How Science Works | Classical Mechanics — Motion, Forces, Conservation and the Discipline of Prediction

1. Why motion needs an explanation, not just a description

A video can show that a trolley moves farther in each successive second. That establishes a pattern in position, but not its cause. Perhaps the track slopes. Perhaps a stretched spring pulls it. Perhaps the camera itself moves. Mechanics separates the record of motion from the explanation of why that motion changes. Position, velocity and acceleration describe the motion; interactions, mass and constraints help explain it.

This distinction matters because very different causes can produce similar short trajectories. Observing downward motion does not by itself establish constant acceleration. Observing constant speed does not establish zero force: a body travelling around a circle changes velocity because its direction changes. A useful explanation identifies the variables that distinguish competing possibilities rather than simply attaching a familiar name to an observation. The organisation of MIT’s classical mechanics course makes the complementary force-based and conservation-based approaches explicit.

2. Choose the system before counting forces

A system is the object or collection of objects placed inside the explanation. For a falling ball, it might be the ball alone. For a collision, it might include both colliding bodies. For an energy balance, it might include the ball, Earth and surrounding air. These choices do not change reality, but they change which interactions count as internal and which transfer momentum or energy across the boundary.

Consider two skaters pushing apart. For either skater considered alone, the other skater exerts an external force. For the two-skater system, their mutual forces are internal. If external horizontal impulse is negligible, the total horizontal momentum remains constant even while each skater’s momentum changes. “Momentum is conserved” is therefore incomplete without the system and direction being named. The conservation argument is developed in Feynman’s lecture on conservation of momentum.

3. A reference frame is part of the measurement

Motion is measured relative to a reference frame. A passenger can be stationary relative to a train seat while moving relative to the platform. Both descriptions can be correct. Problems arise when a calculation quietly combines a velocity measured in one frame with forces interpreted in another. Specify the origin, axis directions, clock and observer before performing the algebra.

Newton’s familiar force law applies directly in inertial frames. In accelerating or rotating frames, additional inertial terms are needed if the same form of equation is to be retained. This does not make those frames illegitimate; it changes the bookkeeping. A turning vehicle is a useful reminder that the observer is not automatically an inertial platform. A clear discussion of inertia and changing motion appears in Newton’s Laws of Dynamics.

4. Position, velocity and acceleration answer different questions

Position tells us where the body is. Displacement tells us the change in position, including direction. Average velocity is displacement divided by elapsed time; instantaneous velocity is the rate at which position changes at a particular moment. Acceleration is the rate of change of velocity. None of these is a synonym for the others, and acceleration need not point in the same direction as velocity.

A rising ball illustrates the distinction. During its ascent its velocity points upward while gravitational acceleration points downward. At the highest point its vertical velocity is momentarily zero, but its acceleration is not. The motion continues because the state includes more than the value of one variable at one instant. The constant-acceleration equations summarised by OpenStax apply only when the relevant acceleration remains constant over the interval.

5. Newton’s laws connect interaction to change

For a particle of constant mass in an inertial frame, the vector sum of external forces equals mass multiplied by acceleration: ΣF = ma. The sum is essential. A large forward force can coexist with a small acceleration when an almost equally large opposing force acts. A body can also move with nonzero velocity while the resultant force is zero.

The momentum form, force as rate of change of momentum, makes the connection to conservation more visible. However, changing-mass systems require careful treatment of momentum carried across the chosen boundary; one should not casually insert a changing mass into the constant-mass formula and assume the result is complete. For the foundations and interpretation of the law, see Newton’s second law.

6. A free-body diagram is a claim about interactions

A free-body diagram isolates one system and shows the external forces acting on it. Each arrow should have an identifiable source: Earth exerts weight, a surface exerts a contact force, a stretched string exerts tension, or air exerts drag. Velocity is not a force and should not be inserted as another force arrow. The diagram is an argument about what has been included, not merely an illustration beside the calculation.

