Dot & Cross Products

Push a stalled car from behind and every newton goes into motion; push it sideways and you achieve nothing. Now grab a spanner: push the handle along its length and the bolt ignores you; push at right angles and it turns. Two situations, two opposite morals — and nature keeps score of both with two different kinds of vector multiplication. The dot product asks: how much do these two vectors cooperate? It rewards alignment, dies at 90°, and hands back a single signed number — which is why work is a dot product. The cross product asks the opposite: how much do these two vectors twist against each other? It rewards perpendicularity, dies for parallel vectors, and hands back a brand-new vector along the axis of the twist — which is why torque is a cross product. Every board asks you to compute both; JEE and NEET also test whether you know which product a situation is silently using — the difference between sin⁡θ\sin\theta and cos⁡θ\cos\theta here is a full mark, several times per paper. This module builds both products from their geometry, derives both component formulas honestly, fixes the right-hand rule before the exam hall does, and drills the classics — including the one-line tan⁡θ\tan\theta trick that settles half the vector-product MCQs ever set.

Key takeaways
  • The scalar (dot) product A⃗⋅B⃗=ABcos⁡θ\vec A \cdot \vec B = AB\cos\theta measures cooperation: maximum when parallel, zero when perpendicular, negative past 90°. Output: a number.
  • The vector (cross) product has magnitude ∣A⃗×B⃗∣=ABsin⁡θ|\vec A \times \vec B| = AB\sin\theta — the parallelogram area the vectors span — and direction perpendicular to both by the right-hand rule. Output: a vector.
  • Component recipes: A⃗⋅B⃗=AxBx+AyBy+AzBz\vec A\cdot\vec B = A_xB_x + A_yB_y + A_zB_z; the cross product comes from the cycle i^×j^=k^\hat{i}\times\hat{j}=\hat{k}, packaged as a 3×33\times3 determinant.
  • The dot is commutative (A⃗⋅B⃗=B⃗⋅A⃗\vec A\cdot\vec B = \vec B\cdot\vec A); the cross is anti-commutative (B⃗×A⃗=−A⃗×B⃗\vec B\times\vec A = -\vec A\times\vec B) — order costs a sign, and the sign is physics (the sense of rotation).
  • Flagship uses: work W=F⃗⋅d⃗W = \vec F\cdot\vec d and power P=F⃗⋅v⃗P = \vec F\cdot\vec v (dot); torque τ⃗=r⃗×F⃗\vec\tau = \vec r\times\vec F and angular momentum L⃗=r⃗×p⃗\vec L = \vec r\times\vec p (cross).
  • Key idea: the dot product extracts the parallel part of one vector along another and reports a scalar; the cross product extracts the perpendicular part and reports a vector along the twist axis — cosine for cooperation, sine for rotation.
The core insight

Two vectors multiply two ways: A⃗⋅B⃗=ABcos⁡θ\vec A\cdot\vec B = AB\cos\theta is a scalar measuring how much they cooperate along a line, while A⃗×B⃗=ABsin⁡θ n^\vec A\times\vec B = AB\sin\theta\,\hat n is a vector measuring how much they twist about the perpendicular axis n^\hat n — and choosing the right product for the physics is itself the exam skill.

Where this lives today

Every frame of every 3D game and film render is a festival of both products. Graphics engines cross two edges of each triangle to get its surface normal — the perpendicular vector Derivation B computes — then dot that normal with the direction to the light to set the triangle's brightness: cos⁡θ\cos\theta shading, computed billions of times per second on a GPU. The operations you practised above are the lighting model of the visual world on your screen. In the physical world, every electric motor and generator on Earth is an engineering of qv⃗×B⃗q\vec v\times\vec B — charges pushed sideways by magnetic fields — and fusion devices like ITER are vessels shaped so that the cross product of particle velocities with a twisted magnetic field keeps a 150-million-kelvin plasma off the walls. The spanner and the sled rope of our opening scale up without changing a single formula.

