Digital TownsCrystalsShop crystals
  1. Crystals & Minerals
  2. Buying guides
  3. Hardness Testing

Mohs Hardness Testing: How to Scratch Test a Crystal Without Ruining It

By Digital Towns Crystals · Last reviewed October 8, 2026

Short answer: a Mohs test asks one question: does a point of known hardness cut a groove into your specimen, or not? Use reference minerals or hardness picks rather than coins and kitchen knives, test only an unpolished, hidden spot on rough material, drag the point once with firm pressure, then wipe and inspect with a loupe. Never scratch a polished sphere, tower or faceted stone. A result tells you a range, such as "between 6 and 7", not an identity.

What does a Mohs number actually measure?

Mohs hardness is resistance to being scratched, nothing more. Friedrich Mohs picked ten common minerals of clearly different hardness and set them in order, talc at 1 and diamond at 10, as a relative ladder rather than a measured quantity. Geology.com's Mohs guide dates the scale to 1812; a USGS illustration gives 1822. The idea is far older: the National Park Service notes that comparing hardness was described as far back as 300 BC by Theophrastus in his treatise On Stones.

Because it is a ranking, the steps are uneven. Geology.com tabulates the same ten minerals on the Vickers indentation scale:

Mohs Index mineral Vickers (kg/mm²)
1 Talc 27
2 Gypsum 61
3 Calcite 157
4 Fluorite 315
5 Apatite 535
6 Orthoclase 817
7 Quartz 1161
8 Topaz 1567
9 Corundum 2035
10 Diamond 10000

The jump from corundum to diamond is larger than the whole range from talc to corundum. For a buyer that has a practical meaning: two stones one Mohs step apart can differ hugely in how easily they wear, and "hard" is not the same as "tough". A hard crystal can still chip or split along cleavage, which is why the scratch test sits beside the other checks in the hardness chart rather than replacing them.

Which tools can you trust?

Household objects are less reliable than the charts say

Classroom charts give tidy values. The NPS chart, for example, lists a masonry drill bit at 8.5, steel nail 6.5, knife or glass plate 5.5, copper penny 3.5 and fingernail 2.5. When Geology.com actually tested the items suggested in introductory textbooks, the numbers spread out badly:

Object Chart value often quoted Range Geology.com measured
Fingernail 2.5 2 to 2.5
Glass 5.5 4 to 7
Knife blade 5.5 5 to 6.5
Steel file 6.5 5 to 6.5
Streak plate about 6.5 5 to 7
Quartz crystal 7 7

Two further traps: United States pennies are now made of zinc, so a modern coin is not the copper reference the old charts assume, and a "glass" test means little when glass itself ranges from 4 to 7. The one object Geology.com found dependable was a quartz crystal, at a reliable 7.

Reference minerals and picks

A real kit is either a set of index minerals or a set of hardness picks. Geology.com describes picks with a plastic point at 2, a copper point at 3 and selected alloys from 4 to 9; they reach small grains and either cut a scratch or leave a trace of metal. The International Gem Society suggests making your own by fixing small sharp pieces of synthetic corundum, topaz and quartz to dowels labeled 9, 8 and 7, adding feldspar at 6 and apatite at 5 for a fuller set. Its reasoning for stopping there: anything that a corundum point cannot scratch must be a diamond or a moissanite.

For crystal shop material, the useful range is lower. Rough selenite (gypsum, 2), cleavage pieces of calcite (3), fluorite octahedra (4) and a quartz point (7) cover most of the questions a buyer actually asks, at a cost of a few dollars.

How do you run the test step by step?

  1. Choose the spot. Find a small, flat, unweathered surface on rough material, ideally a protrusion on the back or base. Weathered skins are often softer than the fresh mineral underneath, so a broken fresh surface gives the truest reading.
  2. Brace the specimen. Hold it flat on a table covered with cardboard or a rubber pad. Geology.com advises dragging the point away from your body and parallel to the fingers holding the stone, so a slip does not end in your hand.
  3. Start with your best guess. The IGS method begins with the tool you think matches the specimen.
  4. Make one deliberate stroke. Press the point firmly and drag it slowly across the surface in a single motion. Rubbing back and forth is the classic beginner error.
  5. Wipe before you judge. A softer point leaves powder on a harder surface, like chalk on a blackboard. Brush it off with a finger; a real scratch is a groove that stays.
  6. Check with a loupe. A 10x loupe separates a cut groove from a smear of residue far better than the eye.
  7. Bracket the answer. If the point scratched, step down to a softer one; if not, step up. You finish with two numbers: the softest point that scratched and the hardest that did not.
  8. Repeat once. A stray grain of quartz embedded in a soft specimen can fake a high result, so confirm before you write anything down.

The IGS gives a worked example. A rough yellow stone that an 8 point scratches but a 7 does not has a hardness of at least 7 and below 8, which rules out topaz at 8 and leaves citrine in play.

How do you read the result?

