Momentum and Collisions

Momentum, impulse, and conservation

Linear momentum is the vector quantity

p = m v .

Its SI unit is $kg m s^{- 1}$ , equal to $N s$ . Newton's second law in momentum form is

\sum F_{ext} = \frac{d p}{d t} .

Integration from $t_{i}$ to $t_{f}$ gives the impulse-momentum theorem,

J = t_{i} \int t_{f} \sum F_{ext} d t = Δ p .

For constant net force, $J = F_{ext} Δ t$ . Impulse is a vector. Component equations retain their signs, so a force that reverses a ball's horizontal velocity provides an impulse whose magnitude exceeds the ball's initial momentum magnitude.

The area under a force-time graph equals the corresponding component of impulse. Two collisions with equal areas have equal momentum changes even when their peak forces and durations differ. Increasing collision time reduces average force for a fixed momentum change. Airbags, crumple zones, padded floors, and a catcher's moving glove use this fact. Peak force cannot be inferred from average force without the force-time profile.

Two force-time histories with equal shaded area deliver the same impulse. The long, low contact reaches the same momentum change as the short, tall one with a smaller peak and average force.

Momentum conservation requires negligible net external impulse over the selected interval:

\int \sum F_{ext} d t \approx 0.

Gravity acts during a ball-bat collision, but its impulse over a few milliseconds is normally negligible beside the contact impulse. During a long descent, gravity's impulse is not negligible. The system boundary and interval determine whether the approximation is appropriate.

Conservation and center-of-mass motion.

A particle system obeys

P = i \sum m_{i} v_{i} .

Internal forces occur in equal-and-opposite pairs, so their vector sum cancels in the rate of change of total momentum. Consequently,

\frac{d P}{d t} = \sum F_{ext} .

If the net external force is zero, total momentum is constant:

P_{i} = P_{f} .

Explosions, recoil, collisions, and spring releases all obey this balance. Internal chemical or elastic energy can become kinetic energy, and kinetic energy can become deformation, sound, and thermal energy while total momentum remains constant.

The center of mass is the mass-weighted position,

R = \frac{1}{M} i \sum m_{i} r_{i}, M = i \sum m_{i} .

Differentiation gives

V_{cm} = \frac{P}{M}, M A_{cm} = \sum F_{ext} .

Its motion obeys the equation for a particle of mass $M$ under the net external force. Fragments of an initially stationary firework shell separate in opposite momentum-weighted directions while their center of mass remains fixed. For a continuous body, $R = M^{- 1} \int r d m$ ; symmetry locates the center of mass only when the mass distribution shares that symmetry.

The center of mass is the mass-weighted mean position, so the marker sits nearer the heavier body. Here the right mass carries three times the weight of the left, placing the mark three-quarters of the way toward it.

Kinetic energy in different frames.

The system kinetic energy is $K = \sum_{i} \frac{1}{2} m_{i} v_{i}^{2}$ . Writing each velocity as $v_{i} = V_{cm} + v_{i}^{'}$ separates this into

K = \frac{1}{2} M V_{cm}^{2} + i \sum \frac{1}{2} m_{i} v_{i}^{'}^{2} = K_{cm} + K^{'} .

The cross term vanishes because $\sum_{i} m_{i} v_{i}^{'} = 0$ . The second term is the kinetic energy measured in the center-of-mass frame and is available to internal collision processes. Internal forces cannot remove the translational energy of the center of mass. Kinetic-energy conservation is frame-sensitive, whereas momentum conservation has the same physical content in every inertial frame.

Collisions and restitution

A one-dimensional two-object collision with negligible external impulse satisfies

m_{1} v_{1 i} + m_{2} v_{2 i} = m_{1} v_{1 f} + m_{2} v_{2 f} .

Two final velocities require a second collision relation in addition to momentum conservation.

Perfectly inelastic: the objects leave together, $v_{1 f} = v_{2 f} = v_{f}$ . $v_{f} = \frac{m _{1} v _{1 i} + m _{2} v _{2 i}}{m _{1} + m _{2}} .$
Elastic: momentum and total kinetic energy are both conserved. $\frac{1}{2} m_{1} v_{1 i}^{2} + \frac{1}{2} m_{2} v_{2 i}^{2} = \frac{1}{2} m_{1} v_{1 f}^{2} + \frac{1}{2} m_{2} v_{2 f}^{2} .$
Inelastic: momentum is conserved for an isolated system, but final kinetic energy is lower than initial kinetic energy. The objects need not stick.