The physical picture

Draw A⃗\vec A and B⃗\vec B tail-to-tail with angle θ\theta between them. For the dot product, drop a perpendicular from the tip of B⃗\vec B onto the line of A⃗\vec A: the shadow has length Bcos⁡θB\cos\theta, and the dot product is AA times that shadow — the overlap of the two arrows, reported as one number. For the cross product, complete the parallelogram the arrows span: its area, ABsin⁡θAB\sin\theta, is the cross product's magnitude — biggest when the vectors are square, zero when they fold onto one line — and because a tilted plane in 3D needs an orientation, the result is delivered as a vector standing perpendicular to the parallelogram.

Figure
Two multiplications — the projection number and the area vector
DOT: a number A B shadow B cos θ θ A·B = A (B cos θ) — overlap CROSS: a vector A B area = AB sin θ θ A×B — out of the plane |A×B| = AB sin θ, direction ⊥ both

Why does physics need both? Because forces do two jobs. Along the motion a force feeds or drains energy — a scalar "overlap" question, answered by the dot. Off the line to a pivot a force turns things — and a turning has an axis and a handedness, so the answer must be a vector, delivered by the cross. The direction needs one more convention: the parallelogram has two perpendiculars, and we must all agree on one. The agreement is the right-hand rule — point the right hand's fingers along the first vector A⃗\vec A, curl them toward B⃗\vec B through the smaller angle, and the thumb gives A⃗×B⃗\vec A\times\vec B.

Figure
The right-hand rule — and why order costs a sign
curl A into B: thumb UP A (first) B (second) fingers curl A×B curl B into A: thumb DOWN A (second) B (first) curl reversed B×A = −A×B

The right panel holds the deepest fact in this module: swapping the order of a cross product reverses the answer. Turning a bolt from A⃗\vec A toward B⃗\vec B is the opposite rotation from B⃗\vec B toward A⃗\vec A, and opposite rotations have opposite axes — B⃗×A⃗=−A⃗×B⃗\vec B\times\vec A = -\vec A\times\vec B is handedness written in symbols.

The scalar product: cooperation as a number

Start with the product that behaves. The scalar (dot) product is defined by the tail-to-tail angle:

A⃗⋅B⃗=ABcos⁡θ(1)\vec A \cdot \vec B = AB\cos\theta \qquad (1)

where: AA, BB — the magnitudes; θ\theta — the tail-to-tail angle (0∘≤θ≤180∘0^\circ \le \theta \le 180^\circ); output — a scalar, positive below 90∘90^\circ, zero at 90∘90^\circ, negative beyond. Read it as a projection: Bcos⁡θB\cos\theta is the shadow of B⃗\vec B on A⃗\vec A's line, so A⃗⋅B⃗\vec A\cdot\vec B is "length of A⃗\vec A times how much of B⃗\vec B agrees with it" — and projecting the other way gives the same number, which is why A⃗⋅B⃗=B⃗⋅A⃗\vec A\cdot\vec B = \vec B\cdot\vec A. Three properties come free from the geometry, each an exam tool: perpendicular vectors have zero dot product (the fastest perpendicularity test there is); a vector dotted with itself gives A⃗⋅A⃗=A2\vec A\cdot\vec A = A^2 (how you find lengths of messy vector sums); and the product distributes over addition, because the shadow of a sum is the sum of the shadows. Distributivity is the engine behind the component recipe. When vectors arrive as components, no protractor is needed:

A⃗⋅B⃗=AxBx+AyBy+AzBz(2)\vec A\cdot\vec B = A_xB_x + A_yB_y + A_zB_z \qquad (2)

where: Ax,Ay,AzA_x, A_y, A_z and Bx,By,BzB_x, B_y, B_z — the rectangular components. Multiply matching components, add, done. No angle appears because the axes have already done the trigonometry — (2) is derived from (1) below, and running it backwards through cos⁡θ=A⃗⋅B⃗/AB\cos\theta = \vec A\cdot\vec B/AB is the standard exam route to the angle between two given vectors. The flagship application is on its way in this track: work, W=F⃗⋅d⃗W = \vec F\cdot\vec d, is a dot product precisely because only the component of force along the motion feeds energy into it; power follows as P=F⃗⋅v⃗P = \vec F\cdot\vec v.

