Mohs Hardness Scale: Interactive Chart of Minerals, Gemstones and Metals
Mohs Hardness Scale: Interactive Explorer
The full Mohs hardness scale with 229 minerals, gemstones, metals, ceramics and everyday objects placed on it. Run a scratch test and narrow an unknown sample down to a shortlist, convert between Mohs, Vickers, Knoop and Rockwell, and see on one chart how badly the ten numbers misrepresent what was actually measured.
Switch the chart axis between the Mohs number and a measured hardness and watch the same materials rearrange. On a computer, scroll the wheel over the chart to zoom and drag to pan. On a phone, pinch with two fingers to zoom and slide two fingers to pan, while one finger scrolls the page. Click any label to open it.
The Mohs hardness scale explained, step by step
The scale has ten points and each one is a named mineral, not a number. Friedrich Mohs published it in 1812 and it has outlived every attempt to replace it, because it asks a question anyone can answer with no equipment and no power: does this scratch that.
Every heading below is live. Click it and the chart above jumps to that step and selects the standard, so you can read the description here and see what else sits at the same hardness at the same time.
Before the list, fix one thing in your head. These are ordinal ranks. Talc is 1 and gypsum is 2 because gypsum scratches talc, not because gypsum is twice as hard. Everything after the ten steps builds on that single point, because the gap between the rank and the measurement is where almost every practical mistake with this scale comes from.
Mohs 1, talc
Talc is the softest mineral anyone agreed to put on a scale, and the reason sits in its structure rather than in its chemistry. It is a phyllosilicate built from sheets of silica tetrahedra sandwiching a layer of magnesium hydroxide, and those sandwiches are electrically neutral. Nothing ionic holds one to the next, only van der Waals forces, which are perhaps a hundredth the strength of the covalent bonds inside the sheet. Drag anything across it and the sheets simply slide.
That is also why it feels greasy and why it takes a fingerprint. A fingernail at 2.5 gouges it without effort, and so does a plastic ruler. If a fingernail marks your sample easily and the mark is a real groove rather than a smear, you are at the very bottom of the scale and the candidate list is short: talc, graphite, molybdenite, some clays and a handful of soft sulfides.
It forms where hot fluids attack ultramafic rock or dolomite, so look for it in serpentinite belts and in metamorphosed magnesian limestone. Industrially the softness is the selling point, not a defect: talc goes into paper, paint, ceramics, cosmetics and plastics as a filler precisely because it does not abrade the machinery it passes through. Soapstone is the massive rock made of it, which is why you can carve it with a knife.
Mohs 2, gypsum
Gypsum is the reference for the fingernail test, and it is where most beginners make their first error. A fingernail is about 2.5, so it scratches gypsum easily and fails on calcite at 3. That single test eliminates a large part of the mineral kingdom in one stroke, which is why it belongs first in any field routine even though the glass plate is the more powerful bracket.
Chemically it is hydrated calcium sulfate, and the water in the formula matters more than the hardness does. Heat it above roughly 100 degrees Celsius and it begins dehydrating towards bassanite, which is the reaction that makes plaster of Paris set and also the reason oven drying a gypsum bearing sample changes what you are holding. Anyone running a moisture content on gypsiferous soil has to drop the drying temperature to about 60 degrees Celsius for the same reason.
Three varieties turn up often enough to be worth naming. Selenite is the clear crystalline form that splits into flexible sheets, alabaster is the fine grained massive form that carvers use, and satin spar is the fibrous form with a silky sheen. All three are the same mineral at the same hardness, and all three will take a fingernail.
Mohs 3, calcite
Calcite is the most useful mineral on the whole scale for a working geologist, and not because of its hardness. It is useful because it arrives with a chemical test attached. Drop cold dilute hydrochloric acid on it and it fizzes hard and immediately. Dolomite, only half a step harder, barely reacts unless you scratch it to powder first. That pairing separates limestone from dolostone in about three seconds, and no other property does it that fast.
Its cleavage is perfect in three directions and none of them meet at right angles, which is why a hammered specimen breaks into rhombs rather than blocks. A clear rhomb shows double refraction so strongly that a printed line under it appears twice, and that is both a party trick and a genuine diagnostic. Iceland spar, the optically clear variety, was used in the first polarising instruments for exactly that reason.
On the scratch ladder, a solid copper coin sits just above calcite at about 3.5, so the two bracket a useful window. Calcite is also the reason a marble worktop scratches when a granite one does not, and the reason a lemon leaves an etched dull patch on marble: you are dissolving the mineral, not scratching it.
Mohs 4, fluorite
Fluorite is where the difference between hardness and toughness first becomes impossible to ignore. It has four perfect octahedral cleavage directions, so it chips and splits far more readily than a hardness of 4 would suggest, and lapidaries will tell you it is one of the least forgiving materials on the wheel. A stone can be perfectly hard enough to resist a scratch and still fall apart under light pressure in a setting.
It is calcium fluoride, cubic, and it comes in almost every colour there is, usually because of defects and trace rare earths rather than a chromophore in the formula. Many specimens fluoresce blue under ultraviolet light, and the word fluorescence was coined from the mineral name rather than the other way round.
A common steel nail at about 4.5 sits just above it. So if a copper coin fails to mark your sample but a nail succeeds, you are in fluorite country, along with dolomite, sphalerite, malachite and siderite. That is a crowded window, which is the point at which hardness stops narrowing things and colour, streak and acid reaction take over.
Mohs 5, apatite
Apatite is calcium phosphate, and it is the mineral your teeth and bones are made of. Tooth enamel measures right at Mohs 5 for that reason, which is a useful anchor to carry around: whatever will scratch your teeth will scratch apatite, and whatever your teeth will scratch is softer than apatite.
In geology apatite matters far beyond hardness. It is the standard host for fission track and uranium lead thermochronology, so a large fraction of what we know about when mountain belts cooled and came to the surface was extracted from apatite grains a few hundred micrometres across. It is also the world source of phosphate fertiliser, which makes it one of the few minerals on this list that global food supply depends on.