Newton’s third-law partners act on different bodies. The force of a book on a table and the force of the table on the book therefore do not cancel on the book’s free-body diagram. The book’s weight and the table’s upward contact force may balance, but they are not that third-law pair. Before solving equations, ask of every arrow: who exerts this force, on what, and in which direction? See the MIT Newton’s laws sequence.

7. Contact forces respond to the situation

The normal contact force is not automatically equal to weight. Its value depends on the surface geometry and the other forces or accelerations. Similarly, static friction is not always equal to a coefficient multiplied by the normal force. In a simple model, static friction adjusts up to a limiting magnitude. A book that does not slide under a tiny push does not experience the maximum possible friction merely because that maximum exists.

Kinetic-friction models often approximate the sliding resistance as proportional to normal force. Such models are useful within a range, but actual surfaces can depend on speed, temperature, wear, lubrication and material state. A coefficient is not a universal property of an object independent of its contact partner. The distinction between an actual friction force and its limiting value is set out in OpenStax’s treatment of friction.

8. Constraints remove possibilities without removing physics

A body moving along a track cannot take every possible trajectory. A rigid rod fixes a distance between its ends. An ideal string can relate the motions of two connected bodies. Constraints reduce the number of independent coordinates needed to describe a system, but they also require interactions that enforce the restriction. Ignoring those interactions can produce a mathematically tidy but physically impossible prediction.

In an ideal circular-motion problem, “centripetal force” names the inward resultant required for the motion; it is not an extra interaction to add after tension, gravity or contact forces have already been included. Ask which real interactions supply the inward component. This is a model-selection habit: begin with a physical inventory and only then choose a convenient mathematical description. The wider framework belongs to classical mechanics.

9. Worked example: one trolley, two independent checks

Constructed teaching example, not a measured experiment. Assume a 4.0 kg trolley starts from rest on a horizontal track. A constant 8.0 N forward pull acts while a constant 2.0 N resistance opposes it. Ignore wheel rotation in this deliberately simplified particle model. The resultant force is 6.0 N, so the acceleration is 6.0 ÷ 4.0 = 1.5 m/s².

After 4.0 s, the predicted speed is v = at = 6.0 m/s. The predicted displacement is s = ½at² = 12 m. The units provide a first check: m/s² multiplied by s² gives metres. The increasing distance covered each second is consistent with acceleration rather than uniform motion.

Now use a different route. The net work over 12 m is 6.0 × 12 = 72 J. The final kinetic energy is ½ × 4.0 × 6.0² = 72 J. Agreement between the force–motion calculation and the work–energy calculation is an internal consistency check. It is not experimental verification: both predictions still depend on the same assumed forces and idealisations. The underlying relation is the work–energy theorem.

10. Work follows force through displacement

Mechanical work measures an energy transfer associated with a force acting through displacement. For a constant force, only its component along the displacement contributes. A force perpendicular to instantaneous motion does no instantaneous work on a point particle, even though it can continually change the direction of motion. This is why energy and force descriptions reveal different features of a trajectory.

The work–energy theorem connects net work to the change in kinetic energy. It can be especially efficient when the question concerns speed after a displacement rather than the complete time history. But energy alone may not supply direction, travel time or individual contact forces. Choose the method for the missing quantity, not because one method is universally more sophisticated. See the work–energy formulation.

11. Potential energy belongs to an interaction

Gravitational potential energy describes the configuration of interacting bodies, not a substance hidden inside one object. Near Earth’s surface, mgh is useful when the gravitational field is approximately uniform over the relevant height range. The zero of potential energy can be chosen conveniently because many predictions depend on differences rather than its absolute value.

Mechanical energy is conserved only under the conditions specified by the model. If friction heats a surface, mechanical energy can decrease while total energy remains conserved in a larger system. Saying that energy was “lost” should therefore prompt the question: lost from which account, and transferred into which form or surrounding system? The first-law perspective in The Laws of Thermodynamics makes the wider accounting clear.