Derivation AWhy multiplying matching components gives the dot product

We want to prove that the geometric definition ABcos⁡θAB\cos\theta and the component recipe AxBx+AyBy+AzBzA_xB_x + A_yB_y + A_zB_z are the same number — the identity every analytical problem stands on. The strategy: write both vectors on the unit-vector basis, expand by distributivity, and let the tiny multiplication table of i^,j^,k^\hat i,\hat j,\hat k kill six of the nine terms. Watch orthogonality, not clever algebra, do all the work.

Step 1. Write the vectors on the basis: A⃗=Axi^+Ayj^+Azk^\vec A = A_x\hat i + A_y\hat j + A_z\hat k and B⃗=Bxi^+Byj^+Bzk^\vec B = B_x\hat i + B_y\hat j + B_z\hat k.

Step 2. Build the multiplication table for the basis using definition (1). Each unit vector dotted with itself: i^⋅i^=(1)(1)cos⁡0∘=1\hat i\cdot\hat i = (1)(1)\cos 0^\circ = 1, likewise j^⋅j^=k^⋅k^=1\hat j\cdot\hat j = \hat k\cdot\hat k = 1. Any two different unit vectors are perpendicular: i^⋅j^=(1)(1)cos⁡90∘=0\hat i\cdot\hat j = (1)(1)\cos 90^\circ = 0, and likewise for every mixed pair. Why this table matters: the coordinate axes were built perpendicular precisely so that this table would be mostly zeros — the basis does the trigonometry once, so every later problem can skip it.

Step 3. Expand A⃗⋅B⃗\vec A\cdot\vec B by distributivity. Nine terms appear: AxBx(i^⋅i^)+AxBy(i^⋅j^)+AxBz(i^⋅k^)+AyBx(j^⋅i^)+AyBy(j^⋅j^)+AyBz(j^⋅k^)+AzBx(k^⋅i^)+AzBy(k^⋅j^)+AzBz(k^⋅k^)A_xB_x(\hat i\cdot\hat i) + A_xB_y(\hat i\cdot\hat j) + A_xB_z(\hat i\cdot\hat k) + A_yB_x(\hat j\cdot\hat i) + A_yB_y(\hat j\cdot\hat j) + A_yB_z(\hat j\cdot\hat k) + A_zB_x(\hat k\cdot\hat i) + A_zB_y(\hat k\cdot\hat j) + A_zB_z(\hat k\cdot\hat k).

Step 4. Apply the table. The six mixed terms carry a factor 0 and vanish; the three matched terms carry a factor 1 and survive: A⃗⋅B⃗=AxBx+AyBy+AzBz\vec A\cdot\vec B = A_xB_x + A_yB_y + A_zB_z — equation (2). Verified: take A⃗=3i^+4j^\vec A = 3\hat i + 4\hat j and B⃗=2i^−j^\vec B = 2\hat i - \hat j. Recipe: (3)(2)+(4)(−1)=2(3)(2) + (4)(-1) = 2. Geometry: A=5A = 5, B=5=2.236B = \sqrt5 = 2.236, and the worked example below finds θ=79.7∘\theta = 79.7^\circ with 5×2.236×cos⁡79.7∘=11.18×0.1789=2.05 \times 2.236 \times \cos 79.7^\circ = 11.18 \times 0.1789 = 2.0 — both routes agree.

The vector product: rotation as a vector

Now the wilder sibling. The vector (cross) product A⃗×B⃗\vec A\times\vec B is a vector, defined by a magnitude and a direction:

A⃗×B⃗=ABsin⁡θ  n^(3)\vec A \times \vec B = AB\sin\theta \; \hat n \qquad (3)

where: ABsin⁡θAB\sin\theta — the magnitude, the parallelogram area; n^\hat n — a unit vector perpendicular to both inputs, on the side the right-hand rule picks. The sine measures the perpendicular part of one vector relative to the other — Bsin⁡θB\sin\theta is the part of B⃗\vec B not lying along A⃗\vec A — so the cross product peaks for perpendicular vectors and dies for parallel ones; in particular A⃗×A⃗=0⃗\vec A\times\vec A = \vec 0. Two properties deserve engraving. First, the anticommutation from the figure: B⃗×A⃗=−(A⃗×B⃗)\vec B\times\vec A = -(\vec A\times\vec B) — same magnitude, reversed direction, because the finger-curl reverses; every year this costs students a sign in torque problems. Second, the cross product distributes over addition, provided you never reorder the factors while expanding. On the coordinate basis the definition compresses into a cycle: ∣i^×j^∣=1⋅1⋅sin⁡90∘=1|\hat i\times\hat j| = 1\cdot1\cdot\sin 90^\circ = 1, and the right hand sends xx into yy with the thumb along zz:

i^×j^=k^,j^×k^=i^,k^×i^=j^(4)\hat i\times\hat j = \hat k, \qquad \hat j\times\hat k = \hat i, \qquad \hat k\times\hat i = \hat j \qquad (4)

where: the pattern is cyclic — x→y→z→xx \to y \to z \to x; neighbours in cycle order give the third with a plus, reversed order a minus (j^×i^=−k^\hat j\times\hat i = -\hat k), and any unit vector with itself gives zero. This one line generates the whole component formula, usually packaged as a determinant:

A⃗×B⃗=∣i^j^k^AxAyAzBxByBz∣=(AyBz−AzBy) i^−(AxBz−AzBx) j^+(AxBy−AyBx) k^(5)\vec A\times\vec B = \begin{vmatrix} \hat i & \hat j & \hat k \\ A_x & A_y & A_z \\ B_x & B_y & B_z \end{vmatrix} = (A_yB_z - A_zB_y)\,\hat i - (A_xB_z - A_zB_x)\,\hat j + (A_xB_y - A_yB_x)\,\hat k \qquad (5)

where: top row unit vectors, second row the first factor's components, third row the second's. Treat it as a recipe: cover each unit vector's row and column, cross-multiply the remaining 2×22\times2 block, alternate signs +,−,++,-,+. Row order is load-bearing — swapping the component rows flips every sign: anticommutation resurfacing in the arithmetic. Where does physics reach for this machine? Anywhere a turning lives. Torque is τ⃗=r⃗×F⃗\vec\tau = \vec r\times\vec F, magnitude rFsin⁡θrF\sin\theta — only the force component perpendicular to the arm turns the bolt, and the vector points along the turn's axis. Angular momentum is L⃗=r⃗×p⃗\vec L = \vec r\times\vec p; the magnetic force on a moving charge is qv⃗×B⃗q\vec v\times\vec B. Each gets its own module later — here you forge the tool they share.

quantityproductresult type
work W=F⃗⋅d⃗W = \vec F\cdot\vec ddotscalar (J)
power P=F⃗⋅v⃗P = \vec F\cdot\vec vdotscalar (W)
torque τ⃗=r⃗×F⃗\vec\tau = \vec r\times\vec Fcrossvector (N·m)
angular momentum L⃗=r⃗×p⃗\vec L = \vec r\times\vec pcrossvector (kg·m²/s)
magnetic force F⃗=qv⃗×B⃗\vec F = q\vec v\times\vec Bcrossvector (N)

The habit to build: energy-flavoured quantities are dots (scalars), rotation-flavoured quantities are crosses (vectors with an axis and a handedness).

Go deeper · Derivation B · 1 minThe determinant formula from the unit-vector cycle
Derivation BThe determinant formula from the unit-vector cycle

We want the component form of A⃗×B⃗\vec A\times\vec B, deriving one component honestly so the determinant stops being an incantation. The strategy: expand over the basis by distributivity — nine terms again — with the cycle (4) as the multiplication table. Watch the three self-cross terms die, and each surviving pair assemble into a difference: that is where the formula's minus signs come from.

Step 1. Expand (Axi^+Ayj^+Azk^)×(Bxi^+Byj^+Bzk^)(A_x\hat i + A_y\hat j + A_z\hat k)\times(B_x\hat i + B_y\hat j + B_z\hat k) by distributivity, keeping factor order scrupulously — cross products must never be reordered mid-expansion.

Step 2. Kill the diagonal. The terms AxBx(i^×i^)A_xB_x(\hat i\times\hat i), AyBy(j^×j^)A_yB_y(\hat j\times\hat j), AzBz(k^×k^)A_zB_z(\hat k\times\hat k) all vanish, because any vector crossed with itself is zero.

