
-
Augite
(Ca,Na)(Mg,Fe,Al)(Si,Al)₂O₆
Stubby dark green-black crystals whose two cleavages meet at close to a right angle, giving a squarish cross-section rather than the amphiboles' diamond.
In the record Mohs 5.5–6 · (Ca,Na)(Mg,Fe2+,Al,Fe3+,Ti)[(Si,Al)2O6] · monoclinic · vitreous lustre check on Macrostrat
-
Apatite
Ca₅(PO₄)₃(F,Cl,OH)
Green, blue or brown hexagonal prisms that a knife will just scratch and that will not scratch glass. Sitting exactly at 5 makes it the calibration point of the scale.
In the record Mohs 5 · Ca5(PO4,SiO4,SO4)3(OH) · hexagonal check on Macrostrat
-
Barite
BaSO₄
Colourless to white bladed crystals that are startlingly heavy for a non-metallic mineral — the heft is what identifies it. Soft, with good cleavage in three directions.
-
Kyanite
Al₂SiO₅
Blue bladed crystals with a unique property — a knife scratches it along the length of the blade and will not scratch it across the width. Nothing else in common collecting does that.
In the record Mohs 5.5–7 · Al2(SiO4)O · triclinic · vitreous lustre check on Macrostrat
-
Tourmaline
Complex borosilicate
Long prisms with a rounded-triangular cross-section and heavy lengthwise striations, black in the common variety and any colour in the gem varieties.
In the record Mohs 7 · A(D3)G6(T6O18)(BO3)3X3Z · trigonal check on Macrostrat
-
Beryl
Be₃Al₂Si₆O₁₈
Six-sided prisms with flat ends, in green, blue-green, yellow or pink, hard enough to scratch quartz. Crystals grow large — metres, occasionally.
In the record Mohs 7.5–8 · Be3Al2(Si6O18) · hexagonal · vitreous lustre check on Macrostrat
-
Corundum
Al₂O₃
Barrel-shaped or tabular hexagonal crystals that scratch everything except diamond and moissanite, distinctly heavy, often with a satiny sheen across the crystal faces.
In the record Mohs 9 · Al2O3 · trigonal · adamantine lustre check on Macrostrat
-
Native sulphur
S
Vivid canary yellow, soft, brittle, greasy-looking, and it smells and crumbles when warmed in the hand. Nothing else is that yellow.
-
Serpentine
Handle with careMg₃Si₂O₅(OH)₄
Mottled green with a waxy or greasy feel and slippery polished surfaces, sometimes with silky white fibrous veins running through it.
-
Chlorite
(Mg,Fe)₃(Si,Al)₄O₁₀(OH)₈
Dark green flakes that look like mica but are not elastic: bend one and it stays bent. That single test separates chlorite from biotite.
In the record Mohs 2–2.5 · (Mg,Fe)3(Si,Al)4O10(OH)2·(Mg,Fe)3(OH)6 · monoclinic check on Macrostrat
-
Epidote
Ca₂(Al,Fe)₃(SiO₄)₃(OH)
A distinctive yellow-green — the shade is called pistachio and once seen is unmistakable — in slender striated prisms, hard enough to scratch glass.
In the record Mohs 6 · {Ca2}{Al2Fe3+}(Si2O7)(SiO4)O(OH) · monoclinic · vitreous lustre check on Macrostrat
-
Gold
Au
Malleable rather than brittle: a needle point dents it instead of chipping it. Streak is gold-yellow, colour does not tarnish, and the density is so high a small grain feels wrong in the hand.
In the record Mohs 2.5–3 · Au · isometric · metallic lustre check on Macrostrat
Nothing in this key matches every filter at once. That usually means one observation is off — a colour read in shade, or a size guessed rather than measured. Clear the filter you are least sure about and look again.
Streak is the property that does not lie
The single most useful thing you can do to an unknown mineral is rub it hard across the unglazed back of a bathroom tile and look at the powder. Colour varies with trace impurities; the streak is the mineral's true colour with the optical effects removed, and it is almost constant within a species.
Hematite is the demonstration case. Specimens can be silver-grey and metallic, dull earthy red, or shiny black — and all three leave the same rusty red-brown streak. Meanwhile the two minerals most commonly mistaken for gold both give themselves away instantly: pyrite streaks greenish-black, gold streaks gold.