On the scratch ladder it sits just below the glass plate at 5.5, and glass is the single most useful reference in the kit. Roughly speaking, the mineral world divides at 5.5 into things that will scratch glass and things that will not, and getting that one observation right removes about half the candidates before you have done anything clever.
Mohs 6, orthoclase feldspar
Feldspar is the most abundant mineral group in the crust, making up roughly 60 percent of it, so Mohs 6 is by far the most crowded neighbourhood on the scale. Orthoclase is the potassium end member, and its two cleavages meet at close to 90 degrees, which is where the name comes from: ortho for straight, clase for breaking.
Plagioclase feldspar sits at the same hardness and looks similar, so hardness will never separate them. What does separate them is a set of fine parallel stripes on one cleavage face, the albite twinning, visible with a hand lens and a good light. If you see the stripes it is plagioclase. If the surface is clean and you can see a flame like or perthitic texture instead, it is alkali feldspar.
At 6 you are past a steel knife and past window glass but not past a hardened file. Practically, if a pale mineral scratches a knife blade and then takes a scratch from the file, feldspar is the first thing to suspect, because statistically it usually is. The step from feldspar to quartz sounds like a single point, and in measured Vickers hardness it is a factor of 1.78, a jump that the single Mohs point hides completely.
Mohs 7, quartz
Quartz is the most consequential number on the entire scale, and the reason is dust. Ordinary airborne household dust and beach sand are largely quartz, so 7 is the line above which a material survives everyday handling and below which it slowly goes dull with no accident required. That one fact governs which gemstones can live in a ring, which countertops scratch, and why sandpaper works.
It has no cleavage at all and breaks with a conchoidal fracture that leaves a shell like curved surface and an edge sharper than a surgical scalpel. That property made it the raw material of choice for stone tools for two million years, and it is why obsidian, which is a natural glass with the same fracture behaviour, was traded across continents.
Quartz is also chemically stubborn. It survives weathering and transport long after feldspar has turned to clay, which is why most sand is quartz sand and why a mature sandstone is nearly pure silica. Set the chart above to Vickers and you will see quartz at 1244 against orthoclase at 700, so a single Mohs step conceals a difference of nearly a factor of two in resistance to indentation, and a factor of more than three on the Rosiwal grinding scale.
One field consequence worth memorising: quartz will scratch a hardened steel file, and very few common rock forming minerals will. If your sample bites into the file rather than the other way round, you have eliminated everything below 6.5 in one test.
Mohs 8, topaz
Topaz is the clearest demonstration on the chart that hardness is not durability. At 8 it will scratch quartz without effort, and yet it has one perfect basal cleavage, so a sharp tap in the wrong direction splits a large crystal cleanly in two. Cutters orient the table facet away from that plane specifically to protect it, and jewellers bezel set it rather than using claws.
It is an aluminium fluorosilicate that crystallises out of fluorine rich vapour in the last stages of granite cooling, which is why it turns up in pegmatites and greisens alongside tourmaline, beryl, cassiterite and wolframite. Where you find one you tend to find the others, and that association is itself a prospecting guide.
Most blue topaz on the market began as colourless material that was irradiated and then heated. Imperial topaz, the sherry to pinkish orange variety, is the naturally coloured one and it commands a very different price. Strong light will fade it over years, so it belongs in a piece worn occasionally rather than in a daily ring.
Mohs 9, corundum
Corundum is aluminium oxide, and the gem varieties are ruby when chromium colours it red and sapphire when it is anything else, including colourless. It has no true cleavage, and that combination of hardness 9 and no cleavage makes it the best natural material for a ring worn every day, better in practice than diamond for anyone who is hard on their hands.
It does have parting along polysynthetic twin planes, which mimics cleavage and can take a chip out of a girdle under a hard enough knock. Parting is not cleavage, because it depends on the twinning being there rather than on the structure itself, but the practical effect on a setting is the same.
Synthetic corundum has been made industrially since the Verneuil flame fusion process arrived in 1902, and most of what is produced never goes near jewellery. It becomes sapphire watch crystals, scratch resistant optical windows, laser hosts, bearing jewels and abrasive hones. Emery, the traditional abrasive of emery paper and emery boards, is a natural mixture of corundum and magnetite, which is why the performance of natural emery varies so much between deposits.
Mohs 10, diamond
Diamond closes the scale, and the last step is where the whole system stops working as a measurement. On the Rosiwal grinding scale, corundum scores 1000 and diamond scores 140000. One step on Mohs, and a factor of 140 in how much work it takes to abrade the material away. Nothing else on the scale comes anywhere near that discontinuity.
The hardness comes from geometry as much as from bond strength. Every carbon atom is covalently bonded to four neighbours at 109.5 degrees in a three dimensional framework, with short bonds and no easy slip plane in any direction. That is close to the theoretical maximum for a material made of light atoms at ordinary pressure.
And yet it cleaves perfectly on the octahedron. That is not really a weakness, it is a manufacturing convenience: cleaving was how large rough stones were split before sawing existed, and the Cullinan diamond was divided that way in 1908. A few laboratory materials beat diamond on particular measures, notably cubic boron nitride, which is softer but does not react with hot steel the way diamond does and therefore outperforms it when machining ferrous metals. On straightforward indentation hardness, diamond is still first.
Where the scale came from
People had been ranking stones by scratching them for a very long time before anyone wrote a scale down. Theophrastus noted relative hardness in On Stones around 300 BC, and Pliny the Elder repeated and extended the observations in the Naturalis Historia in AD 77, including the remark that diamond resisted everything. What nobody had done was fix a reproducible set of reference points.
Friedrich Mohs, born in 1773 in Gernrode and trained at Freiberg, catalogued the Vienna mineral collection of the banker Jakob van der Nüll from 1802, then moved to Graz in 1812 as curator for Archduke Johann, where he needed a systematic way to sort specimens. He published the ten mineral scale that same year. His choices were pragmatic rather than theoretical: ten common minerals, easy to obtain, spread across the range, each reliably scratching the one below.
