The slope, and the two directions that make it easy

A block on a slope has exactly one force on it that anybody chose: its weight, pulling straight down. Everything else — the push of the surface, the grip that stops it sliding — is a reaction, and reactions are awkward because their size is not known until the problem is solved.

The standard method deals with this by choosing axes that suit the slope rather than the room, and the whole difficulty evaporates. It is worth watching it evaporate, because the move is the single most reused technique in mechanics.

Choosing the axes to suit the problem #

Vertical and horizontal are not privileged directions. They feel privileged because gravity points along one of them, but nothing in Newton’s laws mentions the ground.

On a slope, two other directions are far more useful: along the surface, and perpendicular to it. They are useful for a specific reason. The block cannot move perpendicular to the surface — the surface is in the way — so along that axis the acceleration is zero and the forces must cancel exactly. And the block can move along the surface, so that axis carries the entire interesting question.

Resolving the weight $mg$ into those two directions gives $mg\cos\theta$ pressing into the slope and $mg\sin\theta$ pulling down it. Both follow from a single right-angled triangle whose hypotenuse is the weight and whose angle is the slope angle — a fact that looks like a coincidence and is not: the triangle of forces is similar to the triangle of the slope itself, rotated by ninety degrees.

The perpendicular equation is now trivial. Nothing accelerates that way, so the normal force is $N = mg\cos\theta$, and an unknown has been determined without solving anything. The parallel equation is the one to think about.

This is the same decomposition that makes projectile motion tractable, where the horizontal and vertical parts of the motion are solved separately and reassembled. There the axes were chosen to suit the force; here they are chosen to suit the constraint. Choosing them to suit whichever is more annoying is the general rule.

The angle at which it lets go #

Friction is what stops the block sliding, and the model for it is crude and unreasonably effective: the surface can supply a friction force up to a maximum of $\mu N$, where $\mu$ is a number characteristic of the two materials, and it supplies exactly as much as is needed up to that limit.

So the block stays put as long as

$$mg\sin\theta \le \mu\,mg\cos\theta,$$

and the mass cancels out of both sides. Tilt any block of any weight on the same pair of surfaces and it starts to slide at the same angle — the one where $\tan\theta = \mu$.

That the mass cancels is the striking part. It means the sliding angle is a property of the materials, not of the object, and it can be measured with a plank and a protractor. This is the tilting-plane method, and it remains a standard way to get a coefficient of friction, because the alternative — measuring a force while pulling something at constant speed — requires a steady hand and a calibrated spring.

The number that comes out is a decent guide and not a law of nature, in much the way a refractive index summarises a great deal of optics into one symbol. Friction depends on the true area of contact, which depends on how hard the surfaces are pressed together, which is why the simple proportionality to $N$ works at all. It also depends on how long the surfaces have been sitting still, on any film of moisture or oxide between them, and on speed once sliding starts. The single symbol $\mu$ is a summary of a genuinely complicated surface science.

As the slope steepens #

Push the angle up and the two components trade places.

The two effects reinforce each other, which is why sliding, once it becomes possible, becomes possible very suddenly. As $\theta$ grows, $\sin\theta$ rises and $\cos\theta$ falls: the load along the slope goes up while the grip resisting it goes down. There is no gentle transition — the ratio $\tan\theta$ climbs steeply through the region where slopes typically fail, which is the reason a scree slope holds perfectly until it does not.

At the limits the picture degenerates in ways worth noting. At $\theta = 0$ the slope is a floor: all of the weight presses in, none pulls along, and the normal force equals the weight. At $\theta = 90°$ it is a wall: nothing presses in, so nothing is held, and the block falls freely regardless of $\mu$. The general result contains both special cases, which is a decent test of whether a derivation has been done correctly.

The same triangle somewhere else #

The reason this decomposition deserves attention is that it is not about slopes.

A pendulum displaced from the vertical is the same triangle with the constraint curved instead of flat. The rod takes the component of the weight along its own length and the swing is driven by the component across it — $mg\cos\theta$ and $mg\sin\theta$, exactly as on the slope, with the angle measured from the vertical rather than from the horizontal. What changes is that the direction of the constraint moves as the bob does, which is why the pendulum’s equation is not linear and the block’s is.

A pendulum bob is a block on a slope whose angle changes as it moves. The rod plays the part of the surface: it can only push or pull along its own length, so it absorbs $mg\cos\theta$ and leaves $mg\sin\theta$ to drive the motion. That leftover component is the entire content of the pendulum’s equation of motion, including the $\sin\theta$ that makes it hard to solve.