Step 3. Collect the k^\hat k component. Exactly two survivors produce k^\hat k: AxBy(i^×j^)=AxBy k^A_xB_y(\hat i\times\hat j) = A_xB_y\,\hat k and AyBx(j^×i^)=−AyBx k^A_yB_x(\hat j\times\hat i) = -A_yB_x\,\hat k. Together: (AxBy−AyBx) k^(A_xB_y - A_yB_x)\,\hat k. Why a difference: the two routes to k^\hat k traverse the cycle in opposite orders, and the reversed order pays the anticommutation sign. Every component of the cross product is a difference for exactly this reason.

Step 4. Repeat for i^\hat i (from j^×k^\hat j\times\hat k, k^×j^\hat k\times\hat j) and j^\hat j (from k^×i^\hat k\times\hat i, i^×k^\hat i\times\hat k): (AyBz−AzBy) i^(A_yB_z - A_zB_y)\,\hat i and (AzBx−AxBz) j^(A_zB_x - A_xB_z)\,\hat j — the latter is −(AxBz−AzBx)-(A_xB_z - A_zB_x), the determinant's middle minus sign.

Step 5. Assemble all three components and recognize the pattern as the row expansion of the determinant in equation (5). The determinant is nothing but the cycle (4), filed for fast retrieval. Verified: take A⃗=2i^+j^\vec A = 2\hat i + \hat j, B⃗=i^−j^+k^\vec B = \hat i - \hat j + \hat k. Formula: i^ (1⋅1−0⋅(−1))−j^ (2⋅1−0⋅1)+k^ (2⋅(−1)−1⋅1)=i^−2j^−3k^\hat i\,(1\cdot1 - 0\cdot(-1)) - \hat j\,(2\cdot1 - 0\cdot1) + \hat k\,(2\cdot(-1) - 1\cdot1) = \hat i - 2\hat j - 3\hat k. Perpendicularity check by dot products: (i^−2j^−3k^)⋅(2i^+j^)=2−2+0=0(\hat i - 2\hat j - 3\hat k)\cdot(2\hat i + \hat j) = 2 - 2 + 0 = 0 ✓ and ⋅(i^−j^+k^)=1+2−3=0\cdot(\hat i - \hat j + \hat k) = 1 + 2 - 3 = 0 ✓ — the output is genuinely perpendicular to both inputs, as definition (3) demands.

Limiting cases

CaseDot productCross productMeaning
θ=0∘\theta = 0^\circ (parallel)+AB+AB (maximum)0⃗\vec 0full cooperation, no twist
θ=45∘\theta = 45^\circAB/2AB/\sqrt2magnitude AB/2AB/\sqrt2the crossover — equal cooperation and twist
θ=90∘\theta = 90^\circ (perpendicular)00magnitude ABAB (maximum)no cooperation, maximum twist
θ=180∘\theta = 180^\circ (antiparallel)−AB-AB0⃗\vec 0full opposition, still no twist
B⃗=A⃗\vec B = \vec AA2A^20⃗\vec 0self-dot gives length²; self-cross gives nothing

The 45∘45^\circ row is a JEE favourite in disguise — the unique angle where ∣A⃗×B⃗∣=A⃗⋅B⃗|\vec A\times\vec B| = \vec A\cdot\vec B. The zero entries are option-killers: torque from a force along the arm, or work from a perpendicular force, is dead on arrival.

Worked examples

Board level — dot product, angle, and a work bonus

Given A⃗=3i^+4j^\vec A = 3\hat i + 4\hat j and B⃗=2i^−j^\vec B = 2\hat i - \hat j, find (a) A⃗⋅B⃗\vec A\cdot\vec B, (b) the angle between the vectors, (c) A⃗×B⃗\vec A\times\vec B.

Step 1. Dot product by recipe (2): A⃗⋅B⃗=(3)(2)+(4)(−1)=2\vec A\cdot\vec B = (3)(2) + (4)(-1) = \mathbf{2}. Small and positive — the vectors barely cooperate, so expect an angle just under 90∘90^\circ.

Step 2. Magnitudes for the angle: A=32+42=5A = \sqrt{3^2 + 4^2} = 5 (the 3–4–5 triangle) and B=22+(−1)2=5≈2.236B = \sqrt{2^2 + (-1)^2} = \sqrt5 \approx 2.236.

Step 3. Run identity (1) backwards: cos⁡θ=A⃗⋅B⃗AB=25×2.236=211.18=0.1789\cos\theta = \dfrac{\vec A\cdot\vec B}{AB} = \dfrac{2}{5 \times 2.236} = \dfrac{2}{11.18} = 0.1789, so θ=arccos⁡(0.1789)=79.7∘\theta = \arccos(0.1789) = \mathbf{79.7^\circ} — just under 90∘90^\circ, as the small dot product predicted.

Step 4. Cross product. Both vectors live in the xyxy-plane, so only the k^\hat k component of (5) survives: A⃗×B⃗=(AxBy−AyBx) k^=(−3−8) k^=−11 k^\vec A\times\vec B = (A_xB_y - A_yB_x)\,\hat k = (-3 - 8)\,\hat k = \mathbf{-11\,\hat k}. The 11 is the parallelogram's area; the minus says the sweep from A⃗\vec A to B⃗\vec B turns into the page.

Step 5. Cross-check: (ABcos⁡θ)2+(ABsin⁡θ)2=(AB)2(AB\cos\theta)^2 + (AB\sin\theta)^2 = (AB)^2. Indeed 22+112=125=(55)22^2 + 11^2 = 125 = (5\sqrt5)^2 ✓ — dot² + cross² = (AB)², a free consistency check strong students run silently.

Competitive level — the tan-theta classic and a perpendicular unit vector

Problem 1 (JEE-Main pattern): If ∣A⃗×B⃗∣=3 (A⃗⋅B⃗)|\vec A\times\vec B| = \sqrt3\,(\vec A\cdot\vec B), find the angle between the vectors, and the magnitude of A⃗+B⃗\vec A + \vec B in terms of AA and BB.

Step 1. Divide the relation by A⃗⋅B⃗\vec A\cdot\vec B: ABsin⁡θABcos⁡θ=tan⁡θ=3\dfrac{AB\sin\theta}{AB\cos\theta} = \tan\theta = \sqrt3, so θ=60∘\theta = \mathbf{60^\circ}. One line — the magnitudes cancel by construction.

Step 2. The follow-up uses the resultant formula with cos⁡60∘=12\cos 60^\circ = \tfrac12: ∣A⃗+B⃗∣=A2+B2+2ABcos⁡θ=A2+B2+AB|\vec A+\vec B| = \sqrt{A^2 + B^2 + 2AB\cos\theta} = \mathbf{\sqrt{A^2 + B^2 + AB}}. Timing note: 45 seconds total. The whole question is the reflex sin⁡/cos⁡=tan⁡\sin/\cos = \tan; if you find yourself squaring both sides, you have missed the point. Variant to expect: ∣A⃗×B⃗∣=A⃗⋅B⃗|\vec A\times\vec B| = \vec A\cdot\vec B gives tan⁡θ=1\tan\theta = 1, θ=45∘\theta = 45^\circ. *Problem 2 (JEE-Main/Adv pattern): Find a unit vector perpendicular to both A⃗=2i^+j^\vec A = 2\hat i + \hat j and B⃗=i^−j^+k^\vec B = \hat i - \hat j + \hat k.*

Step 1. Perpendicular-to-both is the cross product's home turf. By the determinant (5) — computed in Derivation B — A⃗×B⃗=i^−2j^−3k^\vec A\times\vec B = \hat i - 2\hat j - 3\hat k.

Step 2. Normalize: ∣A⃗×B⃗∣=12+(−2)2+(−3)2=14|\vec A\times\vec B| = \sqrt{1^2 + (-2)^2 + (-3)^2} = \sqrt{14}, so n^=±i^−2j^−3k^14\hat n = \mathbf{\pm\dfrac{\hat i - 2\hat j - 3\hat k}{\sqrt{14}}}.

Step 3. State the ±\pm (both perpendiculars are legitimate unless an orientation is fixed) and verify with dots: n^⋅A⃗∝2−2+0=0\hat n\cdot\vec A \propto 2-2+0 = 0 ✓, n^⋅B⃗∝1+2−3=0\hat n\cdot\vec B \propto 1+2-3 = 0 ✓. Timing note: 90 seconds including the determinant. The verification dots cost five seconds and catch nearly every arithmetic slip — never skip them when a cross product feeds a later part.

Exam traps

The sin–cos swap. The most common error of the topic: torque with cos⁡θ\cos\theta, work with sin⁡θ\sin\theta. Anchor it physically — alignment feeds energy (cos), perpendicularity turns things (sin) — and check an extreme: a force along the arm turns nothing.

The left hand under pressure. In the exam hall the pen lives in the right hand — so students curl the free left hand, silently reversing every direction. Put the pen down; the rule is worth two seconds.

Order in the determinant. τ⃗=r⃗×F⃗\vec\tau = \vec r\times\vec F means rr-components in the second row, FF in the third. Swapped rows give a correct-magnitude, wrong-direction answer — and MCQ setters plant that exact option.

Dropping the direction. A⃗×B⃗\vec A\times\vec B is a vector: reporting 1111 instead of −11k^-11\hat k (or "11, into the page") loses the direction mark. Magnitude-only answers are safe only when the question says "magnitude."

Perpendicular ⇒ dot is AB. Under pressure, "perpendicular" somehow becomes "dot product =AB= AB." Perpendicular is the dot's zero and the cross's maximum — catching yourself writing ABAB there means the wires crossed mid-question.

Cycle in reverse. k^×j^=−i^\hat k\times\hat j = -\hat i, not i^\hat i: against-cycle products carry the minus. In doubt, rebuild from i^×j^=k^\hat i\times\hat j = \hat k and anticommute rather than trusting adrenaline-soaked memory.

Origins · 1 minJosiah Willard Gibbs and how this was discovered
JW
Josiah Willard Gibbs
1839–1903 · Yale

Gibbs never sat the exams you are preparing for, but he wrote their notation. Hamilton's quaternions — the first system that could multiply directed quantities — packaged the dot and cross products inside one unwieldy four-part product, and physicists spent forty years untangling which piece they needed. Gibbs, teaching electromagnetism at Yale, performed the surgery: he separated the two products, gave each its own symbol and geometric story, and printed the result privately for his students in 1881 as Elements of Vector Analysis. The quaternionists were furious — Tait called it "a hermaphrodite monster" — but working physicists voted with their notebooks, and when Gibbs's student E. B. Wilson expanded the notes into the 1901 textbook Vector Analysis, the notation froze into the A⃗⋅B⃗\vec A\cdot\vec B and A⃗×B⃗\vec A\times\vec B you write today.

History
1843
— Hamilton carves the quaternion equations into Brougham Bridge, Dublin; hidden inside the quaternion product are both the dot and cross products.
1844
— Grassmann's Ausdehnungslehre independently builds a general algebra of directed quantities, decades ahead of its audience.
1881
— Gibbs privately prints Elements of Vector Analysis at Yale, splitting the quaternion product into today's separate dot and cross products.
1890
— The "quaternion war" opens: Tait attacks Gibbs's system in print; Gibbs and Heaviside prevail because the split products are simpler to use.
1901
— Wilson's textbook Vector Analysis, from Gibbs's lectures, fixes the modern notation used in every physics course since.

Going further

Texts. NCERT Physics XI Part 1, Ch. 3 · H.C. Verma, Concepts of Physics Vol. 1, Ch. 2 · Halliday, Resnick & Walker, Ch. 3 · Kleppner & Kolenkow §1.4–1.5 at the university register (deep dive: the aq-mech vector-algebra articles).

