Linear Momentum & Impulse

Catch a cricket ball with rigid hands and it stings like a slap; ride your hands back as it arrives and the same ball settles in painlessly. Catch an egg without the ride and you are cleaning the floor. In every case the ball or egg loses exactly the same motion — same mass, same speed, same change — yet the force varies tenfold with how long you take. Whatever "amount of motion" the catch removes, it can be paid for like a bill: in one brutal instalment or many gentle ones. That amount of motion is momentum, $p = mv$ , and the payment is impulse, force × time. Their exchange rate, $F\Delta t = \Delta p$ , is the single most practical equation in this unit: it designs airbags and crumple zones, explains follow-through in tennis and boxing, tells gymnasts to bend their knees, and prices every collision an exam can throw at you. It is also the opening move of the momentum chapter — because once you can track how momentum enters and leaves one body, the next article's miracle becomes visible: between two bodies, momentum never enters or leaves at all, only changes hands. You have already met force as the rate of momentum change ( $F = \Delta p/\Delta t$ , the general second law). This article runs that law in integrated form — impulse as the area under the force–time story — and drills the two skills exams actually test: the cushioning trade-off, and the vector bookkeeping of bounces, where a ball that merely turns around has had its momentum changed by double.

Key takeaways

Momentum $\vec p = m\vec v$ is the directed "quantity of motion" — how much force-time it takes to stop a body, with direction attached. Units $\mathrm{kg\,m/s}$ .
Impulse is the force–time product $\vec J = \vec F_{\text{avg}}\,\Delta t$ — the area under the F–t graph — and it buys momentum change exactly: $\vec J = \Delta\vec p$ .
The cushioning trade: $\Delta p$ is fixed by the situation, so $F_{\text{avg}} = \Delta p/\Delta t$ — stretch the time tenfold, shrink the force tenfold. Airbags, bent knees, follow-through.
Reversal costs double: a ball bouncing back at the same speed takes impulse $2mv$ ; merely stopping takes $mv$ . Signs are the whole game.
Oblique bounce off a smooth wall: only the normal component reverses — $J = 2mv\cos\theta$ along the normal; the along-wall component sails through untouched.
During a violent, brief impact, gravity's impulse $mg\,\Delta t$ is negligible beside the contact spike — that is the impulsive approximation, and it has a legality check.
Key idea: impulse equals change in momentum — the momentum change is non-negotiable, but the force–time price of it is yours (or the engineer's) to negotiate.

The core insight

Impulse — average force times contact time, the area under the force–time graph — equals the change in momentum, $\vec F_{\text{avg}}\Delta t = \Delta \vec p = m\vec v_f - m\vec v_i$ , so for a fixed momentum change the force and the time trade off in exact inverse proportion, and every sign and component of the vectors must be honoured.

Where this lives today

The impulse–momentum theorem is the quiet law of the crash-test industry. Accelerometers in a test dummy integrate force through the milliseconds of impact and report exactly the $\Delta p$ audit of derivation A; safety standards are written as limits on that curve. Airbag controllers make the same bet in real time — sensing the deceleration spike, they must inflate within about twenty milliseconds to stretch the occupant's momentum change over the longest possible time, then deflate through vents during the impact so the bag keeps giving way (a bag that stayed rigid would be a stiff catch). Sport plays the equation in both directions: follow-through in tennis, golf and cricket keeps the ball on the strings or blade longer, pumping in more impulse for the same peak force; gloves, pads and helmets run it the other way. And at the far end of scale and violence, spacecraft engineers designing Mars-lander airbags and naval engineers rating ship fenders are doing the gymnast's knee-bend arithmetic with tonnes in place of kilograms — same bill, negotiated payment.

The physical picture

Momentum measures how hard something is to stop — with a memory of direction. A loaded lorry at walking pace and a rifle bullet can carry the same $mv$ : what they share is the total force-times-time needed to bring either to rest. That product is the natural currency of impacts, because during a real collision the force is a violent spike lasting milliseconds — no ruler or stopwatch in an exam problem can follow it. So don't follow it. The total of the spike — its area on a force–time graph — is pinned by the endpoints alone: it must equal the momentum change it produced. Skip the mayhem; audit the before and after.

