Cleavage, parting or fracture: how to tell
Cleavage repeats, parting does not, fracture never repeats at all. Kyanite's two cleavages meet at 79 degrees; hematite has no cleavage, only parting.
Cleavage is a flat break that repeats: the same flat surface appears on every crystal of the species, as the octahedra a fluorite chips into do. Parting looks identical but repeats only within a twinned specimen. Fracture never repeats — quartz breaks conchoidally at hardness 7, and calcite cleaves into rhombs at 3.
In short
- The single test is repetition. A cleavage plane shows up on every fragment of that species, at the same angles, every time. Parting shows up on some specimens and not others. Fracture is different on every break. Everything else on this page follows from that one distinction.
- Cleavage belongs to the species; parting belongs to the specimen. Every fluorite anywhere splits on the octahedron. Only a twinned hematite parts, and it parts wherever the twin happens to run, which is why parting cannot be predicted from one specimen to the next.
- A flat shiny face is not proof of cleavage. It may be a grown crystal face. A growth face is vitreous and usually carries striations, growth steps or minor irregularity; a cleavage face is pearly or waxy, perfectly flat, and often shows stepped parallel offsets.
- Count the directions and measure the angle. Three directions not at right angles is calcite. One perfect direction plus a second good one at 79 degrees is kyanite. Four directions giving octahedra is fluorite. The angle is more diagnostic than the number.
- This is the property that decides how a specimen is stored, not hardness. Fluorite at hardness 4 and topaz at hardness 8 are both destroyed by a knock in the right direction, because both have a perfect cleavage.
| Species | Cleavage | Parting | Fracture | What you see on a break |
|---|---|---|---|---|
| Fluorite | {111} perfect, four directions | — | — | Octahedral chips; flat triangular faces |
| Calcite | {10-11} perfect, three directions | — | — | Rhombs, at angles that are not 90 degrees |
| Kyanite | {100} perfect, {010} good | {001} | Splintery | Flat lengthwise splits; the two meet at 79° |
| Apophyllite | {001} perfect, {110} imperfect | — | Uneven | A flat pearly top where the termination was |
| Hematite | None | {0001} and {10-11}, from twinning | Uneven to subconchoidal | Flat surfaces on some specimens only |
| Chalcopyrite | {011} and {111}, poor | — | Uneven | No flat surfaces; a brittle brassy powder |
| Quartz | None | — | Conchoidal | Curved, shell-like, glassy; never flat |
| Clinoclase | {001} perfect | — | Uneven | A flat basal split across the needle |
The three-question sequence
Work through these in order on an unknown specimen. It takes under a minute and it resolves most cases without equipment.
Question 1: is the flat surface a break or a grown face? Hold the specimen under a light at a low angle and turn it. A grown crystal face very often carries striations, growth steps, minute etch figures or slight curvature, and it is usually vitreous. A break surface is featureless in that specific way, and a cleavage in particular is often pearly or silky rather than glassy, and may show a flight of parallel steps where the break jumped from one plane to the next. If in doubt, look for the same surface elsewhere on the specimen at a crystallographically sensible angle.
Question 2: does it repeat? This is the whole test. Look at every fragment, every chip, every damaged corner. A cleavage plane repeats — on this specimen, on the next specimen of the same species, and on any small piece broken off. If three separate damaged areas all show flat surfaces parallel to each other, that is cleavage. If one shows a flat surface and the others are irregular, you are probably looking at parting or at a lucky break.
Question 3: how many directions, and at what angles? Count the distinct orientations of flat surfaces. One is a basal cleavage — apophyllite, mica, clinoclase. Three at right angles is galena or halite. Three not at right angles is calcite. Four is fluorite. And where two directions exist, the angle between them is usually more diagnostic than the count: the Handbook of Mineralogy gives kyanite's {100} and {010} as meeting at 79 degrees, which is measurable on a cleaved fragment with a simple contact goniometer or even a protractor and patience.
Parting, and why it behaves so differently
Parting is the property most often mistaken for cleavage, and the distinction is not pedantry — it changes what you can predict about the specimen.
Cleavage is a consequence of the structure. There is a plane of weak bonding in the crystal lattice, it exists in every crystal of that species everywhere, and every specimen behaves the same way. You can state in advance that a fluorite will chip to octahedra.
Parting is a consequence of a defect. Twinning, exsolution lamellae or deformation create a plane of weakness in that particular crystal. A crystal without the defect has no parting at all.
Hematite is the standard example. The Handbook records no cleavage, and parting on {0001} and {10-11} due to twinning. So a twinned hematite plate separates cleanly along a plane, and an untwinned one does not, and there is no way to tell which you have until it happens. The same applies to corundum, where parting on {0001} is common in some material and absent in others, and to pyroxenes.
Practically: treat any flat break on a species known to part as unpredictable, and handle accordingly. A micaceous hematite plate should be supported across its whole area rather than gripped at an edge, because you cannot know where the twin runs. That is a storage decision made from a mineralogical fact, which is what this distinction is for — see handling and storage.
Fracture, and what a conchoidal surface tells you
Fracture is what happens when a mineral breaks in a direction that has no plane of weakness. It is the default: every species fractures, and the ones with no cleavage do nothing else.
Conchoidal is the one worth learning by sight. It is a smooth, curved, shell-like surface with faint concentric ripples, exactly like the break on a piece of thick glass — and for the same reason, because both are isotropic in the relevant sense. Quartz is the reference specimen. If a break is smoothly curved and glassy, the species has no cleavage in that direction, and on a colourless hard mineral that alone narrows the field enormously.
Uneven is the commonest and least informative: a rough, irregular surface. Splintery occurs in fibrous and bladed minerals — kyanite's fracture is splintery, which fits a species that grows in blades. Hackly — jagged, with torn metallic points — is nearly diagnostic of a native metal, and it is one of the identifying properties of native copper. Earthy means the specimen is an aggregate rather than a crystal, and the test is telling you about the aggregate, not the mineral.
The trap: an aggregate breaks like an aggregate. Massive, fine-grained or altered material fractures along grain boundaries whatever the species does, so a botryoidal or earthy specimen will show uneven fracture even for a species with perfect cleavage. Test on a crystal, or accept that the result means nothing.
Why this matters more than hardness for keeping specimens
Collectors reach for hardness when asked which specimens are fragile, and hardness is the wrong property.
Topaz is hardness 8 and has a perfect basal cleavage. A sharp knock in the right direction takes the end off a topaz crystal, and no amount of hardness prevents it. Fluorite is hardness 4 with perfect octahedral cleavage in four directions, which is why a fluorite cube develops chipped, triangular, glassy corners in a drawer. Apophyllite is 4.5 to 5 with a perfect basal cleavage, and the characteristic damage — a flat pearly plate where the termination should be — is that cleavage, not softness.
Meanwhile uvite at hardness 7.5 has no cleavage at all, and hematite at 5 to 6 has none either. Both are far more robust in storage than a much harder cleavable species.
The rule: hardness tells you what will scratch a specimen. Cleavage tells you what will destroy it.
Which means, in a cabinet: cleavable species get individual boxes and fitted beds; species with perfect cleavage in several directions never get stacked or slid; and a specimen is always lifted by the matrix, never by the crystal. Our notes on packing and shipping apply the same reasoning to transit, and the general sequence for unknowns is on how to identify a mineral specimen. British cleavable material we are looking for is listed on the wanted list.