Maxwell's equations and light

Derive the speed of light from the equations alone.

By 1860 electricity and magnetism were a handful of separate laws, each found on a bench with wires, magnets and glass jars. Light was a different subject altogether. Then James Clerk Maxwell wrote the laws down side by side and added a term to one of them. With the new term in place, the equations said something nobody had asked them: that electric and magnetic fields can travel through empty space as a wave, at a speed you can work out from two numbers measured with coils and capacitors. The speed came out as the speed of light.

This lesson gets to that number. Everything in this chapter has been leading here.

Four laws

All of electricity and magnetism fits in four statements about two fields, the electric field \(\mathbf{E}\) and the magnetic field \(\mathbf{B}\). You've met each one already. Here they are as pictures.

One picture for each law (on a narrow screen, one at a time: pick with the buttons). Amber is electric field and blue is magnetic field, as in the last lesson. Drag the white points and the blue one. These are cartoons: flat drawings of fields that fill space, with the number of lines chosen for show.
  • In picture 1, the dashed circle stands for a closed surface, like a bubble. Drag it over the + charge: 12 lines out and none in. Over the − charge: 12 in and none out. Anywhere else, every line that goes in comes out again, and the net is zero.
  • In picture 2, try to do the same to the magnet. Put the surface over its N end, over its S end, across its middle. The two counts are equal every time.
  • Watch picture 3 through a full cycle. The amber circles are only there while the blue field is changing, and they turn the other way when it shrinks. Then drag the blue point on the gauge yourself, fast and slow, and hold it still.
  • In picture 4, drag the dashed loop along the wire and over the gap between the plates. The blue field is there the whole way.
  • Now untick "Maxwell's addition" and drag the loop over the gap again.

In symbols, with a sentence each:

\[ \oint \mathbf{E}\cdot d\mathbf{A} = \frac{Q}{\varepsilon_0} \qquad\qquad \oint \mathbf{B}\cdot d\mathbf{A} = 0 \]
\[ \oint \mathbf{E}\cdot d\mathbf{l} = -\frac{d\Phi_B}{dt} \qquad\qquad \oint \mathbf{B}\cdot d\mathbf{l} = \mu_0 I + \mu_0\varepsilon_0\frac{d\Phi_E}{dt} \]
  1. Gauss's law. The left side is the flux of \(\mathbf{E}\) out through a closed surface: field times area, added up over the whole surface, counting outward as plus. It equals the charge \(Q\) inside, divided by a constant, \(\varepsilon_0\). Electric field lines start on + charges and end on − charges, so what comes out of a surface tells you what's in it.
  2. No magnetic charge. The same sum for \(\mathbf{B}\) is always zero. Magnetic field lines have nothing to start or end on, so whatever goes into a surface comes out again. They only make loops, as you saw inside the bar magnet.
  3. Faraday's law. The left side goes round a closed loop instead, adding up the part of \(\mathbf{E}\) that points along it: a line integral. That's the voltage round the loop, the \(\mathcal{E}\) of the last lesson. It equals minus the rate of change of \(\Phi_B\), the flux of \(\mathbf{B}\) through the loop. A changing magnetic field makes an electric field that circles it.
  4. Ampère's law, with Maxwell's addition. The same trip round a loop for \(\mathbf{B}\). \(I\) is the current passing through the loop, and the first term is why a magnetic field circles a wire. The second term is Maxwell's: \(\Phi_E\) is the flux of \(\mathbf{E}\) through the loop, and if it's changing, that makes a circling magnetic field too.

The two constants set how strong each field is. \(\varepsilon_0 = 8.854 \times 10^{-12}\) C²/(N·m²) is electrical: it's inside the \(k\) of the first lesson of this chapter, as \(k = \frac{1}{4\pi\varepsilon_0}\), and it's what you measure when you charge a capacitor. \(\mu_0 = 1.257 \times 10^{-6}\) T·m/A is magnetic: it's the \(\mu_0\) in the field round a wire, \(B = \frac{\mu_0 I}{2\pi r}\), and it's what you measure from the force between two coils.

