How to Use the Mohs Hardness Scale Properly (And Why It Isn't Linear)

Almost every rockhound can recite the Mohs scale from talc to diamond. Far fewer actually run a scratch test the way it's meant to be run, or could explain why a jump from 9 to 10 on that list means something completely different from a jump from 1 to 2. Both things are worth fixing, and neither takes more than a fingernail, a coin, a knife, and a file.

An ordinal scale, not a ruler

The core thing to internalize is that Mohs hardness ranks order, not magnitude. Friedrich Mohs built his 1812 scale entirely on scratch comparisons — mineral A scratches mineral B, so A is harder — and ordered ten common minerals so each scratches everything below it and is scratched by everything above it. Nothing in that definition says the ten steps are evenly sized. They aren't. Measured against instruments that record actual indentation hardness, the gap between corundum (9) and diamond (10) dwarfs the gap between talc (1) and gypsum (2), even though both are "one step" on Mohs' list. Treat the numbers as a sequence you can compare ("harder than," "softer than," "about the same as") and never as a ratio ("twice as hard"), and you'll avoid the single most common misreading of this scale.

Worked example: bracketing a harder specimen

Say you're testing an unknown gray mineral. You run it against a copper penny (about Mohs 3.5) and it scratches the penny — the specimen is harder than 3.5. You try glass or a steel knife blade (about 5.5) and it scratches that too — harder than 5.5. Finally you try a steel file (about 6.5) and neither one clearly scratches the other — call that a near-tie. Feed those three results into the same bracketing logic the Mohs hardness estimator uses and the specimen brackets to a hardness range of 6 to 7. Check which of the ten index minerals fall in that range and you get exactly two: Orthoclase Feldspar and Quartz. That's a real, useful result even though it isn't a single number — you've gone from "no idea" to "one of two specific candidates" using nothing but a penny, a knife, and a file.

Worked example: bracketing a softer specimen

Run the same logic on a soft, chalky white mineral. A fingernail (about 2.5) scratches it easily — softer than 2.5. It in turn scratches a reference chip of talc (hardness 1) — harder than 1. That brackets the specimen to a range of 1 to 2.5, consistent with something like gypsum or a soft, talc-adjacent mineral. Notice what changed between the two examples: the same three-step bracketing method produced a tight, specific range in the first case and a wider one in the second, purely because of which reference points happened to be available near the specimen's real hardness. That's normal. A field kit only has a handful of reference points, so the width of your bracket depends on how close those points happen to land to the specimen's actual hardness, not on how carefully you ran the test.

Why the substitute-kit numbers are approximate, and what that means

The commonly cited hardness values for a fingernail (2.5), a copper penny (3.5), glass (5.5), and a steel file (6.5) are useful working approximations, not certified constants. A fingernail's actual hardness varies a little person to person and softens noticeably with moisture; a modern penny's plating and alloy differ from older coinage and from country to country, which shifts its real hardness slightly; float glass and tempered glass aren't identical in hardness either; and a worn, dull file behaves differently than a fresh one. None of this makes the substitute kit unreliable for its intended purpose — the four values are close enough together that a specimen genuinely near one of these boundaries might occasionally test on the "wrong" side, which is one more reason to treat a scratch-test bracket as an honest range and to prefer a calibrated hardness pick set when a precise boundary actually matters, rather than trusting either kit's numbers to the second decimal.

The mistakes that quietly wreck a scratch test

Four things ruin an otherwise careful scratch test. First, testing a weathered or coated surface instead of a fresh one — a chalky rind can scratch far more easily than the solid mineral underneath, giving a falsely soft reading. Second, testing along an obvious cleavage plane, where you risk splitting the mineral rather than scratching it, which also reads as falsely soft. Third, confusing a genuine scratch (a groove you can feel with a fingernail, still visible after you wipe the surface) with a streak of the softer material rubbing off onto the harder one, which looks similar at a glance but wipes away completely. And fourth, testing only once: a single ambiguous result is worth rerunning on a different spot before you record it, since a tiny inclusion, a hidden crack, or an off-angle scratch can all skew one attempt without you realizing it.