In an elastic head-on collision, the two equations imply

v_{1 i} - v_{2 i} = - (v_{1 f} - v_{2 f}),

and therefore

v_{1 f} = \frac{m _{1} - m _{2}}{m _{1} + m _{2}} v_{1 i} + \frac{2 m _{2}}{m _{1} + m _{2}} v_{2 i},

v_{2 f} = \frac{2 m _{1}}{m _{1} + m _{2}} v_{1 i} + \frac{m _{2} - m _{1}}{m _{1} + m _{2}} v_{2 i} .

Equal masses with the second object initially at rest exchange velocities in this ideal one-dimensional elastic case. Oblique collisions require component equations: momentum is conserved independently along each axis, while any energy condition applies to the full speed, not separately to arbitrary components.

A perfectly inelastic collision ends with both masses at a common center-of-mass velocity. Momentum is unchanged, while kinetic energy generally decreases through deformation and internal energy transfer.

Variable mass and rocket motion

A rocket alone is an open system because exhaust leaves it. Applying momentum conservation to the rocket plus the exhaust expelled during a short interval gives, with $u$ the exhaust speed relative to the rocket and $d m < 0$ the rocket mass change,

m d v = - u d m + F_{ext} d t .

If external force is zero and $u$ is constant,

Δ v = u ln (\frac{m _{i}}{m _{f}}) .

The logarithm results from continuous change of the accelerating mass. During a vertical launch, gravity and drag are external forces that reduce the vehicle's actual velocity gain. Propellant and rocket together still conserve momentum.

Two-dimensional collisions

Momentum conservation is a vector equation. In a planar collision it yields two scalar equations,

\sum p_{x, i} = \sum p_{x, f}, \sum p_{y, i} = \sum p_{y, f} .

An elastic-collision condition adds one scalar equation for total kinetic energy. Geometry or a specified scattering angle generally adds the remaining condition. Component conservation does not mean kinetic energy is separately conserved in the $x$ and $y$ directions: $v^{2} = v_{x}^{2} + v_{y}^{2}$ couples the components.

Consider an elastic collision of equal masses with an initially stationary target. Suppose the incident particle leaves at right angles to the target particle. Momentum gives

v_{i} = v_{1 f} + v_{2 f} .

Squaring this equation and using perpendicular final velocities gives

v_{i}^{2} = v_{1 f}^{2} + v_{2 f}^{2},

which is the kinetic-energy condition. The right-angle geometry is not a universal feature of elastic collisions; it follows from equal masses and an initially stationary target.

Coefficient of restitution.

For direct impacts along a line, a measured collision can be described by the coefficient of restitution

e = \frac{separation speed}{approach speed} = \frac{v _{2 f} - v _{1 f}}{v _{1 i} - v _{2 i}} .

An elastic collision has $e = 1$ ; a perfectly inelastic collision has $e = 0$ . Values between zero and one represent ordinary inelastic impacts. Momentum conservation plus this relation determines the two final velocities in one dimension when external impulse is negligible. Restitution summarizes the normal relative motion; it does not by itself account for rotation, tangential friction, or deformation details outside the one-dimensional model.

Model scope and impulse records

Momentum equations require a system and time interval. For a two-body collision, forces between the bodies are internal; forces from a floor, wall, tether, or field are external. A floor can supply an impulse large enough to invalidate horizontal and vertical conservation differently. The component with negligible external impulse may still be conserved even when another component is not.

Every term in a momentum equation has units $kg m s^{- 1}$ , and every term in an impulse equation has the same units. For a perfectly inelastic collision, the final velocity must lie between the initial velocities when both masses are positive. An isolated system's center-of-mass velocity is unchanged by its internal collision. These checks expose most sign errors before numerical results are used.

Momentum conservation and energy conservation are compatible laws. Total energy is conserved for an isolated enlarged system, but kinetic energy is only one part of that total. Collision equations distinguish conservation of total momentum, which follows from vanishing external force, from conservation of kinetic energy, which requires the additional elastic-collision condition.

Collision calculations require all velocities in the momentum equation to be measured in one inertial frame. Relative speeds belong in a restitution relation. Mixing a relative speed with an inertial- frame speed is a dimensional-looking error that can still produce an incorrect answer. A final check against the center-of-mass velocity is particularly effective for two-body calculations.

When an initially stationary object explodes into two pieces, their momenta must be equal in magnitude and opposite in direction. Three or more fragments must have a vanishing vector sum. An energy release can make the fragment kinetic energies large, but it cannot create a nonzero total momentum in the original rest frame. Recoil problems, firearm motion, and atomic decay share this structure.

Measurements often report collision duration and average force rather than the full force curve. Their product gives the impulse only when the stated force is the average net external force over that same interval. A contact force alone cannot be substituted if gravity, a support, or another external force provides a nonnegligible impulse. In many short impact problems that correction is small, but its neglect remains a physical assumption rather than an algebraic convenience.

Write the selected system and interval before choosing a conservation statement. The same collision record can support a component impulse calculation while failing an isolated-system approximation in another direction.

Analysis item	Relation	Required record
Momentum state	$p = m v$	inertial frame and object or system boundary
Net impulse	$J = \int \sum F_{ext} d t$	force channel, sign, and interval
Momentum balance	$Δ p = J$	initial and final velocities at matching times
Elastic claim	kinetic-energy comparison	separate uncertainty from momentum residual

Impulse as signed force-time area.

Impulse is the signed area beneath a net-force component plotted against time. The shape of the force history matters for peak loading, deformation, and instrument response, while the total signed area fixes the momentum change. A sharp triangular impact and a broad low rectangular contact can produce the same final velocity change, yet impose very different maximum forces on the objects involved.

When a force changes sign during an interval, its positive and negative areas must be combined algebraically. A rebound often contains a braking impulse that brings momentum to zero followed by a reversal impulse in the opposite direction. The net impulse can exceed the magnitude of the initial momentum when the final momentum has opposite sign. Treating all graph area as positive would understate the force-time requirement for reversal.

Average force is defined by $F_{av} Δ t = J$ , but it need not occur at any instant. A reported average force cannot determine the peak force without a force-time profile. Conversely, an impressive peak force over a very short duration can have a small impulse. Momentum analysis uses the area; injury or structural analysis may additionally need the peak and shape.

A rebound force-time record has a braking part (positive area) and a reversal part (negative area). Their signed sum is the net impulse, equal to final minus initial momentum; the total can exceed the incoming momentum in magnitude when the outgoing velocity reverses direction.

Momentum-system boundaries and external impulse.

Momentum conservation is a statement about a selected system over a selected time interval. The interaction force between two colliding carts is internal only if both carts lie inside the boundary. A wall force is external to a cart system but internal to a cart-wall-Earth system. Enlarging the boundary can turn a troublesome external contact into an internal interaction, but it can also introduce additional momentum carriers whose motion must then be included.

The relevant approximation is net external impulse, not the size of an external force at one instant. During a millisecond collision, gravity can be substantial in magnitude yet supply little impulse compared with the contact force. During a longer push or a bounce on a floor, gravity and support impulses may be important. Momentum can be approximately conserved horizontally while not conserved vertically when the floor supplies a significant vertical impulse but negligible horizontal impulse.

A boundary diagram lists the forces crossing the boundary and their impulses over the selected interval. Their vector sum is compared with the system momentum change. Internal force pairs should not appear in the final external impulse sum. A force need not be zero to be neglected; its integrated effect must be small relative to the momentum scale of the problem.

The collision force is internal when both carts lie inside the dashed system boundary; gravity and any floor contact cross the boundary as external impulses. Whether those external impulses are negligible depends on the chosen interval.

Building a collision balance.

A collision model specifies its system boundary, coordinate direction, time interval, and collision type. Momentum conservation yields one equation for each component with negligible external impulse. A perfectly inelastic model adds a common final velocity. An elastic model adds kinetic-energy conservation or an equivalent relative-speed relation. A coefficient of restitution gives a measured relation for direct impacts. Without one of these additional model statements, momentum alone cannot determine all unknown final velocities.

Momentum components retain their signs throughout the calculation. A negative result indicates motion opposite the selected positive direction. The final centre-of-mass velocity gives a quick check; for an isolated system it must match the initial total momentum divided by total mass regardless of whether the collision is elastic or inelastic.

Kinetic energy is a separate diagnostic. In a perfectly inelastic collision, the final kinetic energy is lower unless the objects already shared a common velocity. The difference becomes internal deformation, thermal energy, and sound in an enlarged system. Momentum conservation remains exact under negligible external impulse even when the collision is highly dissipative.

Collision-balance audit.

After solving, recompute total momentum from the final velocities and compare it with the initial value component by component. Then test the kinetic-energy result against the collision model: it must match for an elastic collision and cannot increase in a passive perfectly inelastic collision. Finally, verify that every velocity belongs to the same inertial frame and that the assumed external impulses were negligible over the stated contact interval. These checks distinguish a physically consistent collision balance from a merely algebraic solution.

The impulse area and the momentum change must carry the same signed component. Keeping this sign visible is especially important when the collision reverses a velocity or when external impulses are resolved along more than one axis.

Force records and isolated systems

Force sensors rarely produce the ideal rectangle used in elementary impulse calculations. Contact force rises as surfaces deform, reaches one or more peaks, and falls as the bodies separate. The momentum change remains the integral of the net force component, so a force record requires its baseline, direction, and time interval have been identified. For a one-dimensional record,

J_{x} = t_{1} \int t_{2} F_{net, x} (t) d t = p_{f, x} - p_{i, x} .

The limits bracket the interaction under study. Extending them into a period of nonzero weight, thrust, or friction changes the impulse being measured. A sensor that reports the force exerted by one object on another records a contact force, not automatically the net force on either object. Its area equals the contact impulse; other force-time areas must be added before using the momentum theorem. Support forces matter for objects pressed against surfaces and for collisions long enough for gravity to contribute appreciable impulse.

Digitized force data are often evaluated by short time strips. If the samples are separated by $Δ t$ , the trapezoidal estimate approximates the interval between two successive readings the area $\frac{1}{2} (F_{k} + F_{k + 1}) Δ t$ . Summing those strips approximates the integral. Finer sampling improves the estimate only when the sensor bandwidth and calibration are adequate; a narrow unrecorded peak cannot be recovered by algebra. Newton seconds and kilogram metre per second are the same momentum unit.

The signed area also exposes cancellation within a record. During a catch, an upward hand force may exceed weight for a short interval while weight acts downward throughout. The upward contact area and downward gravitational area are separate terms. During a bounce, the contact force can remain in one direction while the incoming and outgoing momenta have opposite signs. There is no need for the force curve itself to change sign for the object's momentum to reverse.

A sampled contact-force trace is integrated between the marked limits. Each dot is a reading; the shaded area under the piecewise-linear interpolation approximates the contact impulse over the interval.

Isolated-system momentum conservation.

An isolated system has zero net external impulse over the specified interval. Its total momentum is therefore constant, component by component:

Δ P_{system} = \int F_{ext} d t = 0, P_{f} = P_{i} .

Isolation is an interval-dependent approximation, not a property permanently attached to an object. Two air-track gliders may form an effectively isolated horizontal system during their brief collision even though Earth pulls on them and the track constrains them vertically. Horizontal external impulse is then negligible while vertical external impulse is not. A puck on a rough table fails the same test over a long slide because friction has time to deliver a significant horizontal impulse. The selected components must be stated rather than assumed.

The centre of mass gives the geometric meaning of this result. For total mass $M$ , the system momentum satisfies $P = M V_{cm}$ . When external impulse vanishes, internal forces can redistribute momentum among pieces, eject fragments, or convert translational kinetic energy into deformation, yet they cannot alter the velocity of the centre of mass. In a motion diagram, the centre-of-mass marker moves in a straight line at constant velocity even while the individual carts accelerate violently during contact.

Mass-weighted position matters. A light object far from the rest of the system can shift the centre of mass as much as a heavy object close to it. The point need not lie inside a material object; for separated bodies it may occupy empty space between them. After an explosion from rest, fragments depart in different directions, but their mass-weighted centre remains at rest if external impulse is negligible. This is the same conservation statement as zero total final momentum.