12. Momentum makes short, complicated interactions manageable

Momentum is mass multiplied by velocity for an ordinary nonrelativistic particle. Impulse is the accumulated effect of force over time. A brief collision can involve a complicated force history that is difficult to measure point by point, yet the total momentum change can still be constrained by the net external impulse.

For an isolated system, internal interactions redistribute momentum without changing the total. In practical models, “isolated” often means that external impulse is negligible during the short interval being analysed, not that gravity or air literally cease to exist. State the approximation and its timescale. The distinction between external and internal contributions is central to conservation of linear momentum.

13. Worked example: sticking together does not conserve kinetic energy

Constructed teaching example. A 0.50 kg trolley moving at 2.0 m/s strikes a stationary 1.50 kg trolley, and they stick together. Assume external horizontal impulse is negligible during impact. The initial momentum is 0.50 × 2.0 = 1.0 kg m/s. The combined mass is 2.0 kg, so their common final velocity is 0.50 m/s in the original direction.

The initial kinetic energy is ½ × 0.50 × 2.0² = 1.0 J. The final translational kinetic energy is ½ × 2.0 × 0.50² = 0.25 J. The difference, 0.75 J, has transferred into deformation, internal energy and other channels excluded from the translational-motion account.

This is not a contradiction. Momentum conservation and kinetic-energy conservation are different statements. Using both as mandatory conditions for every collision would overconstrain the problem. The observation that bodies remain joined identifies an inelastic event; the momentum model then predicts their shared velocity. The governing conservation principle is described in Feynman’s momentum lecture.

14. Rotation adds geometry to inertia

A force can change rotational motion depending on where and how it acts. Torque combines force with its lever arm relative to a chosen axis. Rotational inertia depends not only on total mass but on how that mass is distributed around the axis. Moving the same mass farther from the axis generally makes a change in angular speed harder to produce.

For a rigid body rotating about a fixed suitable axis, the familiar relation is net torque equals moment of inertia multiplied by angular acceleration. A wheel can therefore have both translational kinetic energy and rotational kinetic energy. Returning to the trolley example, including real spinning wheels changes the energy allocation and potentially the predicted acceleration. That is an improvement in model resolution, not evidence that the simpler equation never works. See Newton’s second law for rotation.

15. Equilibrium is a test of both force and torque

A rigid object can have zero net force but a nonzero net torque. Equal and opposite forces applied along different lines of action can rotate it without accelerating its centre of mass. Static equilibrium therefore requires the relevant resultant forces and torques to vanish. A bridge or shelf that appears motionless still carries internal stresses and contact forces.

Equilibrium also differs from stability. A balanced object may return after a small displacement, remain in its displaced position, or move farther away. The distinction asks what happens to the forces and potential energy when the state is perturbed. Checking only the original balance misses the behaviour that follows disturbance. The torque framework in rotational dynamics provides the starting point for this analysis.

16. Oscillation reveals a restoring relationship

An oscillator repeatedly exchanges energy between forms because a restoring interaction drives it toward an equilibrium while inertia carries it through. Simple harmonic motion is a particular model in which the restoring force is proportional to displacement and opposite in direction. Real systems can depart from that relationship at large amplitudes or under changing material conditions.

A pendulum illustrates the value and boundary of approximation. At sufficiently small angles, the sine of the angle is approximately the angle in radians, producing a simple period formula. Air resistance, pivot friction, the distribution of the pendulum’s mass and larger angles can matter when greater accuracy is required. A measurement that disagrees with the ideal period invites investigation of these assumptions rather than immediate rejection of mechanics. Oscillations sit within the wider MIT mechanics curriculum.

17. Worked example: a prediction is not a measurement

Constructed teaching example. For an ideal small-angle pendulum of length L = 0.900 m, assume g = 9.81 m/s². The model predicts T = 2π√(L/g) = 1.903 s, approximately. That number has been calculated from assumptions. It must not be presented as a laboratory result.