Gauss's and Stokes's theorems turn each of these into a statement about a single point, using divergence and curl. That's the form you'll usually see printed, with \(\rho\) the charge in each cubic metre and \(\mathbf{J}\) the current through each square metre:

\[ \nabla\cdot\mathbf{E} = \frac{\rho}{\varepsilon_0} \qquad \nabla\cdot\mathbf{B} = 0 \qquad \nabla\times\mathbf{E} = -\frac{\partial\mathbf{B}}{\partial t} \qquad \nabla\times\mathbf{B} = \mu_0\mathbf{J} + \mu_0\varepsilon_0\frac{\partial\mathbf{E}}{\partial t} \]

They say the same four things. Maxwell himself wrote twenty equations in a clumsier notation, and it was Oliver Heaviside, in the 1880s, who boiled them down to these four.

The law that didn't add up

Before Maxwell, law 4 had only its first term: the magnetic field round a loop is set by the current passing through it. "Through it" means through any surface that has the loop as its edge, and for a wire it makes no difference which surface you pick. Lay a flat disc across the loop, or stretch a bag from it like a butterfly net: the wire has to pierce either one.

Now charge a capacitor, two metal plates facing each other across a gap, as in picture 4. Current flows along the wire to one plate and away from the other, but nothing crosses the gap. Put the loop round the wire just before the plate. A flat disc across it is pierced by the wire: current \(I\). A bag stretched from the same loop, passing between the plates, is pierced by nothing: current zero. One loop, two answers for its magnetic field. That's what you saw with the addition unticked: a field that stops dead at the edge of the gap.

Look at what the bag does have passing through it. Between the plates there's an electric field, and while the capacitor charges, it's growing.

Plates of area \(A\) holding charges \(+Q\) and \(-Q\) have an even field between them. This comes from Gauss's law.

\[ E = \frac{Q}{\varepsilon_0 A} \]

So the flux of \(\mathbf{E}\) across the gap is the field times the area, and the area cancels.

\[ \Phi_E = EA = \frac{Q}{\varepsilon_0} \]

The current in the wire is the rate at which charge arrives on the plate, \(I = \frac{dQ}{dt}\). So the flux grows at:

\[ \frac{d\Phi_E}{dt} = \frac{1}{\varepsilon_0}\frac{dQ}{dt} = \frac{I}{\varepsilon_0} \]

Multiply by \(\mu_0\varepsilon_0\), and Maxwell's term for the bag is exactly the wire's term for the disc.

\[ \mu_0\varepsilon_0\frac{d\Phi_E}{dt} = \mu_0 I \]

With the extra term, both surfaces give the same answer and the magnetic field carries on across the gap. Maxwell added it in a paper of 1861–62. He came to it by another road, a mechanical model of how the fields might work, and the argument above is how it's usually justified now. No experiment had asked for it: the effect was far too small to see on a bench at the time.

What the addition buys

Read laws 3 and 4 again with no charges and no currents anywhere. A changing \(\mathbf{B}\) makes \(\mathbf{E}\). A changing \(\mathbf{E}\) makes \(\mathbf{B}\). Each can be the other's source, so once a disturbance exists it can keep itself going, out in empty space, with nothing there at all. All that's needed is something to start it, and a charge that's shaken will do.

A charge shaken up and down, and the wave it sends out along one line. Along the line the picture is to scale and runs in real time: the distances along the bottom are real kilometres. The charge's own swing is not to scale. It's drawn hugely too big, or you couldn't see it. Drag the charge, or let it be shaken for you. A white tick leaves the charge once every second and rides along with the wave. The wave is drawn following where the charge was, which for a steady shake has the right shape (a real wave follows the charge's acceleration), and keeping its height, where a real one fades as it spreads. Both fields are drawn the same size.
  • Let it run. The electric field (amber) waves up and down, and the magnetic field (blue) waves towards you and away, square to it. Both are square to the direction the wave travels, and they rise and fall together.
  • Follow one white tick from the charge to the Moon's distance. It takes 1.28 seconds, whatever else is going on.
  • Slide "Shake it for me" from 0.8 to 1.6 times a second. The wavelength halves, from 374,740 km to 187,370 km. The crests are closer together and no faster: the white ticks haven't changed their spacing at all.
  • Slide it down to 0.2. Now one wavelength is longer than the whole picture.
  • Drag the charge up and down yourself, slowly and then as fast as you can. You set how close the crests are. You have no say in how fast they travel.
  • Hold the charge still. No new wave sets off, and what's already out there carries on without you.