"Scratched" versus "marked"

Equal hardness produces an ambiguous result: two specimens of the same hardness scratch each other poorly, giving faint marks that are hard to judge. That is normal, not a failed test. Read it as "about equal" and move on.

Hardness that changes with direction

Some minerals refuse to give one number. Kyanite is the textbook case: Geology.com reports about 5 when its blades are scratched parallel to their length and about 7 across them, because the bonds differ by direction. Webmineral records kyanite's hardness simply as 4 to 7. Test a kyanite blade lengthwise, get a 5, and you might wrongly conclude it is something else. Diamond cutters exploit the same effect, polishing along the softer cube direction.

Breaking is not scratching

A brittle crystal may chip or crumble under the point. That shows a lack of toughness, not low hardness. If a specimen flakes along flat planes, you have found cleavage; calcite does this readily, which is one of the clearest ways to separate it from quartz in a calcite vs quartz comparison.

Quick reference numbers

From Webmineral's calcite data and its sister pages: gypsum 2, calcite 3, fluorite 4, quartz 7. Geology.com's table adds rhodochrosite 3.5 to 4, malachite 3.5 to 4, sodalite 5.5 to 6, pyrite 6 to 6.5, garnet 6.5 to 7.5 and tourmaline 7 to 7.5.

Where should you never scratch?

The IGS is unambiguous: never perform a scratch test on a finished stone. Pressure can fracture or shatter it, even when tested on an inconspicuous area, because rough and cut material can hold incipient fractures or internal stress that open under load. Its guidance for rough is to test only small protrusions, or better, to saw off a piece. The same organization frames scratch testing, along with streak, acid and hot point tests, as a "last resort" when non-destructive methods fail.

In practice that rules out:

  • Polished spheres, towers, hearts and palm stones. A scratch on a mirror polish is permanent and visible, and it destroys resale value for a result a loupe and a scale could have given you.
  • Faceted or tumbled stones. Tumbled stones are cheap, but the scratch tells you little that their weight and inclusions do not.
  • The face of a cluster or the termination of a point. Those are what you paid for.
  • Coated material. On aura quartz you would be testing a metal film, not the quartz, as the aura quartz guide explains.
  • Anyone else's stock. Scratching a seller's specimen to "check" it is damage, not diligence.

If the stone is finished, use non-destructive checks instead: specific gravity, a loupe for inclusions and polish, and UV. The home tests that do not work guide covers why "it scratched glass, so it is real" proves nothing.

Building a cheap reference kit we stock

A home kit does not need a diamond. What it needs is rough, unpolished reference pieces you are happy to mark, plus a loupe and a soft cloth. Buy references with fresh, flat surfaces, and label each with its number in permanent marker.

  • Selenite, rough wand: selenite is crystalline gypsum, Mohs 2. Your fingernail (2 to 2.5) will just mark it, which makes it a good calibration check for the bottom of the scale. Pick the rough wand variant, not a polished bowl or heart.
  • Optical Calcite: calcite is index mineral 3, and these cleaved pieces show the rhombohedral cleavage that tells you why calcite chips rather than scratches cleanly. Listed in honey and pink pieces from about 1.5 inches.
  • Fluorite Mini Octahedrons: fluorite is the 4 on the scale, and cleaved octahedra give sharp points for testing softer material. The cheapest reference in the set.
  • Blue Kyanite, small blades: not a reference standard but a demonstration piece. Scratch along and across a blade with your quartz point and you can see directional hardness for yourself. Choose a rough small or large piece rather than a tower.
  • Quartz Point Cluster from India: quartz is the dependable 7. A small cabinet cluster supplies points to test with, though for scratching use a broken shard rather than a pristine termination.

Digital Towns Market

Reference minerals for a home hardness kit

See all crystals and minerals

Frequently asked questions

Can I use a knife or a piece of glass to test hardness?

Only as a rough first pass. Geology.com measured knife blades between 5 and 6.5 and glass between 4 and 7, so neither is a fixed reference. A quartz crystal at 7 is far more dependable, and a proper set of picks or index minerals is better still.

Is it safe to scratch test a polished crystal sphere?

No. The International Gem Society says never to scratch test a finished stone, even in a hidden spot, because hidden fractures or internal stress can make it crack. The mark is also permanent. Use specific gravity, a loupe and UV on polished pieces instead.

Why does my kyanite give two different hardness results?

Kyanite has strongly directional hardness: about 5 along the length of a blade and about 7 across it, according to Geology.com. Both readings are correct. Always note which direction you tested.

Does scratching glass prove a crystal is real quartz?

No. Glass varies from about 4 to 7, and many minerals and man-made materials harder than 5.5 will mark it, including glass itself. A scratch on glass narrows the range slightly; it does not identify anything.

What is the difference between hardness and toughness?

Hardness is resistance to scratching. Toughness is resistance to breaking or chipping. A hard stone can still be brittle, and a stone with perfect cleavage can split along flat planes under a blow even though its surface resists scratches.

Sources