How this instrument works
Pour dry sand onto any bench and it refuses to build cones steeper than roughly 34°. Grains above that tilt feel gravity's downhill pull exceed whatever grip holds them, so they slip until slope relaxes back to its limit. That limiting tilt is called an angle of repose, and for one rigid body resting on one plane it drops straight out of statics: motion begins where tan θ overtakes static friction coefficient μ, putting onset exactly at arctan(μ). Charles-Augustin de Coulomb set out that reasoning for the Académie in 1773, inside a memoir written for military engineers who needed to know how steeply earth ramparts could be cut; his friction angle still anchors soil mechanics 250 years on.
Bulk handling lives on this number. Hopper walls get cut steeper than the stored powder's repose angle, or product bridges and hangs halfway down. Stockpiles are costed by treating each heap as one cone of volume π r³ tan θ ⁄ 3, so two degrees of error misprices thousands of tonnes of coal. Pharmacopoeias grade powder flow directly by it — under 30° pours freely, past 56° barely moves. Shipping regulators care too: bulk cargo reposing at 35° or below counts as liable to shift, so that figure gets declared before any hold is filled. Real values cluster tightly. Dry sand 30–35°, gravel and crushed rock 40–45°, wheat near 27°, and Martian dune slip faces photographed from orbit sit around 33°, much as they do in Death Valley.
One arctangent assumes rather much, though. It describes rigid bodies on flat surfaces, with friction proportional to load and indifferent to contact area. Loose grains are not that. They interlock, and packed heaps must dilate — expand slightly, as Osborne Reynolds showed in 1885 — before shearing at all, so measured repose angles usually run several degrees above arctan of any coefficient obtained between two polished samples of identical mineral. Moisture adds cohesion, which no coefficient captures: damp sand will hold a vertical wall and then fail as it dries. Heaps also carry two angles rather than one, letting go near some steeper limit and settling two or three degrees below it. Treat this output as physics' clean lower bound — excellent for checking measurements, no replacement for pouring real material.
- Put your material's friction value into Coefficient of static friction: pure ratio, never percentage, never angle.
- Read Angle of repose (degrees) straight off. It is measured up from horizontal, so 0° means flat ground and 90° means vertical.
- Sanity-check against known pairs: 0.6 returns 30.963757°, near dry sand, and 1 returns exactly 45°.
- Working from heaps already measured? Take tan of that angle instead — 40° of pile implies μ ≈ 0.84.
- Sweep upward and watch returns shrink: each tenth added below μ = 1 buys several degrees, while that same tenth above μ = 2 buys barely one.
Worked example — calibrating a tilt table at μ = 1
Your tilt table gets checked before every run of aggregate samples. One reference block, whose static coefficient is known to be exactly 1, goes onto its plate. Type 1 into Coefficient of static friction and Angle of repose (degrees) returns 45 — arctan(1) is π/4 by definition, so no rounding enters anywhere in it. Raise that plate until your block creeps, and the protractor should read 45.000°. Anything else is instrument error rather than physics.
Exactness is why μ = 1 serves as the bench check, but it marks one crossover worth carrying around as well. Below it, surfaces offer less grip than their load's own weight, so nothing loose stands steeper than 45°. Above it, grip wins and steeper heaps become possible. Rubber on clean dry asphalt sits near 0.9 to 1.0, which puts a stationary tyre on 45° of ramp precisely at its limit. Swap 0.6 in afterwards and this readout drops to 30.963757°, dry-sand territory: surrendering 40% of grip cost only 14 degrees, because arctangent flattens as briskly as it climbs.
Questions
Is an angle of repose measured from horizontal or from vertical?
From horizontal. Quoting 34° means 34° above flat ground; it does not mean leaning 34° off a plumb line. Confusion arrives out of drilling and surveying, where inclination is sometimes quoted from vertical instead. If any figure past 60° turns up for dry granular material, suspect flipped convention — almost nothing loose and dry stands steeper than about 50° unaided.
Why do real sand piles stand steeper than arctan(μ) predicts?
Because grains interlock and must ride up over one another before anything shears. Sliding-block friction assumes one flat contact; heaps instead hold thousands of contacts jammed into force networks, and shearing them demands dilatancy — the slight expansion that makes wet sand blanch and go dry underfoot. Angular, rough grains add several degrees over rounded ones of identical mineral. Expect measurement to land above this output rather than below it.
Does moisture raise or lower it?
Raise it sharply, then destroy it. Traces of water form capillary bridges between grains, adding cohesion that no friction coefficient models — damp sand holds vertical walls, which in repose terms means 90° or beyond. Add more and pores fill, bridges vanish, and saturated material slumps toward 0°. Since cohesion and friction are separate mechanisms, wet piles cannot be described by any single μ, so feed this instrument dry-state values only.
How do laboratories actually measure it?
Three ways, each giving slightly different figures. Pour fixed mass through a funnel from fixed height and measure whatever cone forms — that is ISO 4324 and pharmacopoeial powder-flow practice. Or tilt one shallow box until its surface avalanches, capturing maximum static angle. Or rotate some partly filled drum slowly and record surface tilt just before each slip, capturing dynamic angle typically 2 to 5 degrees lower. Always quote which method produced your number; drum results and funnel results are not interchangeable.
How does this differ from an angle of internal friction?
They agree only for loose, cohesionless material. Internal friction angle φ describes shear resistance inside compacted masses and climbs with density, so densely packed sand can reach 45° while identical sand poured loose reposes near 33°. Repose is instead the surface property of freely formed heaps — effectively φ at its loosest packing. Geotechnical design takes φ from shear-box or triaxial testing; stockpile and hopper geometry take repose. Swapping one for another misjudges slope stability, occasionally by dangerous margins.
Do grain size or pile height change the answer?
Hardly at all, which is the pleasing part. Mass cancels out of tan θ > μ, so ball bearings and mountains of talus reach comparable limits, and pouring on more material widens each cone instead of steepening it. Size enters only indirectly: powders finer than roughly 100 μm pick up van der Waals cohesion and can repose past 50°, while millimetre grains behave as this formula expects. Very tall natural slopes eventually fail through deep rotational slips rather than surface avalanching, which is different physics entirely.