Provenance. All definitions and identities are standard. A verifier should recompute: the nine-term expansions in both derivations; A⃗⋅B⃗=2\vec A\cdot\vec B = 2, θ=79.7∘\theta = 79.7^\circ (cos⁡θ=2/11.18=0.1789\cos\theta = 2/11.18 = 0.1789), A⃗×B⃗=−11k^\vec A\times\vec B = -11\hat k, and the check 22+112=125=(55)22^2 + 11^2 = 125 = (5\sqrt5)^2 for A⃗=3i^+4j^\vec A = 3\hat i+4\hat j, B⃗=2i^−j^\vec B = 2\hat i-\hat j; tan⁡θ=3⇒60∘\tan\theta = \sqrt3 \Rightarrow 60^\circ and ∣A⃗+B⃗∣=A2+B2+AB|\vec A+\vec B| = \sqrt{A^2+B^2+AB}; (2i^+j^)×(i^−j^+k^)=i^−2j^−3k^(2\hat i+\hat j)\times(\hat i-\hat j+\hat k) = \hat i - 2\hat j - 3\hat k with magnitude 14\sqrt{14} and both perpendicularity dots zero.

Prerequisites. Components & Resolution of Vectors · Adding Vectors: Graphical & Analytical. Leads to. Work Done by a Force · Torque & the Turning Effect · Angular Momentum & Its Conservation (the cross product's home chapters).

Why does the cross product point OUT of the plane instead of somewhere inside it?

Because "inside the plane" is already taken — every in-plane direction is some mix of A⃗\vec A and B⃗\vec B, and none can represent the rotation fairly. What a rotation genuinely possesses is an axis: the one line that stays put while everything turns, and that axis IS the perpendicular to the plane. The remaining choice — up or down along the axis — encodes the sense of the turn, and the right-hand rule is the worldwide convention for it.

Is A·B·C legal? What about A×B×C?

A⃗⋅B⃗⋅C⃗\vec A\cdot\vec B\cdot\vec C is illegal as written: A⃗⋅B⃗\vec A\cdot\vec B is a scalar, and you cannot dot a scalar with C⃗\vec C — though (A⃗⋅B⃗) C⃗(\vec A\cdot\vec B)\,\vec C, a number times a vector, is fine. A⃗×B⃗×C⃗\vec A\times\vec B\times\vec C is legal but ambiguous: the cross product is not associative, so the two bracketings give different vectors and honest writing always shows brackets. Both constructions appear at JEE-Advanced as concept checks on whether you track what type each stage produces.

Why sine for the cross product when the dot product got cosine?

Relative to A⃗\vec A, the vector B⃗\vec B has a parallel part Bcos⁡θB\cos\theta and a perpendicular part Bsin⁡θB\sin\theta — together all of B⃗\vec B. The dot product hires the parallel part (cooperation happens along a shared line); the cross product hires the perpendicular part (turning requires pushing off the line). Nothing is wasted: dot² + cross² =A2B2= A^2B^2, an identity that doubles as an exam-hall consistency check.

For B×A, do I use my left hand since the order is reversed?

No — the rule is always right-handed; what changes is which vector the fingers start on. For B⃗×A⃗\vec B\times\vec A, point them along B⃗\vec B (first factor) and curl toward A⃗\vec A: the thumb now points opposite to the A⃗×B⃗\vec A\times\vec B case. Same hand, swapped start, reversed answer — that IS anticommutation. Reserve the left hand for nothing: a left-hand habit is the hardest error to catch because every diagram looks right.

Do I have to expand the full determinant every time, or can I use the cycle directly?

For 2D vectors (both in the xyxy-plane), never expand the full determinant — only the k^\hat k component survives: A⃗×B⃗=(AxBy−AyBx)k^\vec A\times\vec B = (A_xB_y - A_yB_x)\hat k, one line. For genuine 3D vectors the determinant is the fastest reliable route under pressure; the cycle is how you reconstruct it and how you handle pure unit-vector products (j^×k^=i^\hat j\times\hat k = \hat i instantly). Then verify with the dot-product zero-check.

Why is torque zero when the force points along the arm? The force is still strong.

Try opening a door by pushing the handle toward the hinges: full force, zero rotation. Rotation comes only from the force component perpendicular to the arm — exactly what rFsin⁡θrF\sin\theta measures, and at θ=0\theta = 0 there is none. Strength is not the issue; geometry is. It is also why handles live far from hinges (big rr) and why you push a spanner at right angles (sin⁡θ=1\sin\theta = 1): both moves maximize the same cross product.

Before this

Where this leads