The same structure appears wherever a constraint removes a direction: a bead on a wire, a car on a banked curve, a ship’s keel resisting sideways drift, a rocket’s thrust decomposed into what fights gravity and what builds orbital speed. In every case some directions are forbidden by the constraint and carry no acceleration, and the useful axes are the ones aligned with that split.

Resolution is not limited to forces either. Adding two waves is done by resolving each into components and adding those, and the velocities before and after a collision are handled the same way once the problem leaves one dimension. The technique belongs to vectors rather than to mechanics, which is why it transfers without adjustment to fields, currents and probability amplitudes.

The same resolution applies to motion as to force, and for the same reason: a vector equation is several scalar equations wearing one symbol. Resolve a velocity along and across a curved path and the along-component is the speed while the across-component is zero by definition — which is what makes the tangential and normal directions the natural axes for a constrained motion, exactly as the surface’s own directions were the natural axes for the forces on it.

What the method costs #

The free-body diagram is so cheap that its price is easy to miss, and it has one — a hard limit, reached sooner than most of its users expect.

The rule for using it is to isolate one object, draw every force acting on that object, and write down that the forces sum to zero and the torques sum to zero. In two dimensions that is three equations per object. In three it is six. The equations are free; the difficulty is that every contact the object makes with anything introduces unknowns, and contacts multiply faster than equations do.

A block on a slope has one contact and two unknowns — the normal force and the friction — against two useful equations, and it works out exactly. Add a second slope, so the block sits in a V, and there are now four unknowns and still only three equations. The problem has become statically indeterminate: the forces are perfectly definite in reality, and rigid-body statics cannot find them.

The familiar version is a four-legged table on a flat floor. Three legs would be determinate; four is not, and nothing in the diagram says how the weight divides between them. Everyone knows the answer experimentally, which is that it depends on how the table was made and which leg is a millimetre short, and that is exactly the physics the model threw away when it declared the table rigid. Recovering it means giving the legs a stiffness and solving for the deflection that makes them all touch — which is no longer statics but elasticity, with a differential equation where there was an equation in three unknowns.

So the cost of the free-body diagram is paid at the point where a structure has more supports than it strictly needs. Every real building is in that regime deliberately, because redundancy is what keeps a structure up when one member fails, and the entire discipline of structural analysis exists to solve the problems that the method in this essay hands over unsolved.

There is a smaller cost, worth naming because it catches people early. A reaction’s direction is not known before the problem is solved, so the diagram is drawn with a guess, and the guess is validated by the sign of the answer. A negative normal force means the surface would have had to pull to hold the block — which surfaces do not do, so the block has left the slope, and the equation that produced the number no longer describes anything. The algebra does not notice. Checking the sign against what a surface is physically able to do is a step the method does not perform on its own behalf.

The slope used as an instrument #

The condition $\tan\theta = \mu$ has an obvious use in the other direction. It contains no mass, no area and no material property except $\mu$ itself, so tilting a surface until the thing on it slips, and reading the angle, measures the coefficient of friction with a protractor.

That is how it is measured, in practice, far more often than by pulling something with a force gauge. A tilting table is a standard piece of laboratory equipment for it, floor tiles are rated for slip resistance on a ramp that is raised until an operator walking on it loses footing, and packaging is tested the same way to find at what angle a stacked load will slide off a lorry bed.

The reading is unusually honest, because the quantity being measured is the only thing left in the equation. The weight cancels — a heavy block and a light one of the same materials slip at the same angle, which is Amontons’s first law appearing as an experimental result rather than as an assumption.

The granular version is the same statement applied to a heap of loose material rather than to a block on a surface. Poured sand builds up until its surface reaches the angle at which grains slide over one another, and then stops; that angle of repose is what fixes the shape of a scree slope, the capacity of a hopper, and how far a spoil heap spreads.

What the free-body diagram cannot show #

The picture is a block, and a block is a lie of a specific and interesting kind.

The diagram draws all forces acting at a single point, which is only legitimate if the object is small, rigid, and not rotating. A real block on a real slope has its weight acting at its centre of mass and the normal force distributed across the whole contact patch — and the distribution is where the missing physics lives. If the slope is steep enough and the block tall enough, the resultant of the normal force can no longer act under the centre of mass, and the block topples instead of sliding. Toppling and sliding are different failure modes with different thresholds, and the point-mass diagram cannot distinguish them, because it has thrown away the shape.

