Spacetime Metric — Season 1 — 02-special-relativity-and-four-vectors

Transcript

Cold open (≈ 90 seconds)

Two people. Two stopwatches. Two rulers. One of them is standing on a platform. The other is rolling past on a train. They both watch the same event — a flashbulb goes off, somewhere between them — and they both write down what they measured. How long the flash lasted. How far away it happened. How fast the light from it moved.

When they compare notes afterward, they do not agree. The one on the platform writes down one duration; the one on the train writes down a different duration. The one on the platform writes down one distance; the one on the train writes down a different distance. They are honest. Their stopwatches are calibrated. Their rulers are accurate. And they disagree.

For about two hundred and fifty years, physics treated that disagreement as a measurement problem. Somebody had to be wrong. The job of the theorist was to figure out which one.

In 1905, a twenty-six-year-old patent clerk in Bern, Switzerland, published a paper that said something different. He said: neither one is wrong. They are both measuring honestly. The reason they disagree is that they are measuring two different things. And underneath the disagreement, there is something deeper — something neither of them was measuring directly — that the two of them, if they did the algebra carefully, would agree on exactly.

This is Lecture Two. The deeper thing is called the spacetime interval. The geometry that makes it agree is called Minkowski space. And the bookkeeping that lets you write down anything physical in a way that the platform observer and the train observer will both accept — that bookkeeping is called the four-vector.

By the end of the lecture, you will be able to read all of it.

Recap and the stretch (≈ 2.5 minutes)

Last lecture, we built up a single object. A metric. The rule for converting infinitesimally small coordinate steps into a physical squared-distance. On a flat plane we wrote it as ds^2 = dx^2 + dy^2. On a sphere we wrote it as ds^2 = dtheta^2 + sin^2theta, dphi^2. In its full general form we wrote it as ds^2 = g{munu}, dx^mu, dx^nu, where g{munu} is a table of numbers that depends on where you are.

Every example in Lecture One was a metric on space. The flat sheet. The surface of a sphere. A patch of an undulating two-dimensional surface drifting through space. You can stretch any of those examples to three spatial dimensions without trouble — the bookkeeping just gets longer.

But the lecture you are listening to now is about the next stretch. Not from two spatial dimensions to three. From space to spacetime. From a metric that two observers standing still in the same room would agree on, to a metric that two observers moving at different speeds will also agree on.

That second stretch is harder than it sounds. It is harder because at first glance, time and space do not look like things you should be able to put into the same metric. Time has units of seconds. Space has units of meters. A stopwatch is not a ruler. They feel like different kinds of quantity. And before 1905, they were treated as different kinds of quantity.

The leap of 1905 was to put them in the same diagram. To insist that there is a single object, four-dimensional, that contains both, and that the rule for measuring distance in that object is what physical observers actually agree about. The disagreement at the train platform happens because the platform observer and the train observer are projecting that four-dimensional object onto their own private three-plus-one split — three of space, one of time — and their splits are tilted with respect to each other. The four-dimensional object underneath is the same. The split is what changes.

This lecture builds that four-dimensional object. Then it builds the bookkeeping you write physical quantities in, so they live in the four-dimensional object instead of in any particular observer's split. Then it hands you the next-lecture handle: the energy-momentum tensor. That object is going to sit on the right-hand side of Einstein's field equations in Lecture 3, and it is the thing geometry is responding to.

But before any of that — we have to talk about why the everyday rule for adding velocities, the rule you've used since middle school, has to break.

Why Galilean velocity addition has to break (≈ 6 minutes)

Here is the rule you grew up with. You are on a train, going forty miles an hour. You walk forward, inside the train, at three miles an hour. How fast are you moving relative to the platform? Forty-three. You add the speeds. Same idea: a baseball thrown forward at ninety miles an hour from a car moving forward at thirty — the ball is doing a hundred and twenty relative to the road.

This rule has a name. It is Galilean velocity addition, after Galileo, who articulated the underlying principle of relativity in 1632 in his Dialogue Concerning the Two Chief World Systems: the laws of mechanics are the same in any inertial frame, so a sailor below deck on a smoothly-moving ship cannot, by any mechanical experiment, tell whether the ship is moving or at rest. If you accept that principle, and you accept the everyday notion of universal time — the same clock-reading for everybody, regardless of how they're moving — then velocity addition follows.

Your velocity relative to the platform is your velocity relative to the train, plus the train's velocity relative to the platform. Done.