Two caveats. Minerals harder than the tile — roughly 6.5 and above — will not streak at all; they scratch the tile instead, which is itself an observation and puts the specimen straight into the hard group. And a "white" streak means no useful colour rather than a positive result: most non-metallic minerals give white, so a white streak tells you to move on to lustre and hardness.
Hardness, done properly
Mohs hardness is a scratch ranking, not a measurement, and most people do it wrong in the same two ways: they test a weathered surface, and they mistake a smear of metal for a scratch.
Test on a fresh face. Weathering softens the outside of nearly everything.
Rub the mark with your thumb before you judge it. A steel blade drawn across quartz leaves a grey line that looks exactly like a scratch and wipes straight off — that is metal deposited on the mineral, and it means the mineral is harder than the steel.
The tools you already have cover the whole useful range:
| Tool | Approx. Mohs | What it tells you |
|---|---|---|
| Fingernail | 2.5 | Gypsum, talc, graphite, most micas, halite are at or below this |
| Copper coin | 3.5 | Calcite and galena scratch; fluorite does not |
| Steel knife or nail | 5.5 | Apatite and below scratch; feldspar and quartz do not |
| Window glass | 5.5 | The mineral scratching the glass is the informative direction |
| Quartz pebble | 7 | Separates topaz, corundum, garnet and beryl from everything softer |
| Steel file | 6.5 | Useful upper reference before you reach the gem minerals |
Kyanite is worth knowing as the exception that proves the method is real: it is about 4.5 measured along the length of its blades and about 7 measured across them, so a knife scratches it one way and slides off the other. Anisotropic hardness like that is rare and it is diagnostic.
Cleavage versus fracture
How a mineral breaks reflects how its atoms are bonded, so it is a structural property rather than a cosmetic one. Cleavage is breakage along flat planes that reappear every time you break it; fracture is breakage on irregular or curved surfaces.
The distinctive patterns are worth memorising because each one narrows the list to a handful:
- Perfect cubes — galena and halite. Break a cube of galena and you get smaller cubes.
- Rhombs — calcite and dolomite. Leaning parallelograms rather than squares.
- Thin elastic sheets — muscovite and biotite mica. Bend one and it springs back.
- Thin sheets that stay bent — chlorite, and this is exactly how you separate it from biotite.
- Two directions at 90° — feldspars and pyroxenes.
- Two directions at 60° and 120° — amphiboles. The cross-section of a broken hornblende crystal is a flattened diamond, not a square.
- Curved shell-shaped fracture — quartz, olivine, obsidian. No cleavage at all.
Crystal faces and cleavage faces get confused. Crystal faces grew and are often striated, dulled or etched; cleavage faces were made when you broke it and are bright, flat and repeat in parallel sets throughout the specimen.
The extra tests, and when to reach for them
Vinegar. A drop on a fresh face separates the carbonates — calcite, malachite, azurite fizz freely, dolomite only when powdered. Five seconds, and it settles a whole mineral class.
A magnet. Magnetite is strongly magnetic; pyrrhotite is weakly so. Hang the magnet from a thread rather than holding it, so a weak attraction is visible.
Heft. Compare against a similar-sized piece of quartz. Barite is the classic case — a colourless, non-metallic mineral that is obviously too heavy, which is essentially its identification.
Smell. Sphalerite gives off hydrogen sulphide when scratched. Pyrite smells of sulphur when struck. Clay minerals smell of earth when breathed on.
Ultraviolet light. A shortwave UV torch turns a dull box of specimens into a light show and is genuinely diagnostic for some: fluorite glows blue-violet, calcite from some localities glows red, scheelite blue-white, willemite green. Fluorescence varies by locality within a species, so a negative result proves nothing.
Taste. Legitimate for halite and sylvite and nothing else on this list. Do not taste anything with a metallic lustre, and wash your hands after handling galena.
Fool's gold, and the other traps
Gold versus pyrite versus mica. Three tests settle it and none requires equipment. Press a pin or knife point into the grain: gold dents and stays dented, pyrite shatters into powder, mica flakes apart. Streak it: gold leaves gold, pyrite leaves greenish-black. Put it in water and rock the pan: gold sinks immediately and stays put, mica flakes swirl and float away. The colour test people actually rely on — moving into shade — works too: pyrite dulls, gold does not change.