Now consider a separate hypothetical measurement record with L = 0.900 m and T = 1.900 s. Rearranging the model gives g = 4π²L/T² ≈ 9.84 m/s². The small difference from the assumed 9.81 value is not automatically evidence for a new effect. It needs comparison with measurement uncertainty and possible model bias.

For illustration, take independent standard uncertainties of 0.002 m in length and 0.010 s in period. A first-order propagation gives a relative uncertainty √[(0.002/0.900)² + (2 × 0.010/1.900)²] ≈ 1.08%, or about 0.11 m/s² in the inferred g. These invented inputs illustrate uncertainty propagation, not the performance of a real instrument. The reporting discipline follows NIST Technical Note 1297.

18. An experiment should distinguish plausible explanations

Suppose the pendulum model fails more strongly as amplitude increases. That pattern points toward the small-angle approximation rather than a constant clock offset. Suppose every measured period is shifted by nearly the same proportion. Calibration or length definition becomes a different candidate. An informative experiment deliberately changes a variable that causes competing explanations to make different predictions.

Repeated trials estimate variation, but repeating the same biased method does not remove systematic error. Independent timing methods, a clearly defined pivot-to-centre-of-mass length, and an explicit model of the apparatus can test different weak points in the inference chain. This is a design exercise, not an invitation to build hazardous moving equipment. For the role of hands-on comparison in teaching, see MIT’s experimental mechanics syllabus.

19. Graphs should make the model vulnerable

A graph is useful when its axes expose a predicted relationship. Constant acceleration predicts a straight velocity–time graph. A small-angle pendulum model predicts T² proportional to length when other assumptions remain adequate. Plotting quantities selected by the model makes systematic departures easier to see than presenting a collection of unconnected numbers.

Residuals—the differences between observations and model predictions—then reveal what the main curve can hide. A regular pattern in residuals suggests missing structure, whereas a visually attractive fit alone proves little. A flexible curve can fit many datasets without representing the right cause. Keep fitted parameters, measured inputs and genuinely withheld predictions separate. The algebraic predictions for uniformly accelerated motion are stated in motion with constant acceleration; the diagnostic strategy here is an application of that model.

20. Numerical simulation does not remove the need for evidence

Many realistic trajectories cannot be obtained from a convenient closed-form equation. Numerical methods approximate the motion in small time steps, updating forces, velocity and position. The result depends on both the physical model and the numerical method. Making a simulation look smooth is not the same as demonstrating that either is adequate.

One useful check is to reduce the time step and see whether the result converges. Another is to examine energy or momentum conservation where the chosen model predicts it. A third compares against a simpler case with a known solution. These are computational checks, followed by a separate comparison with the real system. Feynman’s discussion of dynamics shows how stepwise calculation can connect a force law to a trajectory.

21. Predictable laws do not guarantee easy long-term prediction

A deterministic model assigns a future state to a specified initial state, but an actual initial state is known only with finite precision. In some nonlinear systems, nearby starting states separate rapidly. Long-term detailed prediction can then become difficult even when the equations themselves are deterministic. The limitation is not necessarily missing randomness in the underlying rule.

The practical response is to report the prediction horizon, uncertainty in initial conditions and the quantities that remain robust. A precise short-term trajectory, a broad range of later positions and a stable long-term statistical property are different kinds of output. The broader modelling discussion on How Mathematical Modelling Works extends this distinction; the mathematical foundations begin with the equations of motion in classical dynamics.

22. Where the classical model stops being sufficient

Classical mechanics is extraordinarily useful for many macroscopic situations at speeds small compared with the speed of light and where quantum effects do not dominate the measured behaviour. It is not the final description of every physical regime. High-speed motion requires relativistic treatment; atomic-scale phenomena can require quantum mechanics; strong gravitational settings require more than a simple Newtonian gravitational model.

The correct lesson is neither “Newton was wrong, so ignore him” nor “Newton works, so nothing deeper matters.” A model has an operating range and a desired accuracy. Using it responsibly means checking that the intended situation remains inside that range. For a next step, use the existing Quantum Mechanics owner; the classical starting point and its qualifications are discussed in Feynman’s dynamics lecture.