Figure

Two ways to pay the same bill — equal areas, different stories

This one picture prices every safety device you own — and every "why did it hurt?" in sport. Your momentum arriving at the ground after a jump is fixed: mass times landing speed. Land stiff-legged and you compress the payment into twenty milliseconds; bend your knees and the same area spreads across three hundred. A helmet's foam, a car's crumple zone, the sand in a long-jump pit: none of them reduce your momentum change by a single unit. They stretch the time, and the peak force — the thing that breaks bones — falls in proportion. The second half of the picture is the vector discipline. Momentum keeps direction, so turning motion around changes it more than stopping does. A ball arriving at $+mv$ and leaving at $-mv$ has changed by $2mv$ — the wall first absorbed the ball's momentum and then loaded it back in reverse. Miss the sign and every bounce problem on the paper comes out half right, which in an MCQ is fully wrong.

The math: momentum, impulse, and the exchange rate

Start with the currency. What makes a moving body hard to stop? Not mass alone (dust at bullet speed is harmless), not speed alone (a glacier is not). The honest measure multiplies them:

\vec p = m\vec v \qquad (1)

where: $m$ — mass in kg; $\vec v$ — velocity, direction included; $p$ in $\mathrm{kg\,m/s}$ . The vector nature is load-bearing: two identical balls moving at $+5$ and $-5\ \mathrm{m/s}$ carry zero total momentum. Compare kinetic energy, its scalar cousin: same two balls carry plenty of that. The two quantities answer different questions and part company completely in collisions — the warning below has the details. Now the exchange rate. The general second law says force is the rate of momentum change, $\vec F = \Delta\vec p/\Delta t$ . Multiply both sides by the duration and the rate statement becomes a total statement — the impulse–momentum theorem:

\vec J = \vec F_{\text{avg}}\,\Delta t = \Delta\vec p = m\vec v_f - m\vec v_i \qquad (2)

where: $\vec J$ — impulse, in $\mathrm{N\,s}$ (identical to $\mathrm{kg\,m/s}$ — same currency, different notation); $\vec F_{\text{avg}}$ — the average force over the contact; $\Delta t$ — the contact time. Read it physically: the left side is what the interaction did (its area on the F–t graph), the right side is what the body shows for it. When the force varies — and in impacts it always does — the product $F_{\text{avg}}\Delta t$ is defined as that area; for the straight-line and rectangular graphs exams draw, the area is triangle-and-rectangle arithmetic, no calculus needed. Rearranged, (2) becomes the design equation of the safety industry:

\vec F_{\text{avg}} = \frac{\Delta \vec p}{\Delta t} \qquad (3)

where: everything above, solved for the thing that injures. $\Delta p$ is set by the crash; $\Delta t$ is engineerable. Ten times the stopping time, one tenth the average force — exactly, not approximately. This is the equation to quote when a question asks why eggs survive soft catches, why boxers roll with punches, why cars are designed to crumple.

the phrase in the problem	the move
"caught / stopped in time $\Delta t$ "	$F_{\text{avg}} = mv/\Delta t$
"rebounds at the same speed"	$\Delta p = 2mv$ — reversal doubles
"rebounds at speed $u$ " (head-on)	$\Delta p = m(u + v)$ — speeds add because directions oppose
"strikes a smooth wall at angle θ to the normal"	normal component reverses: $J = 2mv\cos\theta$ ; tangential survives
"F–t graph given"	impulse = signed area; then $\Delta v = \text{area}/m$
"hit lasts milliseconds — include gravity?"	compare $mg\Delta t$ to the contact impulse; usually negligible

Derivation AThe impulse–momentum theorem — and why the area rule works without calculus

We derive equation (2) from the second law, first for a constant force in two lines, then for a varying force by the slicing argument that legitimises "impulse = area under the F–t graph". The strategy: the second law is a statement about rates; multiplying a rate by its time converts it to a total, and a varying rate is handled by adding many brief constant-rate instalments. Watch how the final statement contains no trace of the force's shape — only its area — which is precisely why violent, unmeasurable collision forces can still be audited exactly.