The rule works. It works for trains. It works for thrown baseballs. It works for cars and bicycles and walking pace. It works for everything in human experience, up to the speed of a fighter jet, up to the speed of a rifle bullet, up to the speed of a hypersonic missile. It is an empirically excellent rule across the full range of speeds anybody had ever measured carefully, for two hundred and fifty years after Galileo.

And then, in the second half of the nineteenth century, somebody started measuring light.

The trouble was this. By the 1860s, James Clerk Maxwell had written down a set of four equations — what we now call Maxwell's equations — that describe how electric and magnetic fields propagate. Out of those equations falls a specific number. A speed. The speed at which an electromagnetic wave travels through empty space. That number is approximately three hundred thousand kilometers per second. We call it c. And Maxwell's equations don't say "c relative to whom." The number just sits there, baked into the structure of electromagnetism.

That is a problem for Galilean addition. Because if light travels at c, and you're moving toward a flashlight at half of c, then by the everyday rule the light should arrive at you doing one and a half c. And if you're moving away from the same flashlight at half of c, the light should arrive doing half c. Those are two different answers, depending on your motion. But Maxwell's equations give one answer. The same one.

So either Maxwell's equations are only true in one special frame — the frame of some hypothetical medium, which physicists in the late 1800s called the luminiferous aether — or Galilean addition is wrong about light.

In 1887, in a basement on the campus of what was then called the Case School of Applied Science in Cleveland, Ohio, two experimentalists — Albert Michelson and Edward Morley — built an apparatus precise enough to settle the question. They split a beam of light into two perpendicular paths, bounced them off mirrors, recombined them, and looked for the interference pattern. As the Earth moves around the Sun at about thirty kilometers per second, the apparatus is moving through the supposed aether in different directions at different times of year.

If the aether existed and Galilean addition were right, the speed of light along the apparatus's direction of motion should differ from the speed of light perpendicular to it, and the interference pattern should shift across the year.

The pattern did not shift. To the precision of the apparatus — which was very good — the speed of light came out the same in every direction, at every time of year, regardless of how the Earth was moving. The experiment was refined and repeated for decades. The null result held.

So now you have a choice, as a physicist in 1900. You can keep Galilean velocity addition and keep universal time and try to patch Maxwell's equations — which several very good physicists, including Lorentz and Poincaré, attempted. Or you can keep Maxwell's equations exactly as Maxwell wrote them, accept that the speed of light is the same in every inertial frame as an empirical input, and figure out what has to give somewhere else.

In 1905 Einstein took the second option, and what had to give was the assumption of universal time. Two observers in motion relative to each other do not share a clock. The clock-reading and the ruler-reading mix into each other in a specific, calculable way when you change from one observer's frame to another's. The mixing rule is called the Lorentz transformation — Hendrik Lorentz, the Dutch physicist, had written down the correct transformation equations in 1904, but he interpreted them as a quirk of how rulers and clocks interact with the aether; Einstein's 1905 paper accepted them as a statement about space and time themselves.

The Lorentz transformation, in its simplest form — one direction of relative motion, call it the x-direction, with relative speed v — is:

$t' = gamma left(t - frac{vx}{c^2}right), quad x' = gamma (x - vt), quad y' = y, quad z' = z$

where gamma = 1/sqrt{1 - v^2/c^2} is the Lorentz factor. For everyday speeds — v much smaller than c — gamma is essentially one, the vx/c^2 term in the time equation is essentially zero, and the equations collapse back to "t' = t, x' = x - vt" — Galileo's rule. The departure from Galileo is hidden inside corrections that scale as v^2/c^2. At everyday speeds those corrections are smaller than parts per trillion. At a tenth of the speed of light, they're a percent. At nine-tenths the speed of light, gamma is about 2.3, and the corrections are the whole picture.

That is the empirical and historical setup. The rest of this lecture is about what those Lorentz transformations are doing geometrically — and the answer, due to Hermann Minkowski in 1908, is that they are rotations in a four-dimensional spacetime, and there is a metric on that spacetime that they leave alone. That metric is what the platform observer and the train observer secretly agree on, even when they're disagreeing about everything else.

The Minkowski line element (≈ 6 minutes)

Here is the central object. The metric on flat spacetime — the spacetime of special relativity, with no gravity yet — written in the same line-element notation Lecture One used. Brace, but the equation is short.

$ds^2 = -c^2, dt^2 + dx^2 + dy^2 + dz^2$

That equation is the entire content of special relativity. Everything else in this lecture is bookkeeping to make it usable. So let's name every piece.

The left side, ds^2 — same little squared-interval we met in Lecture One. The Greek-letter generalization holds. It is the rule for converting four infinitesimal coordinate steps — three in space, one in time — into a single physical quantity that observers will agree on.