Why the ordinal nature matters practically

Beyond the trivia value, treating Mohs as ordinal changes how you should read a "close" result. If a specimen brackets to 8–9, you're in genuinely rarified territory near the top of the scale, and getting the exact answer right (topaz vs. corundum, say) matters more than it would lower down, precisely because the real hardness difference at that end of the scale is so much larger per step than it looks. Conversely, a bracket like 2–3 covers a real but comparatively small difference in absolute terms, even though it also spans a full Mohs point. Don't let the tidy 1-to-10 numbering trick you into thinking every step carries equal weight — it doesn't, and the two worked examples above sit at noticeably different points on that curve even though both only needed three scratch tests to resolve.

Hardness vs. toughness, again

It's worth repeating because it trips up even experienced collectors: hardness is scratch resistance, not resistance to breaking. A high Mohs number doesn't mean a specimen shrugs off drops or hammer blows — diamond, the hardest natural material, can still shatter cleanly along a cleavage plane if struck the wrong way. Treat a hardness bracket as exactly what it measures, and pair it with cleavage, streak, and specific gravity — the mineral property reference lists all three side by side for common minerals — rather than expecting hardness alone to tell you how carefully to handle a specimen.

A quick safety note

The scratch test itself is safe on almost anything, but two related habits are worth building alongside it. Don't scratch-test a specimen you suspect is a toxic ore mineral (galena, cinnabar, realgar, orpiment) more aggressively or more often than you need to, since filing or scratching creates a small amount of dust each time, and wash your hands afterward regardless of what you tested. And never substitute a taste test for a proper one — the old "lick it to see if it's halite" trick doesn't distinguish a safe mineral from a toxic look-alike, and a scratch-and-streak routine gets you a real answer without that risk.

Reading a bracket you can't narrow further

Sometimes a specimen brackets to a range and stays there no matter how many field substitutes you try, simply because your kit doesn't have a reference point that lands closer to the true value. That's not a failure of the method — it's an honest description of the limits of a fingernail-and-penny kit. A calibrated hardness pick set, which includes small points ground to each of the ten index hardnesses (and sometimes half-steps between them), narrows a bracket further than the four-object field substitute kit ever can, because it gives you many more places on the scale to test against. For most identification work a rough bracket is entirely sufficient once combined with streak, luster, and specific gravity; reach for a full pick set only when hardness genuinely needs to be the tiebreaker between two very similar candidates.

Recording brackets instead of guesses

It's tempting to round a bracket to a single tidy number for the sake of a clean notebook entry — call the 6–7 example above "Mohs 6.5" and move on. Resist that urge. The honest record is the range itself, because collapsing it to a midpoint throws away real information: a specimen that bracketed to 6–7 because it clearly scratched a knife and just barely tied with a file is a meaningfully different result from one that landed in the same numeric bracket by a different combination of tests. Write down what actually scratched what, not just the range that resulted, and the raw notes stay useful even if you revisit the specimen months later with a better reference kit.

Putting hardness in its place

None of this makes hardness the most important property a mineral has — it's one input among several, and the two worked examples above only got to "one of two candidates" and "consistent with gypsum," not a certain single answer. That's the realistic ceiling for a scratch test run with ordinary field substitutes, and it's still a genuinely useful ceiling: narrowing an unknown specimen from thousands of possible species down to one or two candidates, with three quick tests and no lab equipment, is exactly what the Mohs scale was designed to do a little over two centuries ago, and it still does that job well today.

Why the ten index minerals still make sense to memorize

Even with a calculator handling the bracketing arithmetic, it's worth actually memorizing the ten index minerals in order — talc, gypsum, calcite, fluorite, apatite, orthoclase feldspar, quartz, topaz, corundum, diamond — because they double as a mental yardstick you can apply to anything, reference kit or not. A specimen you can scratch with a fingernail is softer than gypsum. One that scratches window glass easily is harder than quartz. That rough intuition, built from repetition rather than a lookup table, is often faster in the field than reaching for a kit at all, and it's the same intuition a mineralogist relies on before ever picking up a formal reference set.