The Four-Current and Charge Conservation

Derivation

The Four-Current and Charge Conservation

D-107 Home PU-105 Threads fields · symmetry Depends on Proper Time and the Four-Velocity

Statement

The charge density ρ and the three-current density J are the time and space parts of a single four-vector, the four-current J^μ = (ρc, J), built as J^μ = ρ₀U^μ from the proper charge density and four-velocity. Its vanishing four-divergence, ∂_μJ^μ = 0, is a Lorentz-scalar equation equivalent in every frame to the continuity equation ∂ρ/∂t + ∇·J = 0, which states that electric charge is locally conserved.

Why it matters

Charge conservation is one of the most precisely tested laws in physics, yet in the pre-relativistic form ∂ρ/∂t + ∇·J = 0 it mixes a scalar and a vector and hides its own covariance. Packaging (ρ, J) into J^μ reveals that the continuity equation is nothing but the statement that a four-vector has zero four-divergence — a single scalar equation that automatically holds in all inertial frames.

The four-current is also the source that couples to the electromagnetic potential and appears on the right of the covariant Maxwell equations. Understanding J^μ is therefore the entry point to writing electromagnetism, and later gauge theory, in manifestly Lorentz-invariant language.

Assumptions

Charge is a Lorentz-invariant scalar.If total charge depended on the observer, ρ₀ would not be a scalar and ρ₀U^μ would not transform as a four-vector; the whole construction collapses. Experimentally the electron and proton charges match to better than 1 part in 10²¹, and atoms are neutral to comparable precision, so invariance is on very firm footing.

A well-defined proper charge density ρ₀ exists locally.If the charge distribution has no local rest frame (e.g. two counter-streaming beams occupying the same point), ρ₀ and a single U^μ are not defined; one must instead sum J^μ over species. The single-fluid derivation then applies species by species.

Charge is transported by matter moving on timelike worldlines with a four-velocity U^μ.The prior result U^μ = γ(c, v), U^μU_μ = c² is used. If carriers were massless/luminal, γ diverges and ρ₀U^μ must be replaced by a null current built as a limit.

Spacetime is flat, so partial derivatives form a genuine covariant operator.The four-gradient ∂_μ transforms as a covector only in Minkowski coordinates. In curved spacetime ∂_μ must be promoted to the covariant derivative ∇_μ, and the conservation law becomes ∇_μJ^μ = 0.

Derivation

ρ₀ ≡ dq / dV₀, a Lorentz scalar

Define the proper charge density as charge per unit rest-frame volume. Charge dq is invariant (assumption 1) and dV₀ is the proper volume, so their ratio is a scalar attached to each fluid element. A

J^μ ≡ ρ₀ U^μ

A scalar times a four-vector is a four-vector. Since ρ₀ is a scalar and U^μ a four-vector (prior result), J^μ transforms as a four-vector by construction — this is the whole point. B

U^μ = γ(c, v) ⟹ J^μ = ρ₀γ(c, v) = (ρ₀γ c, ρ₀γ v)

Insert the explicit four-velocity from the assumed prior result proper-time-and-four-velocity, with γ = (1 − v²/c²)^−1/2. Pure substitution. A

ρ ≡ γ ρ₀, J ≡ ρ v ⟹ J^μ = (ρc, J)

Identify the lab-frame charge density. The single factor of γ is length contraction: the same charge dq occupies a volume contracted by 1/γ along the motion, so ρ = γρ₀. The spatial part ρ₀γv = ρv is the ordinary current density. B

∂_μ ≡ ∂/∂x^μ = ( (1/c)∂/∂t , ∇ )

Introduce the four-gradient. Differentiating with respect to the contravariant coordinates x^μ = (ct, x) gives a covariant (lower-index) operator, so the contraction ∂_μJ^μ is a Lorentz scalar. C

∂_μJ^μ = (1/c)∂(ρc)/∂t + ∇·J = ∂ρ/∂t + ∇·J

Contract the four-gradient with the four-current: the time component gives (1/c)∂_t(ρc) and the space components give ∂_xJ_x+∂_yJ_y+∂_zJ_z = ∇·J. The c's cancel in the first term. A

∂_μJ^μ = 0 ⟺ ∂ρ/∂t + ∇·J = 0

Impose local conservation. Integrating ∂ρ/∂t + ∇·J over a fixed volume and using the divergence theorem gives dQ/dt = −∮J·dA: charge only changes by flowing through the boundary, never appearing or vanishing on the spot. Because the left side is a Lorentz scalar, its vanishing in one inertial frame guarantees it in every frame. C

Result

J^μ = ρ₀U^μ = (ρc, J), ∂_μJ^μ = 0 ⟺ ∂ρ/∂t + ∇·J = 0

Reading. Charge density and current density are two faces of one geometric object, the four-current. Local charge conservation is the frame-independent statement that this four-vector is divergence-free in spacetime. What looks like a scalar-plus-vector conservation law in one frame is a single covariant equation: an observer who measures a pure charge density at rest sees, after a boost, that same charge as a current, and the conservation law is preserved exactly.

Units check. [ρc] = (C·m⁻³)(m·s⁻¹) = C·m⁻²·s⁻¹ = A·m⁻², the same as [J] — so all four components of J^μ carry the units A·m⁻², as a four-vector must. Then [∂_μJ^μ] = (m⁻¹)(A·m⁻²) = A·m⁻³ = C·s⁻¹·m⁻³, a rate of change of charge density, matching [∂ρ/∂t].

Limiting cases

Static / steady state: ∂ρ/∂t = 0 ⟹ ∇·J = 0 — steady currents are divergence-free, the foundational hypothesis of magnetostatics and Kirchhoff's current law.
Non-relativistic (v ≪ c): γ → 1, so ρ → ρ₀ and J → ρ₀v; the four-current reduces to the familiar Galilean (ρ, J) and the continuity equation is unchanged in form.
Pure rest charge: v = 0 ⟹ J^μ = (ρ₀c, 0) — a static charge is "all time component"; a boost is what generates the spatial current, just as a boost turns rest energy into momentum.
Point charge: ρ = q δ³(x − r(t)), J = q v δ³(x − r(t)) — continuity is then the statement that the charge follows an unbroken worldline.

Breaks when

Curved spacetime. The flat-space operator ∂_μ is no longer a tensor derivative; ∂_μJ^μ = 0 is not covariant and can even be coordinate-dependent. It must be replaced by ∇_μJ^μ = (1/√−g) ∂_μ(√−g J^μ) = 0, which restores conservation as a genuine geometric law.
Sources or sinks of charge. If a model allows charge creation or destruction (a hypothetical monopole-catalysed decay, or a phenomenological source term), then ∂_μJ^μ = σ ≠ 0. Real electromagnetism forbids this: the U(1) gauge structure makes ∂_μJ^μ = 0 exact.
Chiral / anomalous currents. In quantum field theory a classically conserved axial current picks up the ABJ anomaly, ∂_μJ^μ₅ = (e²/16π²) ε^μνρσF_μνF_ρσ ≠ 0. The electric (vector) current stays conserved, but the naive classical argument fails for chiral currents.
No local rest frame. Where multiple charged streams overlap (or in a plasma treated as several fluids), a single ρ₀U^μ is ill-defined; one must add the four-currents of each species, J^μ = Σ_a ρ_0,aU^μ_a.

Failure modes

Dropping the factor of c in the time component. Writing J^μ = (ρ, J) instead of (ρc, J) gives mismatched units between components and a wrong continuity equation missing the 1/c that cancels; the four "vector" is then not one.
Treating ρ as a Lorentz invariant. Only ρ₀ is invariant. Charge density is the time component of a four-vector and transforms; forgetting this leads to claiming a moving charge distribution has the same density in all frames (it does not — it is enhanced by γ).
Using ρ = γ²ρ₀ by "double counting" length contraction and time dilation. Only one power of γ enters, from volume contraction; the charge itself is invariant, unlike, say, energy density.
Confusing ∂^μ and ∂_μ in the divergence. The scalar is a contraction, so ∂_μJ^μ = ∂^μJ_μ — both give ∂ρ/∂t + ∇·J. Students sometimes flip a sign by mis-raising indices with the metric and get ∂ρ/∂t − ∇·J.
Believing conservation must be assumed separately in each frame. It need not: since ∂_μJ^μ is a scalar, one experimental verification of conservation in the lab frame secures it in all inertial frames automatically.

Discussion

The deepest content of this result is that charge conservation is not an add-on to relativity but is enforced by it. A conserved current is exactly the object whose four-divergence vanishes, and "four-divergence vanishes" is a Lorentz scalar. So the moment you accept that charge is invariant and that it is carried on worldlines, you are forced to conclude that the pre-relativistic continuity equation is covariant. Relativity does not merely tolerate charge conservation; it explains why the mixing of ρ and J across frames is consistent.

The link to symmetry is Noether's theorem. Electric charge is the conserved Noether charge of the global U(1) phase symmetry of a charged field, ψ → e^iαψ. The associated Noether current is precisely J^μ, and ∂_μJ^μ = 0 is the on-shell Noether identity. This is the physics2u "symmetry" thread in its cleanest instance: a continuous symmetry ⟶ a four-current ⟶ a conservation law.

In electromagnetism the four-current is the source of the field. The covariant Maxwell equation ∂_μF^μν = μ₀J^ν makes conservation automatic: because F^μν is antisymmetric, ∂_ν∂_μF^μν = 0 identically (a symmetric pair of derivatives contracted with an antisymmetric tensor), so ∂_νJ^ν = 0 follows from the field equations themselves. Charge conservation is thus not independent of Maxwell's equations — it is a consistency condition baked into them, and this is why one cannot consistently couple electromagnetism to a non-conserved current.

Gauging the symmetry sharpens this further. Local U(1) invariance requires the coupling −A_μJ^μ in the action, and gauge invariance of that term, A_μ → A_μ + ∂_μχ, changes the action by −∫ ∂_μχ · J^μ = +∫ χ · ∂_μJ^μ (after integration by parts). Gauge invariance therefore demands ∂_μJ^μ = 0: charge conservation is the precise price of the consistency of a massless spin-1 gauge field. This is also why the vector current survives quantization while chiral currents may not — an anomaly in ∂_μJ^μ would destroy gauge invariance and render the theory inconsistent, so it must cancel.

Common misconceptions. (i) "Charge density is invariant like charge" — no: charge is invariant, density is the time component of a four-vector and grows by γ for moving distributions. (ii) "The continuity equation is a separate empirical law that happens to be relativistic" — it is the covariant statement ∂_μJ^μ=0, forced by invariance of charge. (iii) "J^μ being conserved means Q is the same in all frames" — total charge Q = ∫ρ d³x is separately invariant, but for a subtler reason (the flux of J^μ through a spacelike slice), not simply because a component is conserved.

Worked examples

Example 1 — Four-current of a boosted charge cloud, and its invariant. A uniform charge cloud has proper charge density ρ₀ = 1.00×10⁻⁶ C·m⁻³ and moves at v = 0.60c along x. Find ρ, J, the components of J^μ, and verify the scalar invariant.

γ = (1 − v²/c²)^−1/2 = (1 − 0.36)^−1/2 = (0.64)^−1/2 = 1/0.80 = 1.25

Lorentz factor for β = 0.60. A

ρ = γρ₀ = 1.25 × 1.00×10⁻⁶ = 1.25×10⁻⁶ C·m⁻³

Lab-frame density, enhanced by one factor of γ (step 4 of derivation). A

J = ρv = 1.25×10⁻⁶ × (0.60 × 3.00×10⁸) = 1.25×10⁻⁶ × 1.80×10⁸ = 225 A·m⁻²

Spatial current density, along x. A

J⁰ = ρc = 1.25×10⁻⁶ × 3.00×10⁸ = 375 A·m⁻²; J^μ = (375, 225, 0, 0) A·m⁻²

Assemble the four components. A

J^μJ_μ = (J⁰)² − |J|² = 375² − 225² = 140625 − 50625 = 90000 (A·m⁻²)²

Contract with metric (+,−,−,−); the result must be frame-independent. C

√(J^μJ_μ) = 300 A·m⁻² = ρ₀c = 1.00×10⁻⁶ × 3.00×10⁸

The invariant length of J^μ is exactly ρ₀c — the value seen in the rest frame, confirming J^μ=ρ₀U^μ and U^μU_μ=c². B

J^μ = (375, 225, 0, 0) A·m⁻², √(J^μJ_μ) = ρ₀c = 300 A·m⁻²

Reading. The moving cloud carries both a raised density and a current; combined into J^μ they have an invariant magnitude equal to the pure rest-frame value ρ₀c. Units check. Every component and the invariant are in A·m⁻², as required for a four-current.

Example 2 — Continuity for a draining charged sphere. A ball of radius R = 0.10 m holds a spatially uniform charge density that is decreasing at ∂ρ/∂t = −2.0×10⁻³ C·m⁻³·s⁻¹ as charge flows radially outward. Use continuity to find the radial current density at the surface and the total current leaving the ball.

∂ρ/∂t + ∇·J = 0 ⟹ ∇·J = −∂ρ/∂t = +2.0×10⁻³ A·m⁻³

Apply the continuity equation; a falling density requires a positive (outward) divergence of current. A

∇·J = (1/r²) d(r²J_r)/dr = const ⟹ J_r(r) = ⅓ (∇·J) r

By spherical symmetry J = J_r(r) r̂; a uniform divergence integrates to a current growing linearly in r. B

J_r(R) = ⅓ (2.0×10⁻³)(0.10) = 6.7×10⁻⁵ A·m⁻²

Evaluate at the surface. A

I = ∮ J·dA = J_r(R) · 4πR² = 6.7×10⁻⁵ × 4π(0.10)²

Total outward current is the flux of J through the surface. A

I = 6.7×10⁻⁵ × 0.1257 = 8.4×10⁻⁶ A

Arithmetic (4π(0.10)² = 0.1257 m²). A

−dQ/dt = −(∂ρ/∂t)V = −(−2.0×10⁻³)·(4/3)π(0.10)³ = 2.0×10⁻³ × 4.19×10⁻³ = 8.4×10⁻⁶ A

Cross-check via the integral form dQ/dt = −I; the two routes agree, confirming conservation. C

J_r(R) = 6.7×10⁻⁵ A·m⁻², I = 8.4×10⁻⁶ A = −dQ/dt

Reading. The charge lost per second equals the current through the boundary — no charge disappears, it merely flows out, exactly as continuity demands. Units check. (A·m⁻³)·m = A·m⁻² for J_r; (A·m⁻²)·m² = A for I.

