This page assumes nothing. We collect every symbol the parent note uses and build each one from a picture before it is ever allowed to appear in an equation. Read top to bottom; each idea rests only on the ones above it.
The crucial idea (look at the figure): the same arrow gives different numbers depending on which set of rulers you use. The arrow to the star doesn't move, but if you tilt your rulers, all three numbers change. This single fact is the reason the whole topic exists — we will need a machine that converts one set of numbers into the other.
Why we need this tool and not another: we want to ask "how much does body-axis b^i line up with world-axis n^j?" The dot product is exactly the instrument that answers "how much do two directions overlap," returning a clean number in [−1,1]. No other simple operation reads off alignment like this.
That overlap number is called a direction cosine — literally the cosine of the angle between one body axis and one world axis. There are 3×3=9 such pairs, and stacking them is what builds C.
Here Cij means "the number in row i, column j." A matrix is just a rectangle of numbers with a rule for multiplying an incoming list of 3 numbers to produce a new list of 3 numbers.
See Skew-symmetric matrices & cross-product operator for the matrix tools we build on next, and Rotation group SO(3) and Lie algebra so(3) for what family C belongs to.
Why this tool for spinning? When a point sits on a rigid spinning body, its velocity is r˙=ω×r. The cross product is the only simple operation that turns "spin axis + position" into the correct "sideways sweep velocity" — perpendicular to both, scaled by how far out you are. That's precisely how rotation moves things.
The parent's whole equation hinges on rewriting the cross product as a matrix. Why bother? Because C is a matrix, and to combine "spin" with "convert" in one equation, spin must also become a matrix.
Why skew captures spin exactly: a spinning frame keeps lengths fixed (no stretching). The instantaneous change it produces must be purely a turn, and skew-symmetric matrices are precisely the matrices whose action is "infinitesimal turn, no scaling." Full detail lives in Skew-symmetric matrices & cross-product operator and Rotation group SO(3) and Lie algebra so(3).
Why a derivative at all? The spacecraft tumbles, so the conversion table C is not frozen. Each entry Cij=b^i⋅n^j shifts as body axes swing. The topic's entire question — "how fast does C change?" — is a question about C˙. The matrix exponential then reverses this: it integrates C˙ back into C(t).
Everything on the left is a picture-level idea; they merge into C˙=−[ωB×]C on the right. See the [[3.5.04 DCM kinematics — Ċ = −[ω×]C (Hinglish)|parent topic]] once these are solid. Downstream this same machine powers Quaternion kinematics — q̇ = ½ Ω(ω) q, Euler angle kinematics & gimbal lock, Poisson's equation for rotating frames, and Attitude propagation & determination (TRIAD, QUEST).