23. The CivDJ reading lens: make the whole explanation inspectable

For this learning series, use the CivDJ lens as an editorial checklist rather than as an additional law of physics. Name the entity, define its present state, identify interactions, follow the transition and compare the predicted return with evidence. A trolley, wheel, pendulum and collision can all be read through these questions, but each still needs its own appropriate mechanics.

For the trolley, the entity is the selected moving system; its state includes position and velocity; the relevant relationship is the resultant force; the transition is acceleration; the return is the measured motion. For a collision, the dominant invariant may be total momentum over a short interval. The checklist is useful precisely when it prevents us from treating different scientific jobs as interchangeable.

24. Learning route: from observation to independent modelling

First encounter. Describe motion in ordinary language, then distinguish distance, displacement, speed and direction. Explain what changes when a pull or surface changes. Use a sketch before symbols. The learner’s first success is being able to tell a description of motion from a claim about its cause.

Secondary progression. Introduce vectors, free-body diagrams, resultant forces, constant-acceleration equations, energy transfer and momentum. Require the student to name the system and assumptions before substituting numbers. Compare two solution routes whenever possible, as in the trolley example, and explain why agreement is not yet a real-world test.

Higher-resolution progression. Add variable forces, integration, torque, angular momentum, differential equations and numerical methods. The final task is to decide which information is missing, which model is justified and which measurement could test it. This proposed route is a teaching design, not a claim that every school follows the same sequence or assessment specification.

25. Five checkpoints, with the reasoning exposed

A trolley moves steadily while someone pulls it. Must the pulling force be zero? No. Steady straight-line velocity requires zero resultant force in the inertial-frame model. A nonzero pull can be balanced by resistance. The question tests whether the learner distinguishes one force from the vector sum.

A ball has zero velocity at its highest point. Is its acceleration zero? No. Near Earth’s surface, neglecting air resistance, gravitational acceleration continues downward. The question tests whether the learner confuses a momentary state with its rate of change.

Two bodies stick together after a collision. What should be checked before conserving momentum? Define the system and establish that the external impulse in the analysed direction is negligible over the collision interval. Do not automatically conserve translational kinetic energy as well.

A simulated orbit slowly gains energy. Has new physics been discovered? Not on that evidence. Numerical error, time-step choice, an incorrect force implementation or an unaccounted energy input must be examined first. A computational output is not automatically an observation of nature.

Which is better: force or energy? Neither in isolation. Force analysis is often needed for acceleration and contact interactions; energy is often efficient for changes in speed or configuration. The right choice follows the unknown and the available information. These answers apply the principles developed above, especially the force law and work–energy theorem.

26. What counts as a complete explanation?

A complete mechanics explanation should let another reader reconstruct the route from assumptions to prediction. It identifies the system and frame, defines quantities and units, represents interactions, selects equations, checks limiting cases and distinguishes calculations from observations. It also says what was omitted and why the omission is acceptable at the requested resolution.

For a teacher, the most revealing question is often not “What answer did you get?” but “What would have to change for your answer to change?” A student who can explain how mass, force, distance, friction or initial state affects the result has begun to understand the mechanism. A student who can identify a measurement that could overturn the explanation has begun to understand how science works.

Sources and deeper study

The article’s worked numerical cases are original, explicitly hypothetical teaching examples. They are not reports of experiments conducted by eduKate. The scientific framework can be checked against MIT OpenCourseWare: Classical Mechanics; The Feynman Lectures, Volume I, Chapter 9: Newton’s Laws of Dynamics; Chapter 10: Conservation of Momentum; and the linked OpenStax sections on forces, friction, work, momentum and rotation. Measurement terminology and uncertainty reporting are supported by NIST Technical Note 1297 and the NIST guide to the International System of Units.

Continue through Physics, Materials Science and Mathematics. The wider How Science Works programme connects this branch to other ways of building and testing knowledge.