Quartz versus calcite. Both are common, clear and glassy. Calcite is soft enough for a coin, breaks into rhombs and fizzes; quartz does none of those things. A clear rhomb that makes text look doubled is calcite, always.
Amphibole versus pyroxene. Both are dark, glassy and prismatic in igneous rock. Look at the end of a broken crystal: about 90 degrees between the cleavages means pyroxene, about 60 and 120 means amphibole.
Rocks presented as minerals. Granite is not a mineral, it is a mixture of three or four. If a specimen contains obviously different materials — pink blocky, glassy grey, black flakes — you are holding a rock, and the rock key is the right tool.
Building a working kit and knowing when to stop
Everything in this key can be done with a fingernail, a copper coin, a steel knife, a piece of window glass, a white unglazed tile, white vinegar in a dropper, a small magnet on a thread and a 10× hand lens. That kit fits in a pocket, costs less than a takeaway, and resolves the great majority of specimens people bring home.
What it will not resolve: the feldspar group beyond the orthoclase and plagioclase split, the garnet and tourmaline series, most clay minerals, and any fine-grained intergrowth. Those need optical work on a thin section, X-ray diffraction, or a specific gravity measurement — and at that point the right move is a local rock and mineral club or a university geology department rather than more staring.
36 entries here, chosen as the rock-forming minerals plus the ores and collector species that actually turn up in a pocket. Mindat lists over 6,000 valid mineral species; almost all of them are microscopic, rare, or both.
Hardness, formula and crystal system
Colour is the least reliable thing about a mineral and hardness is the most reliable, because hardness is a measurement rather than an impression. Every entry in this key that has a formal mineral name is matched against Macrostrat's mineral definitions, which gives 28 of the 36 entries a published Mohs range, a chemical formula, a crystal system and a lustre. The snapshot was taken on 2026-09-04.
Read that figure as a sequence of tests rather than a list of facts. A fingernail is about 2.5 on the Mohs scale and will mark 6 of these 28. A steel knife blade is about 5.5 and splits the rest: 15 can be scratched by one and the remainder cannot, which is a single test that removes most of the list. A piece of quartz at 7 will scratch everything below it. Do the test on a fresh surface and on the specimen rather than the crust, and scratch the mineral with the tool — a streak of metal left on the rock looks like a scratch and is not one. Crystal system is the other free character: 8 of these are isometric, so the habit you can see in a good specimen narrows the field before any test at all.
The reference table
Formula, hardness range, crystal system and lustre for every named mineral in this key, as published. Where a hardness is given as a range, the mineral genuinely varies — usually with the direction of the scratch or the exact composition.
| Entry | Formula | Mohs | Crystal system | Lustre |
|---|---|---|---|---|
| Talc | Mg3Si4O10(OH)2 | 1 | Triclinic | Sub-Vitreous, Resinous, Waxy, Greasy, Pearly |
| Graphite | C | 1–2 | Hexagonal | Sub-Metallic |
| Gypsum | CaSO4 · 2H2O | 2 | Monoclinic | Vitreous, Sub-Vitreous, Silky, Pearly, Dull |
| Chlorite | (Mg,Fe)3(Si,Al)4O10(OH)2·(Mg,Fe)3(OH)6 | 2–2.5 | Monoclinic | — |
| Halite | NaCl | 2.5 | Isometric | Vitreous |
| Galena | PbS | 2.5 | Isometric | Metallic, Dull |
| Gold | Au | 2.5–3 | Isometric | Metallic |
| Calcite | CaCO3 | 3 | Trigonal | Vitreous, Sub-Vitreous, Resinous, Waxy, Pearly |
| Dolomite | CaMg(CO3)2 | 3.5–4 | Trigonal | Vitreous, Sub-Vitreous, Resinous, Waxy, Pearly |
| Chalcopyrite | CuFeS2 | 3.5–4 | Tetragonal | Metallic |
| Sphalerite | ZnS | 3.5–4 | Isometric | Adamantine, Resinous |
| Malachite | Cu2(CO3)(OH)2 | 3.5–4 | Monoclinic | Adamantine, Vitreous, Silky, Dull, Earthy |
| Azurite | Cu3(CO3)2(OH)2 | 3.5–4 | Monoclinic | Vitreous |
| Fluorite | CaF2 | 4 | Isometric | Vitreous, Dull |
| Hematite | Fe2O3 | 5–6 | Trigonal | Metallic, Sub-Metallic, Dull, Earthy |
| Apatite | Ca5(PO4,SiO4,SO4)3(OH) | 5 | Hexagonal | — |
| Magnetite | Fe2+Fe23+O4 | 5.5–6.5 | Isometric | Metallic, Sub-Metallic |
| Augite | (Ca,Na)(Mg,Fe2+,Al,Fe3+,Ti)[(Si,Al)2O6] | 5.5–6 | Monoclinic | Vitreous |
| Kyanite | Al2(SiO4)O | 5.5–7 | Triclinic | Vitreous, Sub-Vitreous, Greasy, Pearly |
| Pyrite | FeS2 | 6–6.5 | Isometric | Metallic |
| Hornblende | (Ca,Na)2–3(Mg,Fe,Al)5(Al,Si)8O22(OH,F)2 | 6 | — | — |
| Epidote | {Ca2}{Al2Fe3+}(Si2O7)(SiO4)O(O | 6 | Monoclinic | Vitreous |
| Olivine | (Mg,Fe2+)2SiO4 | 6.5–7 | Orthorhombic | Vitreous |
| Garnet | X3Z2(SiO4)3 | 6.5–7.5 | Isometric | vitreous to resinous |
| Quartz | SiO2 | 7 | Trigonal | Vitreous |
| Tourmaline | A(D3)G6(T6O18)(BO3)3X| 7 | Trigonal | — | |
| Beryl | Be3Al2(Si6O18) | 7.5–8 | Hexagonal | Vitreous, Sub-Vitreous, Waxy, Greasy |
| Corundum | Al2O3 | 9 | Trigonal | Adamantine, Vitreous, Pearly |
Free to reuse with a link back to this page. The full snapshot behind every Orvik key is on the open field data page.
Questions people ask about this
How do I do a streak test without a streak plate?
Use the unglazed back of a bathroom or kitchen tile, the unglazed foot ring on the bottom of a mug, or the rough side of a piece of porcelain. Rub the mineral hard across it and look at the powder left behind. Minerals harder than about 6.5 will not leave a streak — they scratch the plate instead, which itself tells you the specimen is in the hard group.
Is my gold real, or is it pyrite?
Press a pin or knife point into it. Gold is malleable and takes a dent; pyrite is brittle and crumbles to powder. Then streak it: gold leaves a gold line, pyrite leaves greenish-black. Gold is also far denser, so a grain sinks immediately and stays at the bottom of a pan, and it does not dull when you move it into shade.
What is the difference between a rock and a mineral?
A mineral is one naturally occurring compound with a defined composition and crystal structure — quartz, calcite, pyrite. A rock is an aggregate of minerals: granite is quartz plus feldspar plus mica. If your specimen shows several visibly different materials locked together, identify it as a rock first, then work on the minerals inside it.
What mineral scratches glass?
Anything with a Mohs hardness above about 5.5: quartz, feldspar, garnet, tourmaline, beryl, corundum, olivine, epidote, magnetite and pyrite. The important part of the test is the direction — rub the mark afterwards, because a steel blade drawn across a hard mineral leaves a grey metal smear that looks like a scratch and wipes off.
Which minerals glow under UV light?
Fluorite reliably glows blue-violet and gave fluorescence its name. Calcite, aragonite, scheelite, willemite, sodalite and some feldspars glow in various colours, but fluorescence depends on trace activators that vary by locality, so a specimen that does not glow is not necessarily a different mineral. Shortwave UV shows far more than longwave, and requires eye protection.
Why does colour not identify a mineral?
Because trace impurities and lattice defects control it, and they vary within one species. Quartz alone occurs colourless, purple, pink, smoky, yellow and black. Fluorite comes in every colour there is. Corundum is a ruby when red and a sapphire when blue, and it is the same mineral. That is exactly why streak, hardness and cleavage are the properties a key is built on.
Can I identify a mineral from a photograph?
Rarely with confidence. A photograph gives colour, lustre, crystal shape and sometimes cleavage, which is often enough to narrow the field to two or three candidates. Hardness, streak, heft and acid reaction are the properties that separate the candidates, and none of them survives a camera. Do the physical tests first and photograph a fresh broken face with side lighting.
Still not sure? Point a camera at it
A filter narrows minerals down to a shortlist. Orvik reads the photo itself and names the species, with the look-alikes ranked underneath so you can check the call yourself.