Why there's only one speed

Here's the argument for the speed, using nothing but the two circling laws. Picture a slab of field moving to the right through empty space at some speed \(v\), with a flat front. Behind the front the electric field points up, with strength \(E\), and the magnetic field points towards you, with strength \(B\). Ahead of the front there's nothing yet. At the front, both fields are changing from nothing to something, so both laws apply there. Each has to give back exactly the field the other needs.

Stand a rectangle of height \(h\) upright, half behind the front and half ahead of it, and go round it adding up \(\mathbf{E}\) along each side. The top and bottom run square to \(\mathbf{E}\) and add nothing. The side ahead of the front has no field. Only the side behind counts.

\[ \oint \mathbf{E}\cdot d\mathbf{l} = Eh \]

\(\mathbf{B}\) passes through this rectangle. In a short time \(\Delta t\) the front moves on by \(v\,\Delta t\), so a strip of area \(h \times v\,\Delta t\) fills with field. The flux through the rectangle grows by \(B \times h\,v\,\Delta t\), which is a rate of:

\[ \frac{d\Phi_B}{dt} = Bhv \]

Faraday's law says those two are equal in size. The \(h\) cancels.

\[ Eh = Bhv \quad\Rightarrow\quad E = vB \]

Now lay a second rectangle flat, so that \(\mathbf{B}\) runs along its sides and \(\mathbf{E}\) passes through it. The same reasoning, with the two fields swapped: going round gives \(Bh\), and the flux of \(\mathbf{E}\) grows at \(Ehv\). There's no current, so law 4 is only Maxwell's term.

\[ Bh = \mu_0\varepsilon_0\,Ehv \quad\Rightarrow\quad B = \mu_0\varepsilon_0\,vE \]

Put that \(B\) into \(E = vB\).

\[ E = v \times \mu_0\varepsilon_0\,vE = \mu_0\varepsilon_0\,v^2E \]

\(E\) cancels from both sides, and what's left has no choice in it.

\[ v^2 = \frac{1}{\mu_0\varepsilon_0} \quad\Rightarrow\quad v = \frac{1}{\sqrt{\mu_0\varepsilon_0}} \]

The slab can't travel at any other speed. Slower or faster, and the field each law makes isn't the field the other one started from. The strengths of the fields cancelled out, and nothing about the charge that started the wave was ever in the argument. That's why shaking faster in the figure changed the spacing and never the speed. The same argument fixes the directions: \(\mathbf{E}\), \(\mathbf{B}\) and the direction of travel all have to be at right angles, with the wave moving along \(\mathbf{E}\times\mathbf{B}\), the cross product. And it fixes their sizes: \(E = vB\).

A flat-fronted slab is the easiest case, and a wave of any shape can be built from thin slabs. The full derivation takes the curl of law 3, puts law 4 into it, and arrives at the wave equation, \(\frac{\partial^2 E}{\partial x^2} = \mu_0\varepsilon_0\frac{\partial^2 E}{\partial t^2}\), whose solutions are shapes of any kind sliding along at \(\frac{1}{\sqrt{\mu_0\varepsilon_0}}\). It needs the point-by-point form of the laws and partial derivatives in space and time together. That equation is what the wave equation, later in the course, is about.

The number

In Maxwell's day both constants came from the bench: \(\mu_0\) from the forces between currents, and \(\varepsilon_0\) from charges and capacitors. Neither of those measurements involved light, or anything moving fast.

The two constants, to seven figures:

\[ \mu_0 = 1.256637 \times 10^{-6} \text{ T m/A} \qquad \varepsilon_0 = 8.854188 \times 10^{-12} \text{ C}^2/(\text{N m}^2) \]

Multiply them. The numbers multiply and the powers of ten add: \(1.256637 \times 8.854188 = 11.12650\), and \(10^{-6} \times 10^{-12} = 10^{-18}\).

\[ \mu_0\varepsilon_0 = 11.12650 \times 10^{-18} \text{ s}^2/\text{m}^2 \]

The units have collapsed to seconds squared over metres squared, which is one over a speed squared, as it has to be. Now the square root. The power of ten was left as \(10^{-18}\) because an even power is easy to halve: \(\sqrt{10^{-18}} = 10^{-9}\).

\[ \sqrt{\mu_0\varepsilon_0} = \sqrt{11.12650} \times 10^{-9} = 3.335641 \times 10^{-9} \text{ s/m} \]

That's the time the wave takes to cross one metre, about 3.3 billionths of a second. Turn it over to get metres per second.

\[ v = \frac{1}{3.335641 \times 10^{-9}} = 2.99792 \times 10^{8} \text{ m/s} \]

That's 299,792 km/s. The speed of light is 299,792.458 km/s. From here on the wave's speed gets its usual name, \(c\).

\[ c = \frac{1}{\sqrt{\mu_0\varepsilon_0}} = 299{,}792 \text{ km/s} \qquad\qquad E = cB \]

Maxwell did this sum in that same paper of 1861–62. The best electrical measurements he had were Wilhelm Weber and Rudolf Kohlrausch's, made a few years earlier with a charged jar and a magnetic needle, and they gave him 310,740 km/s. Hippolyte Fizeau had measured the speed of light in 1849, with a spinning toothed wheel and a mirror more than 8 km away, and the figure Maxwell quotes for it is 314,858 km/s. The two had been measured in ways that had nothing to do with each other. Maxwell wrote: "we can scarcely avoid the inference that light consists in the transverse undulations of the same medium which is the cause of electric and magnetic phenomena." Three years later he put it more plainly: light is "an electromagnetic disturbance propagated through the field according to electromagnetic laws".

He died in 1879, aged 48, before anyone had made such a wave on purpose. In 1887 and 1888 Heinrich Hertz did it. A spark jumping across a gap in one piece of apparatus made a much smaller spark jump across a gap in a loop of wire some metres away, with nothing joining the two. He showed that whatever crossed the room could be reflected, focused and bent as light can, and that it travelled at the speed of light. Those were radio waves.

One caution about the sum above. Today the speed of light is fixed by definition, and the value of \(\varepsilon_0\) printed above is worked out from it, so this sum is bound to give \(c\) back. In Maxwell's time the three numbers were separate measurements, and their agreeing was a discovery.

One wave, many names

In the figure, a crest sets off each time the charge completes a shake. If it shakes \(f\) times a second (the frequency, in hertz), the crests set off \(\frac{1}{f}\) seconds apart, and each has travelled \(c \times \frac{1}{f}\) by the time the next one leaves. That distance is the wavelength, \(\lambda\) (lambda):

\[ \lambda = \frac{c}{f} \qquad\qquad c = f\lambda \]

Maxwell's equations set no limit on \(f\). Shake charges at any rate and you get the same kind of wave at the same speed, with a wavelength to match. The names below are only labels for different stretches of one unbroken range, longest waves first:

A radio wave and a gamma ray differ only in wavelength. For the charge in the figure to make visible light, you'd have to shake it about \(5 \times 10^{14}\) times a second. Nothing you can hold moves like that, but the electrons in atoms do.

Electricity, magnetism and light had been three subjects, and now they were one. Physicists have been looking for the next such join ever since. It's also where the trouble behind special relativity starts. The equations give a speed, \(c\), and don't say what it's measured against. The speed of sound is measured against the air, so for decades everyone assumed light had a medium too, and looked for signs of the Earth moving through it. None was ever found. In 1905 Einstein took the equations at their word: light moves at \(c\) for everyone, however they themselves are moving. For that to be true, time itself has to behave differently, and the time dilation lesson shows how.

Problems

A closed surface has 5 electric field lines leaving it and 2 entering. Is the charge inside positive, negative or zero? Could a closed surface have 5 magnetic field lines leaving it and 2 entering?

Answer

Net 3 lines out, so there's positive charge inside. For a magnetic field it's impossible: law 2 says the numbers in and out are always equal, because there's no magnetic charge for the extra 3 lines to have started on.