Step 1. Constant net force on mass $m$ : acceleration $a = F/m$ is constant, so after time $\Delta t$ , $v_f = v_i + a\Delta t$ . Multiply through by $m$ :

mv_f = mv_i + F\,\Delta t \;\;\Rightarrow\;\; F\,\Delta t = \Delta p

Step 2. Varying force: slice the contact into many slivers so brief that $F$ is effectively constant within each. Each sliver delivers its own small instalment, $F_i\,\delta t_i = \delta p_i$ . Why slicing is legal: the second law holds at every instant; within a short enough sliver, "the force now" and "the average force in this sliver" cannot be told apart.

Step 3. Add the instalments. The right side telescopes — each sliver's final momentum is the next one's initial — leaving just $\Delta p$ overall. The left side, a sum of (height × width) strips, is exactly the area under the F–t graph:

\vec J = \text{area under } F\text{–}t = \Delta\vec p

Why the shape drops out: two different force stories with equal areas produce identical momentum changes — the hard and soft catches of the hero figure are indistinguishable to the final ledger.

Step 4. Define the average force as the constant force that would do the same job in the same time: $F_{\text{avg}} = J/\Delta t$ — the rectangle of equal area drawn over the same base. Verified: for a $0.15$ kg ball stopped from $20$ m/s, $\Delta p = 3.0\ \mathrm{kg\,m/s}$ regardless of catch style; a $0.10$ s catch gives $F_{\text{avg}} = 30$ N, a $0.01$ s catch $300$ N — the tenfold trade, matching equation (3). Slicing argument as in NCERT Ch. 5's area treatment; the calculus version $J = \int F\,\mathrm{d}t$ is the UG deep dive's business.

Go deeper · Derivation B · 1 minThe bounce ledger — head-on doubling and the oblique wall

Derivation BThe bounce ledger — head-on doubling and the oblique wall

We derive the impulse for the two rebound geometries every exam draws: head-on (reversal doubles the bill) and oblique off a smooth wall (only the normal component pays). The strategy: apply $\vec J = \Delta\vec p$ component by component, letting the signs do the physics. Watch the smooth-wall assumption work: "smooth" means the wall can push only along its normal, so the along-wall momentum component has nothing acting on it and must survive unchanged.

Step 1. Head-on bounce, speed in $v$ , speed out $u$ , taking outgoing as positive: $p_i = -mv$ , $p_f = +mu$ .

J = p_f - p_i = mu - (-mv) = m(u + v)

Why the speeds add: the wall must first cancel $mv$ of incoming motion, then supply $mu$ of outgoing motion — two jobs in the same direction. Equal-speed rebound ( $u = v$ ) gives the classic $J = 2mv$ .

Step 2. Oblique bounce: ball arrives at speed $v$ at angle $\theta$ to the wall's normal, leaves at the mirror angle with speed unchanged (a smooth, perfectly bouncy wall). Resolve into components: normal $v\cos\theta$ , tangential $v\sin\theta$ .

Step 3. The smooth wall exerts force only along its normal. Tangential ledger: no force, no impulse — $mv\sin\theta$ in, $mv\sin\theta$ out, untouched. Normal ledger: $mv\cos\theta$ in, $mv\cos\theta$ out the other way — a head-on bounce at reduced stakes:

J = 2mv\cos\theta, \quad \text{directed along the normal, away from the wall}

Step 4. The force on the wall is equal and opposite (third law): magnitude $2mv\cos\theta/\Delta t$ on average, pointing into the wall along the normal — never along the ball's path. Verified: limits — $\theta = 0$ (head-on) recovers $2mv$ ✓; $\theta \to 90°$ (grazing) gives $J \to 0$ ✓ (the ball barely notices the wall). Numbers for worked example 3: $2(0.10)(10)\cos 30° = 2 \times 0.866 = 1.73\ \mathrm{N\,s}$ ✓.

