Scalar and Vector Fields

Electromagnetism is a field theory: the influence of charges and currents is not transmitted instantaneously across empty space by mysterious action-at-a-distance, but is carried by fields that exist at every point. To make this precise we need two basic mathematical objects. A scalar field is a function $f(\mathbf{r}) = f(x,y,z)$ that returns one number at each point — temperature, pressure, mass density, or the electrostatic potential. A vector field $\mathbf{F}(\mathbf{r})$ returns a vector at each point — a quantity carrying both magnitude and direction, such as a fluid's velocity, a gravitational field, or the electric and magnetic fields themselves. The distinction is not merely cosmetic. What truly defines a scalar versus a vector is behaviour under rotations of the coordinate system. If you rotate your axes, a genuine scalar field reports the same number at a given physical point; the three components of a genuine vector field rotate into combinations of one another exactly as the displacement vector does. A triple of numbers that does not transform this way is not a vector. This transformation criterion separates real physical vectors from arbitrary lists of three numbers, and it is what guarantees that physical laws written with vectors hold in every frame. This article opens the mathematical bootcamp of the electrodynamics track. Everything that follows — the del operator, gradient, divergence and curl, the integral theorems, potentials, and ultimately Maxwell's four equations — is built on the firm distinction between a number at a point and an arrow at a point. Fix the vocabulary and the pictures now, and the rest of vector calculus becomes a set of natural questions about how these two kinds of field vary from place to place.

Key takeaways

A scalar field assigns a single real number (with units but no direction) to every point of space, e.g. temperature $T(\mathbf{r})$ or the electric potential $V(\mathbf{r})$ .
A vector field assigns a vector — magnitude and direction — to every point, e.g. the electric field $\mathbf{E}(\mathbf{r})$ or the magnetic field $\mathbf{B}(\mathbf{r})$ .
These objects are defined by how they transform under rotations of the coordinate axes: scalars are invariant, vector components mix among themselves.
Scalar fields are visualised with level surfaces (equipotentials); vector fields with arrows or continuous field lines whose density encodes magnitude.
The vector calculus of electromagnetism — gradient, divergence, curl, Laplacian — is precisely the machinery for relating scalar and vector fields.
All four Maxwell equations are statements about the divergence and curl of the vector fields $\mathbf{E}$ and $\mathbf{B}$ , often derived from scalar and vector potentials.
Key idea: A scalar field is one number per point and is invariant under rotations; a vector field is a magnitude-and-direction per point whose components transform like a displacement under rotations. Electromagnetism is the study of how such fields vary in space (via gradient, divergence, curl) and how those variations encode their charge and current sources.

Where this lives today

Field thinking is the default language of modern physics and engineering. Weather models evolve a scalar field (pressure), several more (temperature, humidity), and a vector field (wind velocity) on a global grid; electromagnetic solvers in every phone-antenna and MRI-magnet design team evolve $\mathbf{E}$ and $\mathbf{B}$ on meshes of millions of points. The scalar-versus-vector distinction is hard-coded into the software: a scalar field is stored as one number per grid cell, a vector field as three, and the transformation laws of this article are what guarantee the physics does not depend on how the mesh axes were chosen. The same discipline scales all the way up. The Standard Model of particle physics is a collection of interacting fields classified exactly by their transformation behaviour — scalars like the Higgs field, vectors like the photon field, and spinors in between. When LIGO measures gravitational waves or a lattice-QCD code computes a proton mass, the underlying bookkeeping is still "what kind of field is this, and how do its components transform?" — the question this article taught you to ask about temperature and wind.

The physical picture

Picture a room. At every point there is a temperature — one number. That cloud of numbers, filling the whole volume, is a scalar field. Now picture the air currents in the same room: at every point the air moves with some speed in some direction. That cloud of little arrows is a vector field. Same space, two different kinds of information attached.