Momentum conservation does not say that each object keeps its own momentum. During collision, the internal force on cart one supplies an impulse equal in magnitude and opposite in direction to the impulse on cart two. Their individual momenta change, often substantially; the sum stays fixed. A calculation that sets each cart's momentum unchanged has removed the mechanism of the collision. The conserved quantity belongs to the complete isolated system.

Collision energy and contact impulse

Momentum conservation and kinetic-energy conservation make different claims. For two objects on a line, the total kinetic energy separates into translational motion of the centre of mass and motion relative to it:

K = \frac{1}{2} M V_{cm}^{2} + \frac{1}{2} μ v_{rel}^{2}, μ = \frac{m _{1} m _{2}}{m _{1} + m _{2}} .

Here $v_{rel} = v_{1} - v_{2}$ is the signed relative velocity and $μ$ is the reduced mass. The first term depends only on total momentum. It cannot change in an isolated collision because $V_{cm}$ cannot change. The second term measures kinetic energy available in the centre-of-mass frame, where total momentum is zero. A perfectly inelastic collision leaves the objects with no relative motion, so this entire second term becomes internal energy.

In a passive collision, the kinetic-energy decrease is nonnegative. Deformation, heating, fracture, and sound account for it microscopically; they are not additional forces that spoil momentum conservation. In a superelastic event, stored chemical, elastic, or explosive energy can increase the relative kinetic energy while total momentum still remains fixed. The word inelastic does not identify a destination for all lost kinetic energy, only that mechanical kinetic energy is not conserved.

The laboratory-frame kinetic-energy change can be misleading when the whole system is moving. A pair of carts may retain a large common forward kinetic energy after sticking together, even though all kinetic energy associated with their approach has been dissipated. Subtracting the centre-of-mass translation isolates the part the collision can change. This is why the centre-of-mass frame gives the cleanest account of collision loss.

For the earlier $1.5 kg$ and $2.5 kg$ carts, total momentum is $3.5 kg m s^{- 1}$ , so $V_{cm} = 0.875 m s^{- 1}$ . Their relative speed before impact is $5.0 m s^{- 1}$ and the reduced mass is $0.9375 kg$ . The centre-of-mass kinetic energy available for dissipation is $\frac{1}{2} (0.9375) (5.0)^{2} = 11.7 J$ , exactly the loss found for their perfectly inelastic collision. The remaining $1.53 J$ is translation of the combined mass at the unchanged centre-of-mass velocity.

The initial kinetic energy splits into center-of-mass translation, which no internal force can remove, and a relative-motion part available for dissipation. A perfectly inelastic collision keeps only the translation term.

Extracting impulse from a contact-force history.

A contact-force history becomes a momentum result only after the measured channel is identified. A force plate beneath a ball reports the force of the plate on the ball; a load cell in a tether reports tension; an accelerometer reports neither force directly. The sign convention must match the coordinate used for velocity. For a vertical bounce with upward positive, the net force is commonly $F_{net, y} = N - m g$ ; $N$ alone omits the weight contribution. The baseline subtraction matters most for long contacts and small impulses.

Net impulse is the signed area between the net-force curve and the zero-force line over the contact interval. An offset in the instrument creates a false rectangular area whose size is the offset multiplied by the duration. Drift, trigger thresholds, and finite sampling can therefore produce a credible-looking but wrong impulse. A calibration run with no contact identifies a baseline; an independent change in velocity provides a stronger check, since the two estimates must agree within uncertainty.

Geometry gives an exact integral of a force record approximated by straight segments. A triangular pulse with peak force $F_{p}$ and width $Δ t$ has impulse $\frac{1}{2} F_{p} Δ t$ . A trapezoid contains the average of its two parallel sides times its width. The area must retain its sign. If the curve lies below the baseline, the impulse is opposite the positive coordinate. Several separate contacts require several intervals rather than one wide interval containing unrelated forces.

A force-time graph also separates impulse from peak force. Two catches can give the same momentum change while one spreads the area over ten times the duration and therefore has a much lower average force. That comparison only applies to the same net impulse; changing the rebound speed changes the required area. Reporting a maximum force without duration cannot determine the velocity change, and reporting only duration cannot establish either maximum force or impulse.