Figure

Oblique bounce off a smooth wall — only the normal component pays

One more piece of legal fine print — the impulsive approximation. During a 5-millisecond bounce, does gravity's tug matter? Its contribution to the ledger is $mg\Delta t$ : for a tennis ball, $0.06 \times 9.8 \times 0.005 \approx 0.003\ \mathrm{N\,s}$ , against a contact impulse near $2\ \mathrm{N\,s}$ — a 0.15% correction, rightly ignored. The rule: during a brief violent contact, only the spiking force counts; steady forces (gravity, friction) get their turn before and after. The legality check is always the same division, and long-contact problems (a ball rolling up a cushioned ramp for half a second) can fail it — then gravity's impulse must be counted.

Limiting cases

case	result	meaning
stop, no rebound	$J = mv$	the plain bill
equal-speed rebound	$J = 2mv$	reversal doubles: stop, then relaunch
rebound at $u < v$	$J = m(u+v)$	real bounces pay between $mv$ and $2mv$
grazing incidence $\theta \to 90°$	$J \to 0$	the wall is barely consulted
$\Delta t \to$ long	$F_{\text{avg}} \to$ small	cushioning; gravity's impulse may need counting
$\Delta t \to 0$ at fixed $\Delta p$	$F_{\text{avg}} \to \infty$	rigid impacts spike the force — the breaking regime

Common trapMomentum and kinetic energy are basically the same thing

They are not even the same kind of thing, and the exam knows it. Momentum $m\vec v$ is a vector, linear in speed; kinetic energy $\tfrac12mv^2$ is a scalar, quadratic. The teeth: two identical clay balls at $+5$ and $-5$ m/s carry zero total momentum but $25$ J of kinetic energy — they collide, stick, and stop dead: momentum perfectly conserved (zero before, zero after), every joule of KE gone to heat and deformation. The conflation sells MCQs in both directions. A ball bouncing elastically off a wall keeps its kinetic energy unchanged while its momentum changes by $2mv$ — so "nothing changed, the speed is the same" is wrong about momentum. And a bullet embedding in a block conserves momentum across the crunch while KE collapses — so carrying $\tfrac12mv^2$ through a sticking collision is wrong about energy. File them separately: momentum is the collision-crossing currency (next article makes this exact); kinetic energy is the ledger that pays for damage.

Worked examples

1 · Board level — the bouncing ball's bill (CBSE/ICSE pattern, 3 marks)

A ball of mass $0.20\ \mathrm{kg}$ hits the floor at $6.0\ \mathrm{m/s}$ and rebounds at $4.0\ \mathrm{m/s}$ . Contact lasts $0.050\ \mathrm{s}$ . Find the impulse on the ball and the average net force during contact.

Step 1. Signs first — take upward positive: $p_i = -0.20 \times 6.0 = -1.2\ \mathrm{kg\,m/s}$ ; $p_f = +0.20 \times 4.0 = +0.8\ \mathrm{kg\,m/s}$ .

Step 2. Impulse $J = \Delta p = 0.8 - (-1.2) = \boxed{2.0\ \mathrm{N\,s}}$ , directed upward. (Not $0.4$ : the speeds add because the direction reversed.)

Step 3. Average net force: $F = J/\Delta t = 2.0/0.050 = \boxed{40\ \mathrm{N}}$ upward.

Step 4. For the last mark, the fine print: this $40$ N is the net force; the floor itself pushes $F_{\text{floor}} = 40 + mg = 42$ N. (Check the approximation: gravity's share, $mg\Delta t \approx 0.1\ \mathrm{N\,s}$ , is 5% here — a borderline case worth the sentence.)

2 · Board level — why you bend your knees (IGCSE/AP reasoning-plus-numbers pattern)

A $60\ \mathrm{kg}$ gymnast lands at $5.0\ \mathrm{m/s}$ . Compare the average force on her legs if she stops (a) stiff-legged in $0.020\ \mathrm{s}$ , (b) bending her knees over $0.30\ \mathrm{s}$ . ( $g = 10\ \mathrm{m/s^2}$ ; ignore gravity's impulse during the stop.)

Step 1. The bill is fixed: $\Delta p = 60 \times 5.0 = 300\ \mathrm{kg\,m/s}$ , whichever landing she chooses.