Problems

A proton beam has proper number density n₀ = 5.0×10¹⁴ m⁻³ and moves at v = 0.80c. Taking e = 1.60×10⁻¹⁹ C, write the four-current J^μ in the lab frame.
Solution
ρ₀ = n₀e = 5.0×10¹⁴ × 1.60×10⁻¹⁹ = 8.0×10⁻⁵ C·m⁻³. γ = (1−0.64)^−1/2 = (0.36)^−1/2 = 1/0.60 = 1.667. Then ρ = γρ₀ = 1.333×10⁻⁴ C·m⁻³. J⁰ = ρc = 1.333×10⁻⁴ × 3.0×10⁸ = 4.0×10⁴ A·m⁻². J_x = ρv = 1.333×10⁻⁴ × 2.4×10⁸ = 3.2×10⁴ A·m⁻². So J^μ = (4.0×10⁴, 3.2×10⁴, 0, 0) A·m⁻². Check: √((J⁰)²−J²) = √(16−10.24)×10⁴ = √5.76×10⁴ = 2.4×10⁴ = ρ₀c ✓.
Show explicitly that a travelling-wave charge pulse ρ(x,t) = f(x − ut) carrying current J = ρu x̂ satisfies continuity in one dimension for any profile f.
Solution
∂ρ/∂t = f′(x−ut)·(−u). ∂J_x/∂x = ∂(u f)/∂x = u f′(x−ut). Sum: ∂ρ/∂t + ∂J_x/∂x = −u f′ + u f′ = 0 ✓, for arbitrary differentiable f. Physically, rigid transport at speed u conserves charge because the profile merely translates. Numerically, with a Gaussian f = A e^{−(x−ut)²/2σ²}, both terms equal ±Au(x−ut)σ⁻²e^{−(x−ut)²/2σ²} and cancel identically.
Steady current flows in a wire so that ∂ρ/∂t = 0. A junction splits into three wires carrying current densities that produce net outflow ∮J·dA over a small Gaussian pillbox enclosing the junction. What does continuity require, and how is this Kirchhoff's current law?
Solution
Steady state gives ∇·J = −∂ρ/∂t = 0. Integrating over the pillbox and using the divergence theorem: ∮J·dA = ∫∇·J dV = 0. Writing the flux through each wire cross-section as a current I_k = ∫J·dA_k, the total signed flux out is Σ_k I_k = 0 — the sum of currents into a node equals the sum out. That is exactly Kirchhoff's current law, and it is nothing but charge conservation in the steady state. If, say, I₁ = 3.0 A and I₂ = 2.0 A flow in, the third wire must carry I₃ = 5.0 A out.
In its rest frame a small charged body has J^μ = (ρ₀c, 0, 0, 0) with ρ₀ = 2.0×10⁻⁴ C·m⁻³. Apply a Lorentz boost of rapidity giving β = 0.50 along x and find the transformed J′^μ. Confirm the invariant is unchanged.
Solution
γ = (1−0.25)^−1/2 = (0.75)^−1/2 = 1.1547. The four-current transforms like (ct, x): J′⁰ = γ(J⁰ − βJ^x) = γJ⁰, J′^x = γ(J^x − βJ⁰) = −γβJ⁰ (since J^x=0). With J⁰ = ρ₀c = 2.0×10⁻⁴×3.0×10⁸ = 6.0×10⁴ A·m⁻²: J′⁰ = 1.1547×6.0×10⁴ = 6.93×10⁴, J′^x = −1.1547×0.50×6.0×10⁴ = −3.46×10⁴ A·m⁻². Invariant: (J′⁰)²−(J′^x)² = (6.93²−3.46²)×10⁸ = (48.0−12.0)×10⁸ = 36.0×10⁸, so √ = 6.0×10⁴ = ρ₀c ✓, unchanged. Note the boost has generated a current (negative x, since the charge now streams backward in the new frame).
Starting from the covariant Maxwell equation ∂_μF^μν = μ₀J^ν, prove that charge conservation ∂_νJ^ν = 0 is automatic. State the property of F^μν that makes it work.
Solution
Apply ∂_ν to both sides: ∂_ν∂_μF^μν = μ₀∂_νJ^ν. The operator ∂_ν∂_μ is symmetric under μ ↔ ν (partial derivatives commute), while F^μν = −F^νμ is antisymmetric. The full contraction of a symmetric object with an antisymmetric object vanishes identically: ∂_ν∂_μF^μν = 0. Therefore μ₀∂_νJ^ν = 0 ⟹ ∂_νJ^ν = 0. The key property is the antisymmetry of the field-strength tensor. Consequence: Maxwell's equations are only consistent with a conserved source; one cannot couple electromagnetism to a current that fails ∂_νJ^ν=0.