Biot-Savart Law

Biot-Savart and Ampère's law are mathematically equivalent for steady currents — both compute the magnetic field around currents. Biot-Savart gives the differential contribution dB from each current element via direct integration, useful when the current geometry has no symmetry. Ampère's law (∮ B·dℓ = μ₀ I_enc) bypasses integration when symmetry is high enough that B is constant along a chosen Amperian loop. For a finite arc or a square loop, you reach for Biot-Savart; for an infinite wire or a long solenoid, Ampère is dramatically faster. The two are derivable from each other; both ultimately follow from ∇×B = μ₀J, the magnetostatic Maxwell equation.

The r² in the denominator is the same geometric factor that appears in Coulomb's law and Newton's law of gravity. Field lines from a point source spread over a sphere whose surface area grows as 4πr², so flux density per unit area drops as 1/r². Biot-Savart's source is a current element I dℓ rather than a point charge, but the geometric dilution is identical. The cross product I dℓ × r̂ then encodes the direction of the field — perpendicular to both the current and the line from the element to the field point, which is the right-hand rule. Without 1/r², electromagnetism could not coexist with relativity in flat 3D space.

Point your right thumb in the direction of conventional current flow. Your fingers curl in the direction of the magnetic field circulation around the wire. Equivalently, for the cross product I dℓ × r̂: align fingers along dℓ, curl them toward r̂, and your thumb points in the direction of dB. For a current loop, curl fingers along the current direction; thumb points along the axis where B emerges (the loop's north pole). This convention exists because B is a pseudovector — it flips under mirror reflection, while a real vector wouldn't. The choice of right (rather than left) is arbitrary but universally adopted since André-Marie Ampère.

For a finite straight wire along the x-axis from x=a to x=b with current I, the field at perpendicular distance d from the line is B = (μ₀I)/(4πd) × (sin θ₂ − sin θ₁), where θ₁ and θ₂ are the angles from the perpendicular dropped from the field point to the wire ends. Letting both angles approach ±90° (infinite wire) recovers B = μ₀I/(2πd). The integration uses dℓ × r̂ = sinφ dℓ and dℓ = (d/cos²φ) dφ, after which the integrand becomes simply (μ₀I)/(4πd) cosφ dφ — a textbook example of why parameterizing by the angle, not by length, makes Biot-Savart tractable.

Biot-Savart is exact only for steady currents in magnetostatics. For time-varying currents, retardation effects matter — the field at distance r reflects the source's state at retarded time t − r/c, not the present moment. The fully time-dependent generalization is the Liénard-Wiechert potentials and Jefimenko's equations, which include both retarded charge density and the time derivative of current density. For point charges in motion, replacing I dℓ with qv gives an approximate B = (μ₀/4π) qv × r̂/r², but this misses the radiation field that becomes essential when accelerations are large. The simple Biot-Savart form is a low-frequency, slow-motion limit.

MRI scanners need three things from their gradient coils: a strong main field B₀ (1.5T to 7T), spatially varying gradient fields (a few mT/m for spatial encoding), and shaped RF transmit/receive fields. All three require Biot-Savart integration over coil geometries — typically thousands of loop turns or saddle coils. Designers pick winding patterns that produce a target B(x,y,z) profile, then solve the inverse problem: given the desired field, what wire layout produces it? Modern shielded gradient coils use stream-function optimization driven by Biot-Savart kernels. For superconducting persistent-mode magnets, field homogeneity within parts per million across the imaging volume requires tightly controlled current distributions — every wire's contribution computed via Biot-Savart.