The four terms on the right are the four coordinate steps. One time step, dt, multiplied by -c^2. Three space steps, dx, dy, dz, each just squared with a plus sign. That's the whole metric. In the g_{munu} table notation from Lecture One, the Minkowski metric is a four-by-four diagonal table — -c^2 in the time-time slot, +1 in each of the three space-space slots, zeros everywhere off the diagonal.

Three questions worth asking about this metric. The same three questions we asked of the Pythagorean rule on the flat plane, because we are doing the same kind of work.

First — why the minus sign in front of the time term. This is the single most important sign in twentieth-century physics. The minus sign is the entire reason time is not just a fourth direction of space. If the time term came with a plus sign — if the metric were ds^2 = c^2 dt^2 + dx^2 + dy^2 + dz^2, all four terms positive — then spacetime would be four-dimensional Euclidean space. There would be no light cone. There would be no past and future. There would be no notion of cause and effect built into the geometry.

The minus sign is what makes time different from space, while still letting time and space sit in the same equation.

To see what the minus sign actually does, look at the simplest case. Pretend you have one direction of space and one of time, so the metric simplifies to ds^2 = -c^2 dt^2 + dx^2. Now ask: what does it mean for ds^2 to be zero?

Set -c^2 dt^2 + dx^2 = 0. Rearrange: dx^2 = c^2 dt^2, which means dx/dt = pm c. The coordinate steps that produce a zero interval are exactly the steps that correspond to motion at the speed of light. So the metric, all by itself, picks out the speed of light as the speed for which "distance in spacetime" is zero.

That is the picture physicists call a spacetime diagram. Space across, time up — time multiplied by c so the units match. The two diagonal lines at forty-five degrees are the paths light takes. Everything inside the upper wedge is somewhere you could get to from the origin by moving slower than light. Everything inside the lower wedge is somewhere you could have come from by moving slower than light. Everything in the side wedges, left and right, is somewhere you cannot reach from the origin without exceeding the speed of light.

The minus sign in the metric is what carves the diagram into those wedges. And it gives us a vocabulary for the three kinds of separation between two events.

If two events have ds^2 < 0 — squared interval negative — they are timelike-separated. The time difference between them dominates the space difference. A signal moving slower than light can connect them. The platform observer's stopwatch ticking once and the same stopwatch ticking twice are timelike-separated events: same place in space, different time, and ds^2 = -c^2 dt^2 which is negative.

If two events have ds^2 > 0 — squared interval positive — they are spacelike-separated. The space difference dominates the time difference. No signal traveling at or below the speed of light can connect them. Two simultaneous events at opposite ends of a long room are spacelike-separated.

If two events have ds^2 = 0 — squared interval exactly zero — they are lightlike-separated, or null-separated. A light ray can connect them and nothing slower than light can. The flashbulb going off and the photon from that flash reaching your eye are lightlike-separated events.

Second question — what does it mean for two observers to agree on ds^2 when they disagree on dt and dx. This is where the Lorentz transformation earns its keep. Take any two events. Compute ds^2 in the platform frame, using the platform observer's clock and the platform observer's ruler. You get a number. Then transform the time and space coordinates to the train frame using the Lorentz transformation, and compute ds^2 in the train frame. You get the same number. The proof is two lines of algebra; it falls out of the Lorentz transformation because the Lorentz transformation was designed to leave that combination invariant.

Different observers measure different dt's and different dx's. They write down the same ds^2.

That is the geometric meaning of special relativity. The platform observer and the train observer are two different coordinate systems on a single four-dimensional spacetime. The Lorentz transformation is the change-of-coordinates rule between them. And the Minkowski metric is the structure on the spacetime that survives the change.

Third question — what about the proper time, the thing a clock actually reads. For a timelike-separated pair of events — say, two ticks of a stopwatch carried by a moving observer — there is one frame that is special: the frame in which the stopwatch is at rest. In that frame, the space difference between the two ticks is zero. The metric reduces to ds^2 = -c^2 dt^2, where dt is the time the stopwatch itself registers between the two ticks. Physicists call that dtau, the proper time — the time as read by a clock that is present at both events.

So for timelike intervals, ds^2 = -c^2 dtau^2. The squared interval and the squared proper time differ by a factor of minus c^2. Proper time is the geometry's natural notion of "how much time elapsed for the thing that was actually there." Every observer, regardless of motion, will compute the same proper time for the same clock. That is the deeper agreement underneath the disagreement at the train platform. The platform observer and the train observer disagree on whose clock ticked when. They agree on what the moving clock itself read between any two of its ticks.