That pragmatism is exactly why it survived. Every attempt to improve it has traded away the thing that made it useful. Raymond Ridgway proposed a fifteen point version in 1933 that stretched the crowded top end, putting garnet at 10, fused zirconia at 11, fused alumina at 12, silicon carbide at 13, boron carbide at 14 and diamond at 15. It is more evenly spaced and almost nobody uses it, because you cannot carry a boron carbide reference in your pocket and because two centuries of literature is written in the original numbers.
The measured scales arrived later and from a different direction. August Rosiwal published his grinding hardness work in 1896. Smith and Sandland at Vickers developed the diamond pyramid indentation test in the early 1920s. Frederick Knoop and colleagues at the United States National Bureau of Standards published the elongated indenter in 1939. None of them displaced Mohs in the field, because none of them can be done on an outcrop with a penknife.
What hardness actually is
Scratch hardness measures resistance to plastic deformation and fracture at a point. When one mineral is dragged across another under load, the real contact area is tiny, so the stress under the contact is enormous, far beyond anything the bulk material would survive in compression. Whichever material yields first gets the groove.
Bond strength and bond density
What decides which yields comes down to two things: how strong the bonds are, and how many of them the indenter has to break per unit area.
Diamond is hard because every bond is a short covalent carbon to carbon bond and there are a great many of them per cubic nanometre, arranged so that no plane is weaker than any other. Talc is soft because although the bonds inside each sheet are perfectly respectable, the bonds between sheets are van der Waals forces, and the indenter never has to break the strong ones. It just slides the sheets past each other. Graphite is soft for exactly the same reason, and both minerals make good dry lubricants because of it.
Within a mineral group the trend is usually clear once you look at cation size and charge. Smaller, more highly charged cations pull the structure tighter, shorten the bonds and raise the hardness. Compare the carbonates in the database above: calcite with calcium sits at 3, magnesite with the smaller magnesium ion reaches 3.5 to 4.5, and smithsonite with zinc gets to 4 to 4.5. Same anion group, same crystal structure, different hardness, and the ordering broadly follows cation size, though zinc sits harder than its radius alone predicts.
Why the theoretical models mostly work
Several people have tried to predict hardness from first principles and the successful attempts all reduce to some version of bond energy per unit volume. Povarennykh related hardness to bond strength divided by the volume the bond occupies. Plendl and Gielisse tied it to volumetric lattice energy. Later work by Gilman and by Simunek and Vackar built models around bond density and the resistance of bonds to shear.
These get the ordering right and the magnitudes roughly right, and where they fail it is usually for a revealing reason: the model assumes an isotropic solid and the real mineral is not. That anisotropy is the subject of its own section further down, and kyanite is the case that ruins every simple model.
Instrumented measurement now backs the theory up. Broz, Cook and Whitney published microindentation results for the Mohs minerals in American Mineralogist in 2006 and found Vickers hardness rising from about 0.14 gigapascals for talc to about 19.6 gigapascals for corundum. That is a spread of roughly 140 times across eight supposedly even steps, and it is the cleanest modern demonstration that the scale is a ranking rather than a measurement.
Why the ten numbers lie about the spacing
Set the chart above to the Mohs number and the ten standards sit at even intervals, one unit apart, exactly as every classroom poster draws them. Now switch the axis to Vickers or Knoop. The first nine bunch into the left half of the plot and diamond flies off on its own. That rearrangement is the single most important thing on this page.
Here are the actual numbers. On the Knoop scale the step from calcite at 3 to fluorite at 4 is a rise of 20.7 percent. The step from corundum at 9 to diamond at 10 is a rise of 233 percent. The same one unit on Mohs, more than ten times the real difference.
Mohs Vickers Knoop Rosiwal 1 14 1 0.03 2 62 32 1.25 3 152 135 4.5 4 204 163 5 5 554 430 6.5 6 700 560 37 7 1244 820 120 8 1795 1340 175 9 2000 2100 1000 10 10000 7000 140000
Read the Rosiwal column, which measures how much material a standard grinding operation removes, and the distortion becomes absurd. Talc to diamond is a factor of about 4.7 million. Corundum to diamond alone is a factor of 140. On the Mohs scale those last two are neighbours, one step apart, and a printed chart draws them the same distance apart as calcite and fluorite.
Notice too that the distortion is not monotonic. The step from 3 to 4 is tiny on every measured scale. The step from 4 to 5 is large, roughly 2.7 times on Vickers. The step from 5 to 6 is small again. So you cannot even rescue the scale by applying a single correction curve. The unevenness is irregular, because the ten minerals were chosen for convenience rather than for spacing.
Four consequences follow, and all of them bite in practice.
You cannot average Mohs numbers. A rock with equal quartz at 7 and calcite at 3 does not have a hardness of 5. If you need a composite figure for something like a drillability estimate, convert each mineral to Rosiwal or Vickers, weight those by volume fraction, and convert back at the end.
You cannot use Mohs in a wear equation. Wear rate scales with something closer to the abrasion resistance, which is the Rosiwal column, and that column spans six orders of magnitude while Mohs spans one.
Differences at the top of the scale are far bigger than they look. When a supplier quotes a coating at Mohs 9, ask what measurement that came from. Between 2000 and 10000 Vickers there is only one Mohs step available to hide in, and materials as different as corundum and diamond both round into it.
Differences at the bottom are far smaller than they look. A mineral at 3 and one at 4 are within about 21 percent of each other on Knoop. For most engineering purposes they behave the same, which is why nobody specifies a countertop by whether it is Mohs 3 or Mohs 4.
Vickers, Knoop and Rosiwal, and when each one is the right tool
Three measured scales appear alongside Mohs in this tool. They are not interchangeable and each answers a different question.
Vickers, the general purpose indentation number
A square based diamond pyramid with a 136 degree apex angle is pressed into the surface under a known load, and the diagonal of the resulting dent is measured under a microscope. Hardness is the load divided by the contact area of the impression, quoted in kilograms force per square millimetre or converted to gigapascals.
It works on almost any solid from lead to diamond, and the number is essentially load independent over a wide range, which is what makes it the default for materials engineers. For hardened steels it converts cleanly to Rockwell C, and the converter above does that using the ASTM E140 table, which is valid between about 240 and 940 Vickers. Outside that window the tool leaves the Rockwell field blank rather than extrapolating, because Rockwell C is undefined for soft metals and meaningless for ceramics.
Knoop, for thin and brittle things
Knoop uses an elongated rhombic pyramid instead of a square one, with a long diagonal about seven times the short one. The resulting dent is shallow and thin, so you can test a coating a few micrometres thick or a brittle ceramic without cracking the specimen, and you can fit an indent onto a single small crystal.
The elongation has a second use. Because the indenter is directional, rotating it on a single crystal face reveals hardness anisotropy that a square Vickers indenter averages away. Oriented Knoop testing is how the directional hardness of minerals like kyanite and calcite was quantified in the first place.
Rosiwal, which is really abrasion resistance
Rosiwal is different in kind from the other two. Rather than indenting the surface, it measures how much material a standard grinding operation removes, with corundum normalised to 1000. So it is a resistance to abrasion number rather than a point hardness, and it is the one that actually predicts how fast a drill bit dulls, a countertop wears or a gemstone facet loses its polish.
It is also the scale on which the Mohs distortion is most extreme, and that is not a coincidence. Abrasion involves repeated micro fracture rather than a single indentation, so it is far more sensitive to fracture toughness and to how the material fails than an indentation test is. If you only take one thing from this section: Mohs predicts scratching, Rosiwal predicts wearing out, and they are not the same question.
Converting between them honestly
The converter above interpolates between the ten standards on a logarithmic axis. That is a deliberate choice and it comes with a warning attached.
There is no equation linking Mohs to Vickers, because Mohs is ordinal and Vickers is a ratio scale. Various rules of thumb circulate, the most common being that Mohs is roughly proportional to the cube root of Vickers. Try it: the cube root of 2000 is 12.6, and 0.675 times that gives 8.5 for corundum, which should be 9. Apply it to diamond at 10000 Vickers and it returns 14.5, which is off the scale entirely. The rule is not wrong so much as it is fitted to the middle of the range and diverges at both ends.
Interpolating between measured anchors is the most defensible thing available, and the result is an order of magnitude estimate rather than a measurement. The tool labels every interpolated value as interpolated and every published one as measured, so you can see which is which before you quote it in anything.
Hardness across the mineral groups
Hardness is not scattered randomly through the mineral kingdom. It follows structure and chemistry closely enough that knowing the group tells you roughly where to expect the number, and knowing the number narrows the group. This is the part that turns a scratch test from a party trick into a method.
Silicates, where polymerisation sets the number
Silicates make up over 90 percent of the crust and their hardness tracks one variable more than any other: how thoroughly the silica tetrahedra are linked together.
At the bottom sit the sheet silicates, where tetrahedra link in two dimensions and the sheets are weakly bound to one another. Talc is 1, the micas are 2 to 3, chlorite is 2 to 3, serpentine is 2.5 to 4, and the clays are 1 to 2. Every one of them cleaves perfectly in one direction, and the cleavage and the softness have the same cause.
At the top sit the framework silicates, where every tetrahedron shares all four corners in three dimensions. Quartz is 7, the feldspars are 6 to 6.5, and even the zeolites, which are frameworks riddled with channels full of water, still manage 3.5 to 5.5.
In between, the chain silicates come out at 5 to 6.5. Pyroxenes are single chains and amphiboles are double chains, and both land in the same window, which is why cleavage angle rather than hardness is what separates them: 87 and 93 degrees for pyroxene, 56 and 124 degrees for amphibole.
The isolated tetrahedra silicates break the pattern, and usefully so. Olivine is 6.5 to 7, garnet is 6.5 to 7.5, zircon is 7.5, topaz is 8. They should be soft by the polymerisation rule and they are not, because the tetrahedra are bound together by strong metal to oxygen bonds instead of by shared corners. This is the group that supplies most of the hard gemstones and most of the durable heavy minerals in a sand.
Carbonates, soft and chemically reactive
Carbonates cluster tightly between 3 and 4.5, and the ordering inside that window follows cation size as described earlier: calcite 3, aragonite and dolomite 3.5 to 4, magnesite and siderite 3.5 to 4.5, smithsonite 4 to 4.5. The copper carbonates, malachite and azurite, sit at 3.5 to 4 with them.
The practical point is that no carbonate will scratch a steel nail and every carbonate will effervesce in acid to some degree. Hardness alone will not distinguish members of the group, so acid reaction, colour and crystal habit have to do the work.
Oxides, the hard end of the ore minerals
Oxides span the widest range of any group and they include the hardest common minerals after diamond. Corundum is 9, spinel is 7.5 to 8, cassiterite is 6 to 7, rutile is 6 to 6.5, magnetite is 5.5 to 6.5 and hematite is 5 to 6.5.
Hematite deserves a note because it is the mineral most often misjudged in the field. It comes in a metallic grey specular form and an earthy red ochre form that look nothing alike, and the earthy variety is much softer than the crystalline one, which is why the published range is so wide. The streak settles it: reddish brown either way.
The hydroxides are systematically softer than the oxides, because the hydroxyl groups weaken the structure. Goethite manages 5 to 5.5, gibbsite and brucite only 2.5 to 3.5, and bauxite, which is a rock rather than a mineral, ranges from 1 to 3 depending on what is in it.
Sulfides, soft, heavy and mostly metallic
Most sulfides are soft. Galena is 2.5, molybdenite is 1 to 1.5, stibnite is 2, covellite is 1.5 to 2, chalcopyrite is 3.5 to 4. What makes them recognisable is not hardness but the combination of a metallic lustre, a dark streak and a high specific gravity: galena at 7.5 feels wrong in the hand for its size, and that is the giveaway.
Two sulfides break the pattern and both are iron disulfide. Pyrite and marcasite both reach 6 to 6.5, hard enough to scratch glass, which is the standard way to separate pyrite from the gold it is so often mistaken for. Gold is 2.5 to 3, malleable, and will take a fingernail scratch. Pyrite will scratch your knife.
Sulfates, halides and phosphates
The sulfates are soft and evaporitic: gypsum 2, anhydrite 3 to 3.5, barite 3 to 3.5, celestine 3 to 3.5. Barite is the one worth remembering because of its density. At a specific gravity of 4.5 it feels far too heavy for a pale non metallic mineral, and that mismatch is diagnostic on its own.
The halides are softer still and mostly water soluble. Halite is 2 to 2.5, sylvite is 2, and both taste of what they are, which is a legitimate field test that mineralogists use and rarely admit to. Fluorite at 4 is the exception, harder and insoluble, and it defines the fourth step of the scale.
The phosphates run wider. Apatite anchors 5, monazite reaches 5 to 5.5, turquoise is 5 to 6 but porous and variable, and vivianite is only 1.5 to 2 and oxidises from colourless to deep blue on exposure to air.
Native elements, from the softest to the hardest thing there is
The native elements contain both ends of the scale. Graphite is 1 to 2 and diamond is 10, and they are the same element. That single comparison is the strongest possible argument that hardness is about structure rather than chemistry: identical atoms, one arranged in weakly bound sheets and the other in a rigid three dimensional framework, and the hardest natural substance sits roughly six orders of magnitude on the Rosiwal abrasion scale above the one you write with.
The native metals cluster at 2.5 to 3: gold, silver and copper are all soft, malleable and heavy. Their softness is why they were the first metals worked by hand, and why a gold coin can be tested by biting it, although the tooth enamel at 5 is doing the work rather than any great skill on the part of the biter.
The same mineral, two different hardnesses
Kyanite is the reason the hardness column in any serious mineralogy text sometimes carries two numbers. Along the length of a bladed crystal it measures about 4.5. Across the blade it measures about 7. Same crystal, same chemistry, a difference of two and a half Mohs steps depending on which way you drag the point. The mineral is named for its blue colour, but the older name disthene means two strengths, which is more informative.
The cause is structural. Hardness depends on which bonds the indenter has to break and in what sequence, and in a strongly anisotropic structure that changes with direction. Kyanite is triclinic with chains of aluminium octahedra running parallel to the length of the blade, so a scratch along the chains slides between them while a scratch across them has to break them.
Kyanite is the extreme case but it is not alone. Calcite, sphalerite and several sulfides all measure differently on different faces, and even diamond varies: the octahedral face is harder than the cube face, which is exactly the property that lets one diamond be polished by another. Oriented Knoop testing on cut sections is how all of this gets quantified.
The field consequence is simple. If you get an inconsistent result on a bladed, platy or fibrous mineral, you may not have made a mistake. Turn the specimen ninety degrees and test again before you throw the observation away, and if the two disagree, record both. That disagreement is a diagnostic in itself.
What changes hardness: water, heat and alteration
Published hardness values assume a fresh, dry, unaltered specimen at room temperature. Depart from any of those and the number moves.
Alteration is the big one in the field. A weathered feldspar surface partly converted to clay will take a fingernail even though fresh feldspar is 6. This is why every guide insists on a fresh broken surface, and it is the most common cause of an anomalously soft reading. If your sample reads two steps softer than it should, suspect the surface before you suspect the identification.
Water weakens silicates. Griggs and Blacic showed in the 1960s that trace water dissolved in the quartz structure dramatically lowers the stress needed to deform it, an effect called hydrolytic weakening that matters for how the crust deforms over geological time. At the scale of a scratch test it is negligible, but it explains why laboratory and geological strengths of quartz differ so much.
Temperature matters for anything near its melting point. Ice is the everyday example. At zero degrees Celsius it rates about 1.5, soft enough to scratch with a fingernail. At minus 80 degrees Celsius it approaches Mohs 6, hard enough to scratch glass. Nothing about the chemistry changed, only how close the material sits to its melting point, and every mineral behaves the same way if you heat it far enough.
Porosity and grain size change the answer for rocks and aggregates. A dense fine grained chert made of quartz behaves like quartz. A porous chalk made of calcite is much softer in practice than a calcite crystal, because the indenter is collapsing pore space rather than breaking bonds.
How to run a scratch test that you can trust
Use a fresh surface. Weathered rinds and polished faces both lie, the first because alteration has softened the outside and the second because a good polish can hide a shallow groove. Break a corner off if you are allowed to.
Press once, firmly, in a short stroke of a few millimetres, and look at the result with a hand lens. Then wipe it with a thumb. This is the step people skip and it is the one that matters most. When a softer object is dragged across a harder one it leaves a trail of its own powder behind, and that trail looks exactly like a scratch until you rub it away. A real groove survives the wipe and catches the light from both sides.
Test the mineral, not the rock. If your specimen is coarse enough to see individual grains, aim at one grain. Granite reads 6 to 7 because that is the quartz and feldspar in it, and the biotite flakes in the same rock are 2.5 to 3. Both readings are correct and neither describes the rock.
Now the worked example, which you can reproduce in the Scratch test identifier above by clicking Load a worked example. Suppose your sample scratches a solid copper coin, and a glass plate scratches your sample.
Sample scratches the copper coin so hardness is above 3.0 The glass plate scratches it so hardness is below 5.5 Window Mohs 3.0 to 5.5
That window leaves 95 of the 229 entries in this database, so two tests have eliminated 58 percent of the population. That is a good return, but it is not an identification. The tool then suggests the steel nail at 4.5 as the next test, because it sits nearest the middle of the remaining window and therefore roughly halves what is left whichever way the result goes. That is a binary search, and it is why three well chosen tests beat six random ones.
Note how the logic handles uncertainty. If the sample scratches a reference whose own hardness is quoted as a range, the only guaranteed conclusion is that the sample is above the bottom of that range. The tool takes the conservative bound in both directions, which is why the window it reports is always at least as wide as the truth and never narrower. It also flags contradictions: if you tell it the sample is softer than a fingernail and also that it scratches a steel file, it says so rather than quietly returning nonsense.
One more caution on the references themselves. The traditional copper coin at 3.5 assumes solid copper. A United States cent minted after 1982 is a zinc blank with a copper plating a few tens of micrometres thick, so it behaves like zinc at about 2.5. Use a coin from before 1982, an offcut of copper pipe or plumbing fitting, or drop the coin from your kit entirely and rely on the fingernail and the nail instead.
Hardness is not toughness, and jewellery is where that bites
Hardness resists scratching. Toughness resists breaking. They are separate properties, they frequently pull in opposite directions, and confusing them is responsible for a large share of damaged jewellery. That is why the Gemstone wearability panel above scores on both rather than sorting on hardness.
Jadeite jade is the textbook case in one direction. At 6.5 to 7 it is at or below the hardness of household dust, so a polished surface slowly goes dull with wear. But its structure is a felted mass of interlocking crystals with no through going plane of weakness, and that gives it exceptional toughness. Jade bangles survive impacts that would destroy a diamond, which is why they have been carved into thin rings for three thousand years.
Topaz is the case in the other direction. At 8 it outranks almost everything in the jewellery case, yet one clean tap on the table facet can split it along the basal cleavage. Tanzanite at 6 to 7 has perfect cleavage in one direction and should never go in a daily ring. Kunzite has two. Even diamond cleaves on the octahedron, which is precisely how cutters used to divide rough stones.
There is also a third property that neither word covers: stability. Opal contains water and crazes into a web of cracks if it dries out. Pearl and coral are calcium carbonate and dissolve in perfume, hairspray and sweat. Amber is fossil resin and softens in alcohol. Turquoise is porous and turns green as it absorbs skin oils. None of that has anything to do with hardness.
The number to carry away is 7. Airborne household dust is largely quartz, so anything below that dulls from ordinary handling with no accident involved. Of the 57 gemstones in this database, 18 are softer than the dust they will live in. Opal, turquoise, lapis lazuli, malachite, pearl, amber and jet are all in that group, and none of them belongs in a ring worn every day no matter how well it is set.
The hardness of things people actually ask about
Most searches that land on a hardness scale are not about minerals at all. They are about a specific material and whether it will scratch or be scratched. Here are the ones that come up most.
Steel
There is no single answer, because heat treatment changes it more than composition does. Pure annealed iron is about 62 Vickers and barely reaches Mohs 4. Mild steel, which is what a common nail is, runs 120 to 160 Vickers and lands at 4 to 4.5. Austenitic stainless such as 304 or 316 sits at 210 to 220 Vickers, which is Mohs 5 to 5.5 and explains why stainless kitchen surfaces scratch so readily.
Harden and temper a high carbon steel and you reach 600 to 900 Vickers, which is 55 to 67 on the Rockwell C scale and Mohs 6 to 7. A good file is at the top of that range at 750 to 900 Vickers, and that is why the file is the most useful hard reference in an ordinary toolbox. The spread between soft iron and a hardened file is a factor of about fourteen in Vickers and only three Mohs steps.
Glass
Ordinary soda lime window glass is 500 to 600 Vickers, which puts it at Mohs 5 to 5.5. Borosilicate, the Pyrex type, is slightly harder at 550 to 620 Vickers. Both are far softer than most people assume, which is why a phone screen picks up scratches from sand and grit in a pocket rather than from keys, which are brass or steel and softer than the glass.
A sapphire watch crystal is a different material entirely, synthetic corundum at 1800 to 2200 Vickers and Mohs 9. It genuinely will not scratch on anything you meet day to day, but it is brittle, so it shatters where a mineral glass would only chip.
Granite and marble countertops
Granite is quartz and feldspar, so it comes out at Mohs 6 to 7 and a steel knife will not mark it. That is the standard authenticity check in a showroom: if a knife scratches the sample, it is not granite.
Marble is recrystallised limestone, so it is calcite at Mohs 3 to 4 and a copper coin or a steel nail will mark it. Worse, it is chemically attacked by anything acidic, so lemon juice, wine and vinegar leave dull etched patches that are not scratches at all but dissolution. The etching is often mistaken for scratching and the two need different repairs.
Engineered quartz worktops are roughly 90 percent crushed quartz in a resin binder. The quartz is 7 but the resin is not, so the composite behaves as a compromise: far more scratch resistant than marble, less heat tolerant than granite, and vulnerable to the binder discolouring rather than to the mineral wearing.
Titanium, tungsten and tungsten carbide
These three get confused constantly in jewellery marketing. Commercially pure titanium is only 99 Vickers, about Mohs 4, and Grade 5 alloy reaches around 350 Vickers. It resists denting and corrosion but it scratches more easily than steel.
Tungsten metal is among the hardest pure metals in ordinary use at about 350 Vickers and Mohs 7 to 7.5. Tungsten carbide is not a metal at all, it is a ceramic of tungsten and carbon cemented with cobalt, and it reaches 1300 to 2400 Vickers and Mohs 8.5 to 9. A tungsten carbide ring will scratch glass and resist almost everything, and it will also shatter rather than bend if you hit it hard, which is why they can be cracked off in an emergency.
Moissanite and other diamond alternatives
Moissanite is silicon carbide, laboratory grown for jewellery, and at 9.25 it is the hardest gem material after diamond. It takes daily ring wear without dulling. Cubic zirconia is a different proposition: at 8 to 8.5 it is hard enough on paper, but its facet edges abrade and the polish clouds within a year or two of daily wear, which is a toughness and edge holding problem rather than a hardness one.
Where hardness matters in engineering
Choosing an abrasive
An abrasive has to be harder than the workpiece, and comfortably so. The working rule is that the abrasive should exceed the workpiece by a wide margin in absolute terms, not by one Mohs step, because a marginal difference means the abrasive wears as fast as the work does.
That is why the abrasive market is stratified the way it is. Garnet at 7 to 8 is the standard waterjet and blasting grit and cuts steel happily. Aluminium oxide at 9 handles hardened steel. Silicon carbide at 9 to 9.5 is used on stone, glass and non ferrous metal and will cut corundum, which is why grey wet and dry paper works on a sapphire crystal. Diamond and cubic boron nitride handle everything else, with cubic boron nitride preferred for hot ferrous work because diamond dissolves into hot steel.
Drilling and tunnelling
This is where mineral hardness turns directly into money. Tool wear on a tunnel boring machine or a drill bit is governed largely by the quartz content of the rock, because quartz at 7 is harder than the steel cutting it and everything softer than about 5.5 is not.
Two indices formalise that. Equivalent quartz content weights every mineral in a thin section by its Rosiwal hardness relative to quartz and sums the result, giving a single number that predicts abrasiveness better than modal quartz alone. It works precisely because it uses the abrasion scale rather than the Mohs number.
The Cerchar abrasivity index measures it directly. A steel stylus hardened to Rockwell C 55 is drawn across a fresh rock surface for 10 millimetres under a 70 newton load, and the flat worn onto the tip is measured under a microscope. The wear diameter in millimetres multiplied by ten is the index. Under ASTM D7625 the bands run from 0.3 to 0.5 for very low abrasiveness up to 4 to 6 for extremely abrasive rock and above 6 for quartzitic material. A quartzite will destroy picks that a limestone barely marks, and the whole difference is one mineral.
Coatings, wear parts and dentistry
Hard coatings are specified in Vickers or in gigapascals, never in Mohs, for the reason given earlier: the interesting range above 2000 Vickers is all compressed into the top Mohs step. Hard chrome plating reaches 800 to 1000 Vickers on a substrate that might be 200. Electroless nickel is 500 to 700 as plated and reaches 900 to 1100 after heat treatment. Titanium nitride coatings go beyond 2000.
Dentistry sits at Mohs 5 whether it likes it or not, because enamel is apatite. A restorative material harder than enamel will wear the opposing natural tooth, and one much softer will wear out itself, so the target is to match rather than to maximise. It is one of the few fields where being harder is actively the wrong answer.
Mistakes that catch people out
Reading a smear as a scratch. The most common error by a wide margin, and it biases whichever way the smear was made: powder left on a glass plate makes the sample look harder, while a metal streak left on the sample makes it look softer. Wipe every mark before you record it.
Trusting a modern coin. Covered above and worth repeating, because the 3.5 figure is still printed in field guides written before the 1982 change in composition.
Testing an aggregate instead of a mineral. A rock has no single hardness. If the sample is coarse enough to see grains, test a grain.
Averaging Mohs numbers. Because the scale is ordinal, the arithmetic mean of two Mohs values is meaningless. Convert to Rosiwal or Vickers, combine there, convert back.
Testing a weathered surface. Alteration can drop an apparent hardness by two steps or more. Break a fresh face.
Ignoring direction on a bladed mineral. Kyanite reads 4.5 one way and 7 the other. If your result feels wrong, rotate the specimen and repeat.
Confusing hardness with scratch visibility. A soft dark mineral shows a scratch clearly, a hard pale one hides it. Under a lens on a fresh surface this stops being a problem, but by naked eye it produces a lot of false negatives.
Expecting hardness alone to identify anything. It will not. Even a tight one step window leaves dozens of candidates in this database. Hardness eliminates, it does not confirm. Streak, cleavage angle, specific gravity and lustre are what finish the job, and the diagram below sets out what to reach for next.
Common questions about the Mohs hardness scale
What is the Mohs hardness scale?
It is an ordinal ranking of how well a material resists being scratched, running from 1 for talc to 10 for diamond. Friedrich Mohs defined it in 1812 using ten reference minerals, each of which scratches the one below it and is scratched by the one above. It is a comparison, not a measurement: the numbers say which is harder, not by how much.
What is the hardest mineral on the Mohs scale?
Diamond, at 10. It is the hardest naturally occurring mineral, and its hardness comes from every carbon atom being covalently bonded to four neighbours in a rigid three dimensional framework with no easy slip plane. A few laboratory materials come close: moissanite is 9.25, boron carbide is about 9.5, and cubic boron nitride is about 9.9 and outperforms diamond for machining hot steel because it does not react with iron.
What is the softest mineral on the Mohs scale?
Talc, at 1. Its structure is built from electrically neutral sheets held to each other only by weak van der Waals forces, so the sheets slide past one another under almost no load. That is why it feels greasy and why it is used as a lubricant and a filler.
Is the Mohs scale linear?
No, and it is not even smoothly non linear. On the Knoop scale the step from 3 to 4 is a rise of about 21 percent while the step from 9 to 10 is a rise of about 233 percent. On the Rosiwal grinding scale, corundum at 9 scores 1000 and diamond at 10 scores 140000, a factor of 140 for a single step. Switch the chart above from the Mohs axis to Vickers or Rosiwal and you can watch the ten standards rearrange themselves.
Why does the Mohs scale have no units?
Because it is an ordinal scale rather than a ratio scale. The numbers encode an order, the way finishing positions in a race do, and there is no physical quantity attached to them. A hardness of 8 is not twice a hardness of 4 in any sense, and there is nothing to attach a unit to. Vickers and Knoop are ratio scales with real units of kilograms force per square millimetre, which is why arithmetic is legal on those and not on Mohs.
What is a fingernail on the Mohs scale?
About 2.5. It will scratch talc and gypsum and it will fail on calcite at 3, which makes it the fastest way to find out whether a sample is very soft. Dry nails test slightly harder than nails that have been in water, which is one more reason to treat all the common field references as approximate.
What is glass on the Mohs scale?
Ordinary soda lime window glass is 5 to 5.5, or 500 to 600 Vickers. It is the single most useful reference to carry, because it divides the mineral world almost exactly in half. Borosilicate glass is slightly harder at 5.5 to 6. A sapphire watch crystal is not glass at all, it is synthetic corundum at 9.
What is steel on the Mohs scale?
It depends entirely on the heat treatment. Mild steel such as a common nail is 4 to 4.5. A stainless kitchen knife is about 5 to 6. A hardened file is 6.5 to 7. Pure annealed iron is only about 4. The spread between soft iron and a hardened file is a factor of roughly fourteen in Vickers hardness and just three Mohs steps.
What is quartz on the Mohs scale, and why does 7 matter so much?
Quartz is 7, and it matters because ordinary household dust and beach sand are largely quartz. Anything softer than 7 will slowly go dull just from being handled and wiped, with no accident required. That is why 7 is the practical dividing line for whether a gemstone can live in a ring worn every day, and why 18 of the 57 gemstones in this database are softer than the dust they will spend their life in.
How do I convert Mohs to Vickers, Knoop or Rockwell?
There is no equation, because Mohs is ordinal. The most defensible approach is to interpolate between the ten measured standards on a logarithmic axis, which is what the converter above does. A rule of thumb sometimes quoted is that Mohs is roughly proportional to the cube root of Vickers, but it is fitted to the middle of the range and fails badly at both ends: apply it to diamond and it returns a Mohs number of about 14.5.
Rockwell C is a separate matter. It is defined for hardened steels only and the standard ASTM E140 conversion covers roughly 240 to 940 Vickers. Outside that window the tool leaves the field blank rather than producing a number that means nothing.
Can two minerals with the same Mohs hardness scratch each other?
Yes, and it is a normal result rather than an error. Published hardness figures carry real spread because specimens vary in purity, alteration and crystal orientation. If two materials have overlapping ranges, a scratch test between them can go either way depending on the individual samples. The Compare two materials panel above says so explicitly whenever the ranges overlap.
Why does my mineral give a different hardness in different directions?
Because hardness can be anisotropic. Kyanite is the extreme example: about 4.5 along the length of a blade and about 7 across it. Calcite, sphalerite and even diamond all vary between faces. If a bladed or platy mineral gives you inconsistent results, rotate it ninety degrees and test again, and record both numbers.
Is hardness the same as toughness or strength?
No. Hardness resists scratching, toughness resists chipping and breaking, and compressive strength resists crushing. They are separate properties and they often disagree. Jadeite jade is softer than quartz and yet exceptionally tough, so bangles survive knocks that would ruin harder stones. Topaz is Mohs 8 and splits cleanly along a basal cleavage under a single tap. Diamond itself cleaves on the octahedron.
Who invented the Mohs hardness scale and when?
Friedrich Mohs, a German born mineralogist, published it in 1812 from Graz, a decade after cataloguing the Vienna mineral collection of the banker Jakob van der Nüll. He was born in 1773 and died in 1839. The idea of ranking stones by scratching was much older, appearing in Theophrastus around 300 BC and in Pliny the Elder in AD 77, but Mohs was the first to fix a reproducible set of reference minerals.
Is there a Mohs scale above 10?
Not in the standard scale. Ridgway proposed a fifteen point extension in 1933 that put garnet at 10, fused zirconia at 11, fused alumina at 12, silicon carbide at 13, boron carbide at 14 and diamond at 15, and it never displaced the original because the references are not things anyone carries. For materials harder than corundum the sensible answer is to quote Vickers or Knoop rather than to stretch an ordinal scale beyond its design.
Does the Mohs scale work on metals and plastics?
Partly. It ranks them correctly against minerals, which is useful, and the tool includes the common metals for that reason. But metals deform plastically instead of fracturing, so a scratch on soft metal is a ploughed furrow with a raised lip rather than a broken groove, and the test tells you less than it does on a brittle mineral. For metals, use Vickers, Brinell or Rockwell. For polymers, use a pencil hardness test or Shore durometer instead.
What this tool does not do
It does not measure anything. Every value here is a published figure, and published hardness figures for minerals carry real spread because specimens vary, faces vary and observers vary. Where sources disagree the range is shown rather than a single number, and a range that spans a whole Mohs step is telling you something honest about the mineral rather than something sloppy about the data.
The Vickers and Knoop values for anything other than the ten standards and the metals are interpolated rather than measured, and the panel labels them as such wherever they appear. The Rosiwal figures for non standard materials are interpolated too. Treat all of them as order of magnitude guides.
It also does not cover microindentation properly. Nanoindentation on oriented crystal faces gives numbers that differ from bulk figures, sometimes substantially, and reconciling that literature is a research problem rather than a lookup.
What it does do is run entirely offline. All 229 entries, the scratch logic, the conversions and the chart work in your browser with nothing uploaded and no network call after the page loads. Save the page and the whole database comes with you into a valley with no signal. The download buttons give you the same data as CSV or JSON if you would rather have it in a spreadsheet or a script.
Related Calculators and Reference
- Rock Density Calculator works out bulk density from mineral composition, which is the property you reach for after hardness has narrowed the field.
- Interactive Geological Time Scale is the companion reference chart, with every eon, era, period and epoch and the golden spike behind each boundary.
- Stereonet handles the orientation data once you know what the mineral is and need to know how the fabric is arranged.
- Outcrop Photo Annotator puts a scale bar and north arrow on the field photograph that goes with your identification.
- Soil Description Generator covers the same describe it properly problem for soils rather than rock.
- Hardness Conversion Calculator: move from scratch hardness to Rockwell, Brinell and Vickers numbers for steel.
- International Mineralogical Association is the body that approves the mineral names and species definitions this database follows.