Step 2. Stiff: $F = 300/0.020 = \boxed{15\,000\ \mathrm{N}}$ — twenty-five times her weight ( $mg = 600$ N), the bone-fracture regime.

Step 3. Bent knees: $F = 300/0.30 = \boxed{1000\ \mathrm{N}}$ — under twice her weight.

Step 4. The sentence that earns the reasoning mark: the momentum change cannot be reduced, but stretching the stopping time by 15× reduces the average force by exactly 15× — the same trade bought by crash mats, airbags and sand pits.

3 · Competitive level — the oblique wall (JEE-Main pattern, ~2.5 minutes)

A $0.10\ \mathrm{kg}$ ball moving at $10\ \mathrm{m/s}$ strikes a smooth wall at $30°$ to the <strong>normal</strong> and rebounds at the same speed and mirror angle, contact time $0.020\ \mathrm{s}$ . Find the impulse on the ball and the average force on the wall.

Step 1. Resolve about the wall: normal component $v\cos 30° = 8.66$ m/s reverses; tangential $v\sin 30° = 5.0$ m/s survives (smooth wall — no tangential force possible).

Step 2. Impulse, normal direction only: $J = 2mv\cos\theta = 2(0.10)(10)(0.866) = \boxed{1.73\ \mathrm{N\,s}}$ , along the outward normal.

Step 3. Average force on the ball: $1.73/0.020 = 86.6$ N along the outward normal; on the wall, by the third law, $\boxed{86.6\ \mathrm{N}}$ into the wall, along the normal.

Step 4. MCQ audits: the answer must vanish as $\theta \to 90°$ (graze) ✓ and hit $2mv/\Delta t = 100$ N at $\theta = 0$ ✓; and the direction is along the normal, never along the incoming path — a drawn-option trap. Timing note: ~2.5 min. Everything is decided the moment you resolve along and perpendicular to the wall — do that first, always. Watch the wording: " $30°$ to the wall" and " $30°$ to the normal" are complementary angles, and both phrasings are in circulation; the cosine belongs with the angle from the NORMAL.

4 · Competitive level — the crumple-zone estimate (NEET/JEE reasoning pattern, ~3 minutes)

A $1200\ \mathrm{kg}$ car hits a barrier at $15\ \mathrm{m/s}$ and stops. Estimate the average force on the car if it (a) crumples through $0.90\ \mathrm{m}$ , (b) hits rigidly and stops in $0.05\ \mathrm{m}$ . (Hint: with steady deceleration, stopping time $\Delta t = 2s/v$ .)

Step 1. The bill: $\Delta p = 1200 \times 15 = 18\,000\ \mathrm{kg\,m/s}$ .

Step 2. (a) Crumple: $\Delta t = 2(0.90)/15 = 0.12$ s, so $F = 18\,000/0.12 = \boxed{1.5 \times 10^5\ \mathrm{N}}$ — about $12.5g$ of deceleration, survivable when belted.

Step 3. (b) Rigid: $\Delta t = 2(0.05)/15 = 0.0067$ s, so $F = 18\,000/0.0067 = \boxed{2.7 \times 10^6\ \mathrm{N}}$ — $18$ times larger, deep in the fatal range.

Step 4. Cross-check by the energy route (Force-vs-Energy dispatch: average force also equals $\text{KE}/s$ ): $\tfrac12(1200)(15^2)/0.90 = 1.5\times10^5$ N ✓ — the two books agree, as they must for uniform deceleration. Timing note: ~3 min. The $\Delta t = 2s/v$ move (distance ÷ average speed $v/2$ ) is the standard bridge from a crumple distance to a stopping time — memorize it; it appears in every safety-estimate problem. Note the design lesson in the ratio: force scales inversely with crumple distance, which is why centimetres of steel deformation are worth lives.

Exam traps

Half the bounce. $\Delta p$ for a reversal is $m(u+v)$ , not $m(v-u)$ — the speeds add when the direction flips. The seller's version: a ball bouncing at equal speed "has no momentum change since the speed is unchanged" — wrong twice (it's a vector; it changed by $2mv$ ).