The diagram also assumes the surface is rigid. Push hard enough and it deforms; the block sinks, the contact area grows, and the friction model that was proportional to load stops being proportional to anything simple. Soil mechanics is largely the study of slopes where this matters.

And it assumes the block is not rolling. A cylinder on the same slope obeys a different equation, because some of the released potential energy goes into spin instead of speed — which is why a hoop, a solid cylinder and a sphere released together on a ramp arrive in a fixed and initially surprising order, decided entirely by how the mass is distributed relative to the axis. That is an energy argument rather than a force argument, and it reaches the answer without ever writing down the friction that makes the rolling possible.

Where the friction law stops #

The single symbol $\mu$ carries two claims that the figures use without stating: that friction does not depend on the contact area, and that it does not depend on the speed. Both are approximations of the useful kind — good enough to design with, wrong in stateable places.

Area-independence holds because surfaces touch only at their high points. The true contact area is a small fraction of the apparent one, it is proportional to the load because the high points deform until they can carry it, and so the friction ends up proportional to the load rather than to the size of the block. The argument fails as soon as the material is soft enough for the true area to saturate: a rubber tyre at high load has already flattened everything there is to flatten, so widening the tyre genuinely does increase grip. Racing tyres are wide for a reason the textbook law says cannot exist.

Speed-independence fails at both ends. At very low speeds friction rises with how long the surfaces have been in contact, because contacts creep and grow — which is why a static coefficient exceeds a kinetic one at all, and why the difference produces stick-slip. At high speeds the interface heats, and if it can melt it does: an ice skate slides on a film of water it makes itself, and $\mu$ falls by more than a factor of ten in the process.

Three further places the law simply does not apply. Below a few micronewtons of load, adhesion dominates and friction is proportional to true contact area with no load term at all — this is the regime a gecko’s foot works in, and the regime in which every microelectromechanical device has to be designed. In a hard vacuum, clean metal surfaces cold-weld on contact, and $\mu$ effectively becomes infinite; spacecraft mechanisms are built around this, and a jammed hinge on a satellite has killed missions. And for very soft materials the whole model is replaced, since energy is lost by internal damping in the bulk rather than at the interface.

The domain of validity is therefore something like: hard, dry, unlubricated surfaces, at loads well above adhesion and speeds well below anything that heats them, in an atmosphere that keeps an oxide film in place. Nearly every everyday slope is inside it. Almost nothing at the frontiers of engineering is.

The machine hiding in the picture #

The inclined plane is one of the classical simple machines, and the free-body diagram shows why. Raising a load straight up requires a force equal to its full weight. Pushing it up a slope requires only $mg\sin\theta$ — less force, over a longer distance, for the same energy in the end.

That trade is not a discount. The work done is the same either way, because the distance grows by exactly the factor the force shrinks by, and that exactness is energy conservation making an appearance in a problem that was set up purely with forces. The height climbed is all that the energy accounting sees, and the route is invisible to it. Every simple machine has this structure: a ratio traded against its reciprocal, with the product held fixed by a conservation law that the machine’s designer need never think about.

A screw is an inclined plane wrapped around a cylinder. A wedge is two of them back to back. A road switchbacking up a mountainside is choosing a value of $\theta$ that keeps $\tan\theta$ below what tyres and engines can manage, and the hairpin bends are the price of that choice.

The friction that resists the block is also what makes the slope usable: without it nothing could be pushed up at all, since any pause would let the load slide back. The same force appears twice with opposite signs, once as the obstacle and once as the enabling condition, which is a pattern worth watching for. Air resistance ruins a projectile’s parabola and is the only reason a parachute works.

The ladder from here #

Later rungs: the block that is already sliding, where kinetic friction differs from static and the difference produces stick-slip — the squeal of a brake and the note of a violin. Two blocks connected over a pulley, where the constraint couples two free-body diagrams. The banked curve, where the same triangle is solved for the angle rather than the force. Rolling on the slope, and the moment of inertia that decides the order of arrival. The toppling condition, which needs the shape the point mass discarded. And friction from below — the atomic-scale contact that explains why $\mu$ is roughly constant, which was not understood for the two centuries during which the law worked perfectly well.

Amontons stated the friction laws in 1699, Coulomb refined them in 1785, and the physical explanation arrived in the 1950s. A law can be useful for a very long time before anybody knows why it is true.