How to use a streak plate
Of 108 mineral species we checked, 78 record a streak and exactly half are white. Only quartz, topaz, corundum and schorl are too hard, and they streak white.
A streak plate is unglazed porcelain of about hardness 6.5. Drag the specimen across it and the powder it leaves is the streak. Of 108 species this site covers, 78 record a streak in the Handbook of Mineralogy, and exactly 39 of those are white or colourless — half the test results are no result. Only four species are too hard to mark the plate.
In short
- Half of all streaks are white, and white tells you almost nothing. We read the Handbook sheet for 108 species and found 78 with a recorded streak: 39 white or colourless, 39 a real colour. The test is a coin toss on whether it will answer at all.
- The hardness limit costs you nothing, which is the surprising part. Only four species in the set are harder than the plate — corundum, topaz, quartz and schorl — and three of the four have a white streak anyway. The one case the plate cannot do is a case where it would not have helped.
- Every coloured streak in the set belongs to a crystal of hardness 6.5 or less. Pyrite and marcasite, both 6 to 6.5, are the hardest minerals in the set that give a useful streak, and both sit just inside the plate's limit.
- Streak beats colour because it removes the variables. Hematite is steel-grey, black, dull or brilliant, and its streak is cherry-red every time. Body colour depends on grain size, surface and impurities; powder colour largely does not.
- Use the back of a tile, not a purchased plate, if you like — any unglazed ceramic of about hardness 6.5 works. What matters is that it is unglazed, because a glaze is harder and smoother and will not take powder off the specimen.
| Species | Streak | Hardness | Body colour | What the streak settles |
|---|---|---|---|---|
| Hematite | Cherry-red or reddish brown | 5 to 6 | Steel-grey to black | Separates it from magnetite, whose streak is black |
| Goethite | Brownish yellow to ochre | 5 to 5.5 | Blackish brown | Separates it from hematite and from black manganese oxides |
| Pyrite | Greenish to brownish black | 6 to 6.5 | Pale brass-yellow | Separates it from gold, which streaks gold |
| Cobaltite | Greyish black | 5.5 | Silver-white | Confirms a metallic species against a white-streaking silicate |
| Molybdenite | Bluish grey | 1 to 1.5 | Lead-grey | Separates it from graphite, which streaks black |
| Crocoite | Yellow-orange | 2.5 to 3 | Hyacinth-red to orange | Confirms a chromate against red silicates and oxides |
| Azurite | Pale blue | 3.5 to 4 | Deep azure-blue | Separates it from other blue crusts of the copper suite |
| Malachite | Pale green | 3.5 to 4 | Bright to blackish green | Separates it from black-streaking manganese coatings |
What we counted, and what it says about the test
Advice on streak testing is usually given as though the test always returns an answer. We wanted to know how often it does, so we counted.
Method, so you can repeat it. On 18 September 2026 we took the Handbook of Mineralogy sheet for every species this site covers — 108 sheets after removing duplicates and the non-species files — converted each to text, and extracted the Streak field and the Hardness field from each.
78 of the 108 record a streak at all. Of those 78, 39 are white or colourless and 39 give a real colour. That is an exact half, and it is the single most useful thing to know before you pick up a plate: the likeliest outcome of a streak test on an unknown specimen is a white mark that eliminates nothing.
The 30 sheets with no streak field at all are mostly transparent silicates and zeolites where the property is not diagnostic and the Handbook does not bother. In practice those behave like the white group — the powder is white and the test does not help.
None of which is an argument against the test. It is an argument for knowing in advance whether your candidate species has a coloured streak, which takes ten seconds on a species page and saves a mark on a specimen face.
The hardness limit, and why it costs nothing
A streak plate is unglazed porcelain at about hardness 6.5. The standard caveat is that anything harder will scratch the plate instead of leaving powder, so the test fails on hard minerals. That is true, and we wanted to know how much it actually costs.
Of the 108 species in the set, exactly four are harder than 6.5: corundum at 9, topaz at 8, quartz at 7, and schorl at 7.
Three of those four have a white streak recorded, and the fourth is given no streak at all. Corundum: white. Quartz: white. Schorl: white. Topaz: no streak field.
So the conclusion is unusually clean. The streak plate fails only on species where a streak would have told you nothing anyway. Every one of the 39 coloured streaks in the set belongs to a mineral of hardness 6.5 or less, with pyrite and marcasite at 6 to 6.5 sitting right at the boundary and still working.
The practical consequence: if you get a scratch on the plate rather than a mark, you have learned that the specimen is harder than 6.5, which is a genuine hardness result and a useful one. You have not lost a streak result, because there was none to lose. Our note on testing hardness properly covers what to do with that.
How to do it without damaging the specimen
Streak is a destructive test. It is the gentlest of the destructive tests, but it removes material and it leaves a mark.
Use an edge or an underside, never a display face. A visible scar on the front of a specimen is permanent and it is visible in every photograph of it thereafter. Pick a fresh point on the back where the loss shows on nothing.
One firm stroke, not a scrub. Repeated rubbing grinds plate material into the mark and pales it. A single decisive drag of two or three centimetres gives the truest colour.
Wipe the mark with a damp finger and look again. This is the step people skip and it is the one that matters, for the same reason it matters in a hardness test: a lot of what looks like streak is plate dust or a metallic smear that comes straight off. What survives the wipe is the streak.
Look at it in daylight, next to a known one if you can. The distinction between molybdenite's bluish grey and graphite's black is real and it is not large; the eye judges two greys against each other far better than it judges one in isolation. See our molybdenite note for that specific pair.
Do not streak a soluble, radioactive or arsenic-bearing specimen. The test makes dust by design, which is precisely the wrong thing to do with an arsenide or a uranium mineral. Our notes on toxic minerals and on radioactive specimens set out that group.
Where the streak test is the best test available
For all its limitations, there are questions where streak is the fastest and cleanest answer a collector has, and they are worth naming.
Hematite against magnetite and against black manganese oxides. Hematite streaks cherry-red no matter what colour the specimen is, which is the classic case and still the best one. Body colour ranges from steel-grey and iridescent to dull red; the streak does not move.
Pyrite against gold. Pyrite streaks greenish to brownish black; gold streaks gold. Our note on fool's gold works through the whole comparison, and streak is the cheapest of the tests in it.
Molybdenite against graphite, where density is decisive but needs a scale, and streak needs nothing.
Any metallic mineral against any non-metallic one. This is the underrated use. A specimen that looks metallic and streaks white is a silicate or a carbonate with a lustrous face, not an ore mineral — and that single fact eliminates most of the candidates before you have done anything else. Our note on identifying a specimen puts the test in the order it belongs in.
We keep no catalogue and nothing on this site is offered for sale. Our wanted list sets out the material and the provenance we look for — and a specimen with a streak scar on a display face is worth less than the same specimen with the scar on the back, which is the whole argument for the paragraph above. Our collecting notes cover the other non-destructive tests to run first.