Impulse along the path instead of the normal. Off a smooth wall the impulse is along the normal: the tangential component survives untouched. Options drawn along the incoming or outgoing path are traps.

Angle to the wall vs angle to the normal. Complementary angles, both used by setters. The $\cos$ goes with the angle measured from the normal. Re-read the sentence before resolving.

Peak force for average force. The F–t spike's peak can be several times $J/\Delta t$ . Questions asking "average force" want the equal-area rectangle; if a graph is given and the peak is asked, read the graph, don't divide.

Forgetting the net-vs-contact distinction. $J/\Delta t$ gives the net average force; the floor or bat supplies that plus $mg$ (if the contact is long enough for gravity to matter — run the $mg\Delta t$ check rather than guessing).

Signed areas on F–t graphs. Area below the time axis is negative impulse. A force that spikes positive then negative can leave $\Delta p = 0$ with plenty of drama in between.

"Momentum is conserved for the ball." During a bounce the ball alone is not an isolated system — the wall's impulse is exactly what changes its momentum. Conservation belongs to the ball-plus-wall(-plus-Earth) system, next article's business. Quote the impulse–momentum theorem for one body, conservation only for closed systems.

Units. $\mathrm{N\,s}$ and $\mathrm{kg\,m/s}$ are the same thing — write either, but not $\mathrm{N/s}$ (a real option on real papers). And $g$ : CBSE/ISC $9.8$ , JEE/NEET usually a stated $10$ , A-Level $9.81$ — as ever, the paper's word is law.

Origins · 1 minRené Descartes and how this was discovered

René Descartes

1596–1650 · La Flèche, Amsterdam, Stockholm

Descartes made the first great bet on momentum: in his Principia Philosophiae (1644) he argued that God had endowed the universe with a fixed total "quantity of motion" — mass times speed — that no collision could alter. It was the right instinct with a fatal flaw: he reckoned motion as an unsigned quantity, so his collision rules predicted nonsense (two equal balls meeting head-on carry plenty of Cartesian "motion" but zero of the real thing). The repair came from Christiaan Huygens and the Royal Society's 1668 collision inquiry: treat motion as directed, and the books balance in every experiment. Newton then grounded the balanced books in his third law and made $m\vec v$ — signed, vectorial — the central currency of the Principia. The arc from Descartes' beautiful wrong idea to Newton's exact ledger is the arc of this article: momentum only works if you keep the signs.

History

1644

— Descartes proposes conservation of "quantity of motion,"

m|v|

— right instinct, missing sign.

1668

— Wallis, Wren and Huygens answer the Royal Society's collision challenge; Huygens' directed momentum makes the accounts balance.

1687

— Newton's Principia defines force as the rate of change of momentum and grounds the bookkeeping in the third law.

1950

s — crash-test instrumentation turns the impulse–momentum theorem into automotive law: crumple zones are legislated areas under force–time curves.

1998

— airbag controllers fire within ~20 ms of impact — real-time engineering of Δt, straight from equation (3).

Going further

Texts. NCERT Physics Class 11, Ch. 5 (Laws of Motion — impulse) · HC Verma, Concepts of Physics Vol. 1, Ch. 9 · Morin, Introduction to Classical Mechanics, Ch. 4 — and the momentum chapters of any JEE archive, where the oblique-wall and F–t-graph patterns repeat annually.

Provenance. All results are standard. A verifier should recompute: the theorem via the slicing argument; the bounce ledgers $m(u+v)$ and $2mv\cos\theta$ ; Example 1 ( $J = 2.0\ \mathrm{N\,s}$ , net $40$ N, floor $42$ N); Example 2 ( $15\,000$ vs $1000$ N from $\Delta p = 300$ ); Example 3 ( $1.73\ \mathrm{N\,s}$ , $86.6$ N along the normal); Example 4 ( $1.5\times10^5$ vs $2.7\times10^6$ N, energy cross-check agreeing); the tennis-ball legality check ( $mg\Delta t \approx 0.003\ \mathrm{N\,s}$ vs $\sim 2\ \mathrm{N\,s}$ ).

Prerequisites. Momentum & the General Second Law (F = dp/dt) · Newton's Second Law · Vector Components. Leads to. Conservation of Linear Momentum · Inelastic Collisions & Coefficient of Restitution · The Bouncing Ball (archetype) · Linear Momentum & Impulse (UG deep dive).

Is impulse a force? The units and the name both confuse me.

It is not a force — it is a force's accumulated effect over time, the way distance is speed's accumulated effect. A force of 100 N tells you the rate momentum is flowing right now; an impulse of 100 N·s tells you the total that flowed. The unit N·s unpacks as $\mathrm{kg\,m/s^2 \times s = kg\,m/s}$ — exactly momentum's unit, as it must be, since impulse is a momentum change. When a question asks for a force, divide the impulse by the time; when it asks "which delivered more impulse", compare areas, not heights.

Why can I ignore gravity during a bounce but not during the flight before it?

Because impulse is force × time, and the bounce gives gravity almost no time. During a 5 ms contact, gravity contributes $mg \times 0.005$ — thousandths of a N·s — while the contact force, spiking to hundreds of times the weight, delivers the full $\sim 2mv$ . Before and after the bounce, gravity has the stage to itself for whole seconds and completely controls the motion. The discipline: don't memorize "ignore gravity in collisions"; run the comparison $mg\Delta t$ vs the contact impulse once, so you'll recognize the rare long-contact problem where the approximation fails.

How does follow-through actually make the ball go faster? The bat isn't pushing harder.

Impulse is area: height × width. Follow-through doesn't raise the force (your muscles set that ceiling); it widens the time the bat and ball stay in contact, so the same force acts through more milliseconds and deposits more momentum. Stopping the swing at contact is cropping the area's right-hand edge. The same logic runs a javelin thrower's long final pull and a rocket engine's long burn: when force is capped, time is the only knob left.

The ball's momentum changed during the bounce. I thought momentum was conserved — which is it?

Both, once the system is named. The ball alone is not isolated during the bounce — the wall hammers it with an external impulse, and its momentum duly changes by exactly that impulse (this article's law). The ball plus the wall-and-Earth is isolated — the wall's push on the ball and the ball's push on the wall are a third-law pair — so the total is conserved: the Earth picks up the $2mv$ the ball handed over, at an utterly invisible velocity ( $2mv/M_{\text{Earth}} \sim 10^{-24}$ m/s). "Conserved" is always a claim about a named system with no external force; the next article builds everything on that discipline.

Can an impulse change the direction of motion without changing the speed?

Yes — that is precisely what a perpendicular impulse does, and the elastic oblique bounce is the worked case: speed in equals speed out, yet the impulse $2mv\cos\theta$ is real and the wall genuinely got shoved. The extreme version is uniform circular motion: the string delivers a continuous stream of tiny impulses, each perpendicular to the velocity, endlessly redirecting a momentum vector of constant magnitude. Momentum is a vector; "changed" includes "turned". Kinetic energy, blind to direction, is what stays constant in these cases — one more reason the two ledgers must be kept apart.

Why do problems give contact times in milliseconds — who measures those?

High-speed cameras and piezoelectric force plates, routinely. A tennis ball stays on the strings ~4–5 ms; a golf drive is under half a millisecond; a football kick ~10 ms; a car crash 100–150 ms. Exam values are realistic. And the measurements matter commercially: racket and club designers tune contact time (through string tension and face flex) precisely because the impulse–momentum theorem says time-on-ball is money. When a problem's contact time looks absurd — a "0.5 s" cricket catch — that's the setter signalling a soft catch, and often that gravity's impulse deserves a look.

Is J = FΔt exact or an approximation when the force varies?

Exact, provided $F$ means the average force over the interval — that is the definition of the average. What would be an approximation is substituting the peak force, or a guessed "typical" force, for the average. The area statement is the honest general form: impulse = area under the F–t curve, computed by geometry for the shapes exams draw (rectangles, triangles, trapezia) and by integration in the UG deep dive. Every exam use falls into one of two patterns: given $\Delta t$ and asked for $F_{\text{avg}}$ (divide), or given the graph and asked for $\Delta p$ (find the area, respecting signs).