Vector collisions, recoil, and model checks

Planar collisions are solved from component equations, but the coordinate axes should serve the geometry of the problem. A convenient choice places one axis along a known incoming velocity or along a smooth surface. Every velocity is then resolved before the momentum equations are written. Speeds are nonnegative magnitudes; components carry the signs. An angle measured above the positive horizontal gives $v_{x} = v cos θ$ and $v_{y} = v sin θ$ , whereas an angle below it has a negative vertical component.

For two particles with negligible external impulse,

m_{1} v_{1 x, i} + m_{2} v_{2 x, i} = m_{1} v_{1 x, f} + m_{2} v_{2 x, f},

m_{1} v_{1 y, i} + m_{2} v_{2 y, i} = m_{1} v_{1 y, f} + m_{2} v_{2 y, f} .

The equations conserve the vector sum, while individual Cartesian magnitudes may change. They do not conserve an individual particle's horizontal and vertical momenta. In a glancing collision, a contact impulse transfers momentum along the local normal. A smooth contact supplies little tangential impulse, while a rough contact can change both components and create rotation. Those physical details determine which supplementary relations are legitimate after the two conservation equations have been written.

The incoming horizontal momentum equals the head-to-tail sum of the two outgoing momenta. The vertical components are equal and opposite and cancel, while the horizontal components add to the initial momentum.

Recoil is the momentum response of the remaining material when another part is expelled or when an internal energy release separates a system. The correct system contains every part whose momentum is counted after the event. Treating only the recoiling remnant as the system leaves the ejected material's impulse external and forbids momentum conservation for that remnant alone. During a short explosion, gravity and air resistance often contribute negligible impulse, but that estimate belongs to the stated time interval.

The initial kinetic energy was zero, yet the final kinetic energy is positive. That does not violate conservation of energy. An explosion converts stored internal energy into translational kinetic energy while preserving total momentum. The centre of mass remains at rest because the final momentum sum is zero. If the device were moving before separation, every fragment would share the centre-of-mass translation in addition to its motion relative to the centre of mass; total momentum would then equal total mass times that common centre-of-mass velocity.

Measured recoil can also identify an unmeasured ejection. If the remnant mass and velocity are known, its negative momentum is the vector sum of all expelled material's momentum, provided external impulse is negligible. This inference cannot uniquely identify several individual fragments without additional measurements, but it constrains their resultant. The same accounting applies to firearm recoil, molecular breakup, and spacecraft maneuvers: internal forces redistribute momentum; they do not generate net momentum for the full closed system.

Three fragment momenta from a device that explodes from rest sum to zero. The remnant momentum is the exact negative of the resultant of the two expelled pieces (dashed), so the center of mass stays at rest despite the fragments' motion.

Building and checking a momentum model.

A reliable momentum calculation begins with a physical statement that can be tested: the system, the interval, the coordinate axes, and the external impulses retained in the model. Those four choices determine the equation before any numbers are substituted. The same pair of bodies can be an isolated system during a short collision, a non-isolated system during a long slide, and part of a larger Earth-object system when a support impulse must be accounted for. Momentum conservation is therefore never a label attached permanently to a diagram. It is a claim about one boundary over one interval.

The interval should begin before the relevant interaction has supplied appreciable impulse and end after it has ended. An arbitrarily long interval can hide the collision inside unrelated gravity, drag, thrust, or support impulses. An interval that is too short can omit part of a contact force history. For a measured force record, the start and end times should be visible in the data or justified by the contact geometry. For a word problem, the stated event usually specifies those limits: impact, separation, explosion, capture, or a specified push.

The interaction stage fixes the duration over which external impulse is evaluated. A ball and bat can have a large internal contact impulse while the ball alone has a large external impulse. For the ball-bat pair, that same contact force is internal and cancels from the system total. Enlarge the boundary only when the momentum of every newly enclosed object is included.

With a fixed coordinate system, total initial momentum plus total external impulse equals total final momentum component by component. Drawing those three vectors as a head-to-tail relation exposes a missing sign or a neglected force: the polygon fails to close. In two dimensions, a result can satisfy the horizontal equation and still be wrong because the vertical residual has been ignored. The residual vector represents a required unmodelled impulse or a measurement error.

When external impulse is negligible relative to the momentum scale, the initial and final vectors coincide to the precision justified by the data. When it is retained, the final vector is displaced by that impulse. The same vector relation covers a falling object, a cart pushed by a fan, and a collision on a sloped track. A conservation equation may hold in one direction but fail in another when only one component of the external impulse is small.

A push, coasting interval, and bumper contact require separate momentum balances unless every external impulse across all three stages is included. The final momentum of one stage becomes the initial momentum of the next. Internal details of one stage need not be carried into the next once their net impulse has been recorded.

A negative momentum component is not an error if the chosen positive axis was retained consistently; it states that the cart now moves opposite the original direction. By contrast, reversing an arrow halfway through a calculation without changing the algebraic sign creates a hidden contradiction. Stage-by-stage balances prevent this mistake because every transition has an explicit initial state, external impulse, and final state.

Measurements impose another level of discipline. A velocity from video tracking is a displacement divided by a frame interval and has a finite uncertainty; a mass has calibration uncertainty; a force sensor can have an offset. A result quoted to many more digits than the inputs is not a stronger momentum balance. The relevant test is whether the measured momentum residual is smaller than the uncertainty propagated from masses, velocities, angles, and any external impulse estimate. A small residual supports the model; a persistent residual points to an omitted force, an unmeasured momentum carrier such as rotation or expelled material, or a sign and timing error.

Agreement among the assumptions, calculation, and diagnostic checks supports the model. Recalculate the final momentum vector from the reported velocities; compare it with the initial vector plus external impulse; then test any separate collision statement. An elastic model requires kinetic-energy agreement within uncertainty. A perfectly inelastic model requires a common final velocity. A recoil or explosion from rest requires a zero final momentum sum, even though its final kinetic energy is positive. Each test addresses a distinct physical claim.

The final report can keep those claims in separate rows. A passed momentum balance does not establish an elastic collision, and a kinetic-energy decrease does not identify the source of every missing channel.

Claim	Test quantity	Required agreement
Momentum balance	$p_{f} - p_{i}$	measured external impulse
Elastic collision	initial and final kinetic energy	stated uncertainty band
Perfectly inelastic collision	final object velocities	common final velocity
Recoil from rest	final vector sum	zero within measurement resolution

Keep the boundary, interval, and sign convention visible beside the equations. That record shows whether gravity was legitimately neglected, whether a wall impulse belongs in the balance, and whether a negative answer represents reversal or a bookkeeping mistake. An explicit model makes the momentum result reproducible and states the physical approximation on which it depends.

Linear momentum alone has clear limits. A spinning puck, a rolling wheel, or an off-centre impact can conserve translational system momentum while transferring substantial energy into rotation. The missing energy is not evidence against the momentum balance; it signals that kinetic energy has channels beyond centre-of-mass translation. Angular momentum and torque are then needed to determine the spins or to predict how the contact force was distributed. Similarly, a collision that ejects spray, chips, or exhaust is not fully represented by two visible bodies unless the escaped material is included as a third momentum carrier or its impulse is measured.

The model can be enlarged without abandoning the original calculation. Adding the floor and Earth converts a floor impulse from external to internal, but introduces Earth's extremely small recoil momentum. Adding a launcher and its expelled gas closes a rocket momentum balance, but requires the exhaust velocity in the same inertial frame. Each enlargement moves a force across the boundary and adds an object whose momentum must be tracked. Choose the smallest boundary for which the remaining external impulse is known or negligible at the required precision.

The hierarchy avoids two opposite mistakes: declaring conservation where an unmeasured external impulse is decisive, and constructing a needlessly enormous system whose additional momenta obscure the question. The momentum theorem remains exact for any complete system. Approximation enters through the selected boundary, the selected interval, and the decision to neglect or estimate the external impulse crossing that boundary.

External impulse changes the accounting in a defined way. If a known impulse acts during separation, add that vector to the initial momentum before equating the final fragment sum. The centre-of-mass velocity changes by external impulse divided by total mass, component by component. Internal explosion forces still cancel in the system sum, even when they are much larger than the external force at an instant.