The whole edifice of special relativity sits on that one invariance. Three questions in, we now have the geometry. The next move is to bring in a physicist who has spent his career thinking about what this geometry buys you, mathematically and physically.

The Thorne handoff (≈ 6 minutes)

The voice I want to bring in here belongs to a physicist whose textbook is the reason most working relativists know what they know. Kip Thorne. Richard P. Feynman Professor of Theoretical Physics Emeritus at Caltech. BS Caltech 1962, PhD Princeton 1965 under John Wheeler. 2017 Nobel Prize in Physics, shared with Rainer Weiss and Barry Barish, for decisive contributions to the LIGO detector and the observation of gravitational waves. Co-author, with Charles Misner and the same John Wheeler, of the textbook Gravitation, published by W. H.

Freeman in 1973 — a book physicists call "MTW," after its authors' initials, and which remains, fifty-three years on, the canonical graduate text in general relativity.

Thorne has spent six decades thinking about what the metric language does for physics. In Black Holes and Time Warps, his 1994 trade book published by W. W. Norton, he sets out the geometric picture of relativity for a general audience in language that has been borrowed by every popular treatment since. What follows is a paraphrase, in his published-position voice — not a verbatim quotation, but a synthesis of the framing he has used across MTW, the Nobel-lecture material, and Black Holes and Time Warps — of what the Minkowski metric is doing.

The point I want you to take from the Minkowski metric is this. Special relativity does not say that observers in motion measure the same things. It says the opposite. Two observers in relative motion will measure different elapsed times between the same pair of events; they will measure different spatial distances; they will measure different sequences of which event happened first. They will disagree on essentially every coordinate-dependent statement you can ask them about.

What they will not disagree on is the geometry. There is a single four-dimensional spacetime. Each observer has their own way of slicing it — three dimensions they call space, one they call time — and their slicings are tilted with respect to each other. The slicing is observer-dependent. The four-dimensional thing being sliced is not.

The metric is what tells you what is observer-dependent and what is not. A coordinate step like dt or dx is observer-dependent; it depends on the slicing. The combination -c^2 dt^2 + dx^2 + dy^2 + dz^2 — the squared interval — is observer-independent. It is the same number for every inertial observer. That is the content of Lorentz invariance.

Think about what that does to your notion of physical reality. Before 1905, physicists treated time and space as separate. Each had its own metric. The clock that ticked on the platform was the same clock as the one that ticked on the train, just read by two different people. The relativity of simultaneity changed that. The clock on the train is not the clock on the platform. They are two clocks, two different slicings of the same underlying object.

Hermann Minkowski, in 1908 in Cologne, drew the right conclusion. He said: from now on, space by itself, and time by itself, are doomed to fade away into mere shadows. Only a union of the two will preserve an independent reality. That is the famous Minkowski opening line, and it is not rhetoric. It is the geometric statement that what is real, in special relativity, is the four-dimensional thing — and what each observer measures is a projection of it.

The reason this matters for everything that comes after is that once you have a metric on a four-dimensional object, all the geometric machinery you developed for two-dimensional surfaces in Lecture One — what's a straight line, what's a distance, how do you parallel-transport a vector — carries over. The mathematics does not care that one direction has a minus sign. The mathematics cares that you have a metric. You have a metric. So you can do geometry on spacetime exactly the way you did geometry on a sphere.

When Einstein went from special relativity in 1905 to general relativity in 1915, the move he made was to let the metric vary from point to point. The flat Minkowski metric — the one you have just learned — is the spacetime of special relativity. The general g_{munu} — the one that varies, that responds to mass and energy — is the spacetime of general relativity. The Minkowski metric is what you recover, locally, in any small enough patch of any spacetime, because every smooth manifold looks flat if you zoom in enough.

That is the geometric architecture the Misner-Thorne-Wheeler textbook lays out. Spacetime is the arena. The metric is the structure that makes it a geometry rather than a set of points. Lorentz invariance is the local symmetry of every patch. And every observer who carries a clock and a ruler is measuring projections — projections of a four-dimensional object onto their particular slicing of it. The four-dimensional object is what you should think of as physically real.

The geometric statement Thorne has just walked through is what makes the bookkeeping of the next section possible. If observers disagree about coordinates but agree about geometry, then the way to write down a physical quantity — a velocity, a momentum, an energy — is in a form that lives in the geometry, not in any particular observer's coordinates. That form is the four-vector.