Find the wavelength of an FM radio station at 100 MHz (\(1.00 \times 10^8\) Hz), and of a microwave oven at 2.45 GHz (\(2.45 \times 10^9\) Hz). Use \(c = 3.00 \times 10^8\) m/s.

Answer

\(\lambda = \frac{c}{f}\). For the radio station, \(\frac{3.00 \times 10^8}{1.00 \times 10^8} = 3.00\) m. For the oven, \(\frac{3.00 \times 10^8}{2.45 \times 10^9} \approx 0.122\) m, about 12 cm.

The Sun is \(1.50 \times 10^{11}\) m away. How long does its light take to reach us? Use \(c = 3.00 \times 10^8\) m/s.

Answer

\(t = \frac{d}{c} = \frac{1.50 \times 10^{11}}{3.00 \times 10^8} = 500\) s, a little over 8 minutes. If the Sun went out, the sky would stay bright for 8 minutes more.

In full sunlight the electric field of the light swings up to roughly 1,000 V/m. How strong does the magnetic field get? Compare it with the Earth's magnetic field, about \(5 \times 10^{-5}\) T. Use \(c = 3.00 \times 10^8\) m/s.

Answer

\(E = cB\), so \(B = \frac{E}{c} = \frac{1{,}000}{3.00 \times 10^8} \approx 3.3 \times 10^{-6}\) T. That's about a fifteenth of the Earth's field. In everyday units the magnetic part of light looks tiny, though it carries exactly half the wave's energy.

The first lesson of this chapter gave the constant in Coulomb's law as \(k = 8.99 \times 10^9\) N·m²/C², and \(k = \frac{1}{4\pi\varepsilon_0}\). To the accuracy needed here, \(\mu_0 = 4\pi \times 10^{-7}\) T·m/A. Show that \(c = \sqrt{\frac{4\pi k}{\mu_0}}\), and get the speed of light from \(k\).

Answer

Rearrange \(k = \frac{1}{4\pi\varepsilon_0}\) to get \(\varepsilon_0 = \frac{1}{4\pi k}\). Then \(\mu_0\varepsilon_0 = \frac{\mu_0}{4\pi k}\), and \(c = \frac{1}{\sqrt{\mu_0\varepsilon_0}} = \sqrt{\frac{4\pi k}{\mu_0}}\).

Put in the numbers. The \(4\pi\) cancels: \(\frac{4\pi k}{\mu_0} = \frac{4\pi \times 8.99 \times 10^9}{4\pi \times 10^{-7}} = 8.99 \times 10^{16}\). An even power of ten again, so \(c = \sqrt{8.99} \times 10^8 \approx 3.00 \times 10^8\) m/s. The force between two charges sitting still already has the speed of light inside it.

A capacitor is being charged by a steady current of 0.50 A. How fast is the flux of \(\mathbf{E}\) between its plates growing? What is the magnetic field 2.0 cm from the wire, and what is it 2.0 cm from the centre of the gap, if the plates are smaller than that? Use \(\varepsilon_0 = 8.85 \times 10^{-12}\) C²/(N·m²) and \(\mu_0 = 4\pi \times 10^{-7}\) T·m/A.

Answer

\(\frac{d\Phi_E}{dt} = \frac{I}{\varepsilon_0} = \frac{0.50}{8.85 \times 10^{-12}} \approx 5.6 \times 10^{10}\) V·m/s (a volt-metre is the N m²/C of the Gauss's law lesson).

Round the wire, \(B = \frac{\mu_0 I}{2\pi r} = \frac{4\pi \times 10^{-7} \times 0.50}{2\pi \times 0.020}\). The \(\pi\) cancels, leaving \(\frac{2 \times 10^{-7} \times 0.50}{0.020} = 5.0 \times 10^{-6}\) T.

Round the gap, the loop of radius 2.0 cm has all of the changing flux passing through it, and Maxwell's term \(\mu_0\varepsilon_0\frac{d\Phi_E}{dt}\) is exactly \(\mu_0 I\). So the field is the same, \(5.0 \times 10^{-6}\) T. That was the whole point of the addition.

Learn more

An unhandled error has occurred. Reload 🗙