Figure

Two kinds of field on the same space — one number per point versus one arrow per point

To see a scalar field, draw its level surfaces — the surfaces on which the field is constant (isotherms, equipotentials). To see a vector field, draw field lines that are everywhere tangent to the arrows, and let the density of lines represent the magnitude. Where lines crowd together the field is strong; where they spread out it is weak. These two visual idioms are the working pictures for the entire electromagnetism track: you will read equipotential maps and field-line diagrams far more often than you will read component formulas. The key test that distinguishes a true vector from three random numbers is rotation. Turn your head (rotate the axes) and a temperature reading does not change, but the components of the wind velocity get reshuffled in a precise, lockstep way — the same way the components of an arrow pointing in a fixed physical direction get reshuffled. Physical vectors must obey this rule.

Figure

The rotation test — the arrow is the physics, the components are bookkeeping

Electromagnetism lives entirely in this language, and each later operator earns an immediate physical payoff. The potential $V$ is a scalar field whose gradient gives the electric field, $\mathbf{E}=-\nabla V$ . The divergence of $\mathbf{E}$ will become Gauss's law, tying field lines to the charge that emits them. The curl of $\mathbf{E}$ will become Faraday's law of induction, and the curl of $\mathbf{B}$ will wrap around the currents that create it. The whole subject is about the operators that translate between scalar and vector fields and connect them to charges and currents.

The two field types, precisely

A scalar field is nothing more exotic than a real-valued function of position. If you stand at the point $(x,y,z)$ and ask "what is the temperature here?", the field answers with one number. Formally,

f:\ \mathbb{R}^3 \to \mathbb{R}, \qquad f = f(x,y,z) \qquad (1)

where: $f$ — scalar field — assigns one real number per point; $(x,y,z)$ — Cartesian coordinates of the point. The number carries units (kelvin, volts, kg/m³) but no direction; asking "which way does the temperature point?" is meaningless. A vector field packs three numbers into each point, but the packaging matters: the three numbers are the components of a single geometric arrow. Fluid velocity is the cleanest mental model — at each point the water moves at some speed in some direction. In Cartesian form,

\mathbf{F}:\ \mathbb{R}^3 \to \mathbb{R}^3, \qquad \mathbf{F} = F_x\,\hat{\mathbf{x}} + F_y\,\hat{\mathbf{y}} + F_z\,\hat{\mathbf{z}} \qquad (2)

where: $\mathbf{F}$ — vector field — assigns a vector per point; $F_x,F_y,F_z$ — Cartesian components, each itself a scalar function of position; $\hat{\mathbf{x}},\hat{\mathbf{y}},\hat{\mathbf{z}}$ — orthonormal Cartesian basis vectors. Note the layering: a single vector field contains three component functions, but they are not three independent scalar fields — they must transform together. From any vector field we can extract two derived quantities: its length and its direction. Both are fields in their own right, and it pays to know which kind:

|\mathbf{F}(\mathbf{r})| = \sqrt{F_x^2 + F_y^2 + F_z^2}, \qquad \hat{\mathbf{F}} = \frac{\mathbf{F}}{|\mathbf{F}|} \qquad (3)

where: $|\mathbf{F}|$ — magnitude of the field vector at each point — a genuine scalar field, as we prove below; $\hat{\mathbf{F}}$ — unit vector field giving the local direction. The field-line picture splits a vector field exactly this way: tangency encodes $\hat{\mathbf{F}}$ , line density encodes $|\mathbf{F}|$ .

What makes a vector a vector

Here is the question that separates physics from bookkeeping: if I hand you three functions $(f_1, f_2, f_3)$ , is that a vector field? The answer cannot come from the functions alone — it comes from how they respond when you rotate your axes. A scalar field responds in the simplest possible way: not at all. The same physical point $P$ carries labels $(x,y,z)$ in one frame and $(x',y',z')$ in a rotated frame, and the field cares only about $P$ :

f'(x',y',z') = f(x,y,z) \qquad (4)

where: $f'$ — the field function expressed in the rotated frame; both sides refer to the same physical point. This is exactly what makes density $\rho$ and potential $V$ scalars — observers with differently tilted axes agree on their values, point by point. A vector field responds the way a displacement arrow does. Rotate the axes by an orthogonal matrix $R$ and the components remix:

F_i' = \sum_{j} R_{ij}\, F_j, \qquad R^{\mathsf{T}} R = \mathbb{1} \qquad (5)

where: $R_{ij}$ — rotation matrix relating the two frames; $F_j$ — components in the original frame; $F_i'$ — components in the rotated frame; the orthogonality condition $R^{\mathsf{T}}R=\mathbb{1}$ says rotations preserve lengths and angles. Equation (5) is the definition of a vector in physics: a triple that transforms this way. Three unrelated readings — say (temperature, pressure, humidity) — stacked into a column fail the test, and any "law" written with them would change its content when the axes turned.

Derivation AThe magnitude of a vector field is a scalar field

We are proving that $|\mathbf{F}|$ , built by squaring and adding the three components at each point, passes the scalar test of equation (4): its value at a physical point is unchanged by any rotation of the axes. The strategy is direct — transform the components with equation (5), square, sum, and let the orthogonality condition $R^{\mathsf{T}}R=\mathbb{1}$ collapse the double sum. The one identity to watch is $\sum_i R_{ij}R_{ik}=\delta_{jk}$ , which is nothing but the matrix statement $R^{\mathsf{T}}R=\mathbb{1}$ written in components. This small calculation is the prototype for every invariance proof in the subject.

Step 1. Square and sum the rotated components:

\sum_i (F_i')^2 = \sum_i\Big(\sum_j R_{ij}F_j\Big)\Big(\sum_k R_{ik}F_k\Big) = \sum_{j,k}\Big(\sum_i R_{ij}R_{ik}\Big)F_jF_k

Why: each factor of $F_i'$ carries its own dummy index; regrouping isolates the purely geometric sum over $i$ .

Step 2. Apply orthogonality:

\sum_i R_{ij}R_{ik} = (R^{\mathsf{T}}R)_{jk} = \delta_{jk}

Why: this is the component form of $R^{\mathsf{T}}R=\mathbb{1}$ , with $\delta_{jk}$ the Kronecker delta — 1 when $j=k$ , 0 otherwise.

Step 3. Collapse the double sum with the delta:

\sum_i (F_i')^2 = \sum_{j,k}\delta_{jk}F_jF_k = \sum_j F_j^2

The squared length computed in the rotated frame equals the squared length in the original frame, so $|\mathbf{F}|$ assigns a frame-independent number to each point — a genuine scalar field. Verified: this is the standard invariance argument of Griffiths §1.1.5; setting $F=(4,3,0)$ and rotating by any angle reproduces magnitude 5.

The operators that connect them

With the two field types defined, all of vector calculus is the study of three differential machines, each answering a natural question. Each also carries an electromagnetic payoff that we preview here and develop in its own article. The gradient asks of a scalar field: in which direction does $f$ increase fastest, and how fast?

\nabla f = \frac{\partial f}{\partial x}\hat{\mathbf{x}} + \frac{\partial f}{\partial y}\hat{\mathbf{y}} + \frac{\partial f}{\partial z}\hat{\mathbf{z}} \qquad (6)

where: $\nabla f$ — the gradient, a vector field pointing along the steepest increase of $f$ , with magnitude equal to that maximal rate; $\partial f/\partial x$ — partial derivative, the rate of change along $\hat{\mathbf{x}}$ with the other coordinates held fixed. The gradient turns scalar into vector: applied to the potential it will hand us the electric field. The divergence asks of a vector field: does this point act as a source or a sink?

\nabla\cdot\mathbf{F} = \frac{\partial F_x}{\partial x} + \frac{\partial F_y}{\partial y} + \frac{\partial F_z}{\partial z} \qquad (7)

where: $\nabla\cdot\mathbf{F}$ — divergence, a scalar field measuring net outward spreading per unit volume. Its electromagnetic payoff is Gauss's law, $\nabla\cdot\mathbf{E}=\rho/\varepsilon_0$ : charge density is precisely the source strength of the electric field. The curl asks of a vector field: does it circulate around this point, and about which axis?

\nabla\times\mathbf{F} = \left(\frac{\partial F_z}{\partial y}-\frac{\partial F_y}{\partial z}\right)\hat{\mathbf{x}} + \left(\frac{\partial F_x}{\partial z}-\frac{\partial F_z}{\partial x}\right)\hat{\mathbf{y}} + \left(\frac{\partial F_y}{\partial x}-\frac{\partial F_x}{\partial y}\right)\hat{\mathbf{z}} \qquad (8)

where: $\nabla\times\mathbf{F}$ — curl, a vector field whose direction is the local rotation axis and whose magnitude is the circulation per unit area. Its payoffs are Faraday's law, $\nabla\times\mathbf{E}=-\partial_t\mathbf{B}$ , and Ampère's law, where the curl of $\mathbf{B}$ wraps around currents.

operator	input → output	question it answers	E&M payoff
gradient $\nabla f$	scalar → vector	which way is uphill, how steep?	$\mathbf{E}=-\nabla V$
divergence $\nabla\cdot\mathbf{F}$	vector → scalar	source or sink here?	$\nabla\cdot\mathbf{E}=\rho/\varepsilon_0$ , $\nabla\cdot\mathbf{B}=0$
curl $\nabla\times\mathbf{F}$	vector → vector	circulation about which axis?	$\nabla\times\mathbf{E}=-\partial_t\mathbf{B}$

The single most used scalar-to-vector conversion in electrostatics deserves to be written now, even though its full justification comes later in the track:

\mathbf{E}(\mathbf{r}) = -\nabla V(\mathbf{r}) \qquad (9)

where: $\mathbf{E}$ — electric field (vector field); $V$ — electric potential (scalar field); the minus sign encodes that $\mathbf{E}$ points downhill in potential, from high $V$ toward low $V$ , just as heat flows from hot to cold. One scalar function generates all three field components — the economy that makes potential theory worth an entire chapter.

Figure

One scalar field generating a vector field — equipotentials of a point charge and the E arrows perpendicular to them

Go deeper · Derivation B · 1 minThe point-charge field from its scalar potential

Derivation BThe point-charge field from its scalar potential

We are deriving the Coulomb field $\mathbf{E}=q\,\hat{\mathbf{r}}/(4\pi\varepsilon_0 r^2)$ from the single scalar function $V=q/(4\pi\varepsilon_0 r)$ , as a concrete first exercise in equation (9). The route is pure differentiation: compute the three partial derivatives of $1/r$ , assemble them into the gradient, and attach the minus sign. Watch the chain rule at the first step — everything hinges on $\partial r/\partial x = x/r$ , and a dropped factor here is the classic source of wrong powers of $r$ . The result exhibits the central economy of potential theory: one scalar function stores an entire three-component vector field.

Step 1. Write the potential and the radius explicitly:

V(\mathbf{r}) = \frac{1}{4\pi\varepsilon_0}\frac{q}{r}, \qquad r = \sqrt{x^2+y^2+z^2}

Why: to differentiate with respect to $x,y,z$ we need $r$ expressed in Cartesian coordinates; $\varepsilon_0$ is the permittivity of free space.

Step 2. Differentiate $1/r$ with respect to one coordinate using the chain rule:

\frac{\partial}{\partial x}\frac{1}{r} = -\frac{1}{r^2}\frac{\partial r}{\partial x} = -\frac{1}{r^2}\cdot\frac{x}{r} = -\frac{x}{r^3}

Why: $\partial r/\partial x = x/r$ follows from differentiating $r^2=x^2+y^2+z^2$ implicitly; the $y$ and $z$ derivatives are identical with $x$ replaced.

Step 3. Assemble the three partials into the gradient:

\nabla\frac{1}{r} = -\frac{x\hat{\mathbf{x}}+y\hat{\mathbf{y}}+z\hat{\mathbf{z}}}{r^3} = -\frac{\mathbf{r}}{r^3} = -\frac{\hat{\mathbf{r}}}{r^2}

Why: the numerator is exactly the position vector $\mathbf{r}$ , and $\hat{\mathbf{r}}=\mathbf{r}/r$ converts $\mathbf{r}/r^3$ into $\hat{\mathbf{r}}/r^2$ .

Step 4. Apply $\mathbf{E}=-\nabla V$ :

\mathbf{E} = -\frac{q}{4\pi\varepsilon_0}\nabla\frac{1}{r} = \frac{1}{4\pi\varepsilon_0}\frac{q}{r^2}\,\hat{\mathbf{r}}

A radial, inverse-square vector field generated entirely from a scalar field — the prototypical scalar/vector pairing of electrostatics, and the field whose divergence and flux we will dissect in the divergence and Gauss's-law articles. Verified: matches Coulomb's law as stated in Griffiths Ch. 2 and Purcell Ch. 2; the gradient identity $\nabla(1/r)=-\hat{\mathbf{r}}/r^2$ is Griffiths problem 1.13.

Go deeper · Derivation C · 1 minThe dot product of two vector fields is a scalar field

Derivation CThe dot product of two vector fields is a scalar field

We are proving that $\mathbf{A}\cdot\mathbf{B}$ — the pointwise dot product of any two genuine vector fields — passes the scalar test, so that quantities like $\mathbf{E}\cdot\mathbf{E}=E^2$ (which feeds the electromagnetic energy density) are legitimate scalar fields. The strategy mirrors the magnitude proof: transform both factors with the vector law (5), regroup the rotation matrices, and let orthogonality produce a Kronecker delta. This consistency check matters — it shows the scalar and vector transformation laws are not independent postulates but a mutually closed algebra: products of vectors built with the dot product automatically land back in the scalar class.

Step 1. Transform both vectors' components with the same rotation:

A_i' = \sum_j R_{ij}A_j, \qquad B_i' = \sum_k R_{ik}B_k

Why: both are vector fields in the same space, so the same matrix $R$ acts on each — there is only one rotation of the axes.

Step 2. Form the dot product in the rotated frame and regroup:

\mathbf{A}'\cdot\mathbf{B}' = \sum_i A_i' B_i' = \sum_{j,k} \Big(\sum_i R_{ij}R_{ik}\Big) A_j B_k

Why: the sum over $i$ involves only the rotation matrices, so it can be evaluated once and for all, independent of the fields.

Step 3. Use orthogonality, $\sum_i R_{ij}R_{ik} = \delta_{jk}$ , and collapse:

\mathbf{A}'\cdot\mathbf{B}' = \sum_{j,k}\delta_{jk}A_j B_k = \sum_j A_j B_j = \mathbf{A}\cdot\mathbf{B}

The value is frame-independent: the dot product of two true vector fields is a genuine scalar field. This is why the energy density $u=\tfrac{1}{2}(\varepsilon_0 E^2 + B^2/\mu_0)$ , built entirely from dot products, is a scalar — every observer agrees on how much energy sits in each cubic metre. Verified: standard result, Griffiths §1.1.5 and Jackson's introductory discussion of transformation properties; reduces to the magnitude proof when $\mathbf{B}=\mathbf{A}$ .

Limiting cases

Fields are classified by their behaviour under coordinate transformations and by their physical character:

Case	Result	Meaning
Scalar (rotation-invariant)	$f'(\mathbf{r}) = f(\mathbf{r})$	Temperature, charge density $\rho$ , electric potential $V$ . The value at a physical point is unchanged when axes are rotated.
Vector (rotates with axes)	$F_i' = \sum_j R_{ij}\, F_j$	$\mathbf{E}$ , $\mathbf{B}$ , force, velocity. The three components transform with the rotation matrix $R$ .
Pseudoscalar / pseudovector	$\mathbf{B}\to +\mathbf{B}\ \text{under parity}$	Quantities like $\mathbf{B}$ pick up an extra sign under reflection (parity) because they arise from cross products; an honest 'polar' vector like $\mathbf{E}$ flips sign under parity.
Uniform field (special case)	$\mathbf{F}(\mathbf{r}) = \mathbf{F}_0 = \text{const}$	A field that does not vary in space; its gradient, divergence and curl vanish but the field itself need not.
Static vs dynamic	$f(\mathbf{r})$ vs $f(\mathbf{r},t)$	Electrostatics uses time-independent fields; the full Maxwell theory promotes every field to a function of space and time.

Worked examples

Example 1 — reading a scalar field and its induced vector field (confidence builder). The electrostatic potential of a $q = 2.0\ \mathrm{nC}$ point charge is $V(r) = q/(4\pi\varepsilon_0 r)$ , with $1/(4\pi\varepsilon_0) = 8.99\times 10^9\ \mathrm{N\,m^2/C^2}$ .

Step 1. Evaluate the scalar field at $r_1 = 0.50\ \mathrm{m}$ :

V(0.50\ \mathrm{m}) = \frac{(8.99\times 10^9)(2.0\times 10^{-9})}{0.50} = 36.0\ \mathrm{V}

One number, no direction — a scalar field value.

Step 2. Evaluate the vector field it generates at the same point. From the derivation above, $E = q/(4\pi\varepsilon_0 r^2)$ :

E(0.50\ \mathrm{m}) = \frac{(8.99\times 10^9)(2.0\times 10^{-9})}{(0.50)^2} = 71.9\ \mathrm{V/m}

pointing radially outward — magnitude and direction, a vector field value.

Step 3. Cross-check the two with a finite difference, since $E_r=-\mathrm{d}V/\mathrm{d}r$ . At $r_2=0.51\ \mathrm{m}$ , $V = 35.3\ \mathrm{V}$ , so

-\frac{\Delta V}{\Delta r} = -\frac{35.3 - 36.0}{0.01} \approx 70\ \mathrm{V/m}

within 3% of the exact $71.9\ \mathrm{V/m}$ — the gradient really is the downhill rate of the scalar field.

Example 2 — the rotation test with numbers (exam level). In frame $S$ a vector field at some point has components $(F_x, F_y) = (4.0, 3.0)$ (units arbitrary), so $|\mathbf{F}| = \sqrt{16+9} = 5.0$ . Rotate the axes by $\theta = 30^\circ$ about $z$ ; the 2D rotation law gives $F_x' = F_x\cos\theta + F_y\sin\theta$ and $F_y' = -F_x\sin\theta + F_y\cos\theta$ .

Step 1. Compute the new components with $\cos 30^\circ = 0.866$ , $\sin 30^\circ = 0.500$ :

F_x' = 4.0(0.866) + 3.0(0.500) = 4.96, \qquad F_y' = -4.0(0.500) + 3.0(0.866) = 0.598

Step 2. Check the invariant:

|\mathbf{F}'| = \sqrt{4.96^2 + 0.598^2} = \sqrt{24.6 + 0.36} = \sqrt{25.0} = 5.0

The components changed drastically — $(4.0, 3.0) \to (4.96, 0.60)$ — but the magnitude is exactly 5.0 in both frames. A stack of three unrelated readings would not have passed this test; only the lockstep remixing of equation (5) preserves the geometry.

Where the two field types run electromagnetism

Electrostatic potential and field. The electrostatic potential $V(\mathbf{r})$ is the workhorse scalar field of electromagnetism; the electric field is the vector field $\mathbf{E} = -\nabla V$ . Working with the scalar $V$ is far easier (one function instead of three), and the gradient operation recovers the full vector field. Equipotential surfaces (level surfaces of $V$ ) are always perpendicular to field lines of $\mathbf{E}$ .

Charge and current sources. Charge density $\rho(\mathbf{r})$ is a scalar field, while current density $\mathbf{J}(\mathbf{r})$ is a vector field. Maxwell's equations relate them to $\mathbf{E}$ and $\mathbf{B}$ through divergence ( $\nabla\cdot\mathbf{E}=\rho/\varepsilon_0$ ) and curl ( $\nabla\times\mathbf{B}=\mu_0\mathbf{J}+\dots$ ). Recognising which quantity is scalar and which is vector dictates which operator can legally connect them.

Visualising fields. Engineers and physicists map scalar fields as colour plots or contour (equipotential) lines and vector fields as arrow plots or streamlines. A magnetic field's lines, an electric dipole's pattern, and a flowing fluid's streamlines are all the same visual idiom applied to a vector field — magnitude shown by line density, direction by tangency.

Energy density and Poynting flux. The electromagnetic energy density $u = \tfrac{1}{2}(\varepsilon_0 E^2 + B^2/\mu_0)$ is a scalar field built from the magnitudes of the vector fields, while the Poynting vector $\mathbf{S} = \tfrac{1}{\mu_0}\mathbf{E}\times\mathbf{B}$ is a vector field describing energy flow. Together they show how scalar and vector fields combine to describe energy transport.

Origins · 1 minMichael Faraday and how this was discovered

Michael Faraday

1791–1867 · Royal Institution, London

Faraday is the reason this article exists. A bookbinder's apprentice with no formal mathematics, he taught himself science from the books he bound, talked his way into Humphry Davy's laboratory, and became the greatest experimentalist of his century. Where his contemporaries computed forces acting at a distance between charges, Faraday insisted on what he could see in iron filings: lines of force filling the space around magnets and charges, bending, crowding, and pulling like elastic bands. Space itself, he argued, was in a state of strain — the field was the physical object. He could not write the equations, but Maxwell later said that Faraday's method was mathematical to its core, and built the entire theory of electromagnetism by translating Faraday's field pictures into the vector calculus you are about to learn. Every arrow diagram and every equipotential map in this track is Faraday's way of seeing, formalised.

History

1785

— Coulomb establishes the inverse-square force between charges, framed as action-at-a-distance with no field concept

1831

— Faraday discovers electromagnetic induction and begins describing electric and magnetic effects with lines of force filling space

1852

— Faraday publishes his mature view that the lines of force are physically real states of the intervening medium

1864

— Maxwell's dynamical theory recasts Faraday's pictures as differential equations for continuous scalar and vector fields

1881

— Gibbs and Heaviside independently distil quaternion methods into modern vector analysis: fields, gradient, divergence and curl in the notation used today

1905

— Einstein's relativity makes the field the irreducible physical object: the E and B fields of one observer mix into each other under boosts

Going further

References. Griffiths, Introduction to Electrodynamics, 4th ed., Ch. 1 · Jackson, Classical Electrodynamics, 3rd ed., Introduction and Ch. 1 · Purcell & Morin, Electricity and Magnetism, 3rd ed., Ch. 2 · Zangwill, Modern Electrodynamics, Ch. 1

Provenance. Definitions and transformation laws follow Griffiths Ch. 1 (§1.1.5 on how vectors transform) and Jackson's introductory transformation discussion. A verifier should: (1) confirm the gradient of $1/r$ gives $-\hat{\mathbf{r}}/r^2$ and hence the Coulomb field $\mathbf{E}=q\hat{\mathbf{r}}/(4\pi\varepsilon_0 r^2)$ (Griffiths Ch. 2, Purcell Ch. 2); (2) verify the orthogonality identity $\sum_i R_{ij}R_{ik}=\delta_{jk}$ and the resulting invariance of the dot product; (3) check the divergence and curl component formulas against any standard text. No numerical constants beyond $\varepsilon_0$ and $\mu_0$ (standard SI) are used.

Builds on: Vectors and vector algebra · Functions of several variables · Cartesian coordinates Leads to: The Gradient Operator · The Del Operator (Nabla) · The Divergence of a Vector Field · The Curl of a Vector Field · Spherical Coordinates

Is temperature really a scalar field even though it changes from place to place?

Yes — "scalar" does not mean "constant". A scalar field can vary wildly across space; what makes it a scalar is that at any one physical point, every observer agrees on its value regardless of how their coordinate axes are tilted. Contrast that with the wind velocity at the same point: observers with rotated axes assign it different components (though the same arrow). Varying-in-space and transforming-under-rotation are two completely independent questions, and the scalar/vector classification only concerns the second.

Why do physicists obsess over the transformation law instead of just saying a vector is an arrow?

"An arrow" is the right intuition, but the transformation law is how you check that a candidate quantity actually is one. Plenty of triples look like vectors and are not — the three Euler angles of a rigid body, for instance, do not remix like displacement components under rotation. The law also generalises where pictures fail: it is how one discovers that $\mathbf{B}$ is a pseudovector, that energy density is a scalar, and (in relativity) that $\mathbf{E}$ and $\mathbf{B}$ are really six components of one object. The test scales; the picture alone does not.

Are the three components of a vector field three separate scalar fields?

Each component is a perfectly good function of position, but it is not a scalar field, because its value at a fixed physical point changes when you rotate the axes ( $F_x$ becomes a mixture of the old $F_x, F_y, F_z$ ). The honest statement: a vector field is one geometric object whose description requires three functions, and those functions are tied together by the transformation law. Genuine scalar fields built from a vector field include its magnitude $|\mathbf{F}|$ and its dot product with any other vector field.

If the potential V is easier to work with, why do we bother with the vector field E at all?

Because force — the thing that actually pushes charges — is a vector. $V$ is a brilliant computational device: one function, no components, easy to add over many charges. But the physics deliverable is $\mathbf{F}=q\mathbf{E}$ , so at the end of every potential calculation you take a gradient to recover the field. There are also situations the scalar potential alone cannot handle: once fields change in time, $\nabla\times\mathbf{E}\neq 0$ and you need the vector potential too. The track teaches both representations precisely so you can pick the cheaper one per problem.

What does it mean physically that equipotentials and field lines are perpendicular?

Along an equipotential, $V$ does not change, so no work is done moving a test charge along it: $\mathbf{E}\cdot\mathrm{d}\boldsymbol{\ell}=0$ for any step $\mathrm{d}\boldsymbol{\ell}$ within the surface. That is only possible if $\mathbf{E}$ has no component along the surface — i.e. it is perpendicular to it. Equivalently, the gradient always points along the steepest change, which is at right angles to the directions of no change. This is the same geometry as contour lines on a hiking map: the steepest path crosses the contours perpendicularly.

Is the magnitude of a vector field always a smooth scalar field?

It is a scalar field wherever it is defined, but smoothness can fail where the field vanishes. $|\mathbf{F}|=\sqrt{F_x^2+F_y^2+F_z^2}$ involves a square root, and at a zero of the field (a neutral point between two like charges, say) the magnitude has a kink even though the components are perfectly smooth there. The direction field $\hat{\mathbf{F}}$ is worse — it is undefined at zeros. This is why careful field-line drawings show special "X" and "O" patterns near nulls.

Why does B not flip under reflection when E does?

Because $\mathbf{B}$ is built from cross products. Reflect a current loop in a mirror and the current's velocity vector (a polar vector) reflects normally, but the cross product $\mathbf{r}\times\mathbf{v}$ that sets the loop's magnetic moment picks up an extra sign — two reflected polar vectors leave their cross product with a reversed handedness. Quantities behaving this way are called axial vectors or pseudovectors. Nothing is wrong with $\mathbf{B}$ ; you just must not equate a polar vector to an axial one, or your equation fails in the mirrored world.

Do I need the pseudovector business at this stage, or can I file it away?

File it away, but keep the headline: there are two sub-species of "vector", distinguished by their behaviour under reflection, and $\mathbf{E}$ (polar) and $\mathbf{B}$ (axial) are one of each. For all rotation-based manipulations in the coming articles the two behave identically, so nothing you compute will care. It resurfaces exactly twice in the track: when we ask why $\mathbf{B}$ 's field lines circulate rather than radiate, and when parity symmetry appears in the relativistic formulation.

Scalar and Vector Fields

Where this lives today

The physical picture

The two field types, precisely

What makes a vector a vector

The operators that connect them

Limiting cases

Worked examples

Where the two field types run electromagnetism

Going further

Where this leads