Four-vectors as bookkeeping (≈ 6 minutes)

Here is the bookkeeping. In ordinary three-dimensional physics, a vector is a quantity with three components — three numbers, one for each direction of space. The velocity of a thrown baseball has three components: how fast it's moving east, how fast north, how fast up. The momentum of the baseball is mass times velocity, also three components. When you switch from one observer to another — say, you rotate your coordinate axes — the three components of the vector mix together in a calculable way, but the vector itself, the underlying arrow, doesn't change.

Special relativity asks you to do the same trick with four components instead of three. A four-vector has one time component plus three space components, and when you switch from one inertial observer to another by a Lorentz transformation, the four components mix together — but the four-vector itself, the underlying arrow in spacetime, doesn't change.

The simplest four-vector is the one that records where and when an event happens. Call it the position four-vector. In the platform observer's frame, its four components are (ct, x, y, z). The factor of c in front of t is there so all four components have the same units. Physicists usually write this with a Greek index, so the position four-vector becomes x^mu, where mu runs from zero to three: x^0 = ct, x^1 = x, x^2 = y, x^3 = z. The Greek index is the same kind of bookkeeping label we met in Lecture One — it labels which component, not what number.

Now do the same trick with motion. In ordinary physics, velocity is the rate of change of position — derivative of (x, y, z) with respect to time. In special relativity, there is a question of whose time: the platform observer's t, or the train observer's t, or the proper time tau that the moving thing itself carries?

The right answer is proper time. If you take the derivative of the position four-vector with respect to the proper time of the moving object — the time its own clock reads — you get a quantity that all inertial observers will agree about, because they all agree about that clock. That quantity is the four-velocity, u^mu:

$u^mu = frac{dx^mu}{dtau}$

The four-velocity is what you get when you ask, "how is the moving thing's position in spacetime changing, per unit of its own clock-reading?" Its zero component, u^0, is c cdot dt/dtau = c cdot gamma — gamma factor times the speed of light. Its space components are gamma times the ordinary three-velocity. So at everyday speeds where gamma is essentially one, the four-velocity is basically (c, vx, vy, v_z) — your usual velocity, with the speed of light tacked on as the zero component. As speed approaches c, gamma blows up, and the four-velocity gets more interesting.

Next: momentum. In ordinary physics, momentum is mass times velocity. In special relativity, four-momentum is mass times four-velocity:

$p^mu = m u^mu$

where m is the rest mass — the mass an observer measures when the thing is at rest in their frame. The four-momentum has four components. The zero component is m c gamma. The three space components are gamma m times the ordinary three-velocity. If you do a Taylor expansion of gamma for small v/c, the zero component becomes approximately mc + frac{1}{2} m v^2 / c + ldots, and you recognize the second term: kinetic energy, divided by c. So the zero component of the four-momentum, multiplied by c, is the total energy of the moving object. Physicists write this as

$p^0 = E/c$

and the four-momentum, in its standard form, is

$p^mu = (E/c, px, py, p_z)$

That arrangement is what makes special relativity say something striking. Energy and momentum are not separate quantities. They are four components of a single four-vector. An observer at rest with respect to a particle measures one energy and zero momentum. An observer moving with respect to the particle measures a different energy and a non-zero momentum. The two observers are measuring different splittings of the same underlying four-momentum into its time piece and its space pieces. The four-momentum itself is the same.

Now the punchline. Once you have a four-vector and a metric, you can compute a squared length of the four-vector — using the same metric that gave you the squared interval between events. For the four-momentum, that squared length, in the convention of this lecture, comes out to:

$p^mu p_mu = -m^2 c^2$

The downstairs index here — pmu with the index lowered — is the metric-contracted version of p^mu, which in flat Minkowski space is pmu = (-E/c, px, py, pz) — the time component flips sign, the space components don't. When you compute p^mu pmu, you sum over mu and use Einstein's summation convention from Lecture One. Three lines of algebra later you get -E^2/c^2 + px^2 + py^2 + p_z^2 = -m^2 c^2. Rearrange, multiply through by -c^2, and you have the equation that students of physics learn under a slightly different alias:

$E^2 = (pc)^2 + (mc^2)^2$

That is the relativistic energy-momentum relation. For a particle at rest — no momentum, p = 0 — it collapses to E = mc^2, the most famous equation in physics. For a massless particle like a photon — no rest mass, m = 0 — it becomes E = pc, the energy-momentum relation of light. For everything in between, this is the equation that ties energy, momentum, and rest mass into a single invariant.

And the geometric content is the part to hold onto. The right-hand side, -m^2 c^2, is observer-independent — rest mass squared is the same number in every frame. The left-hand side, p^mu p_mu, is built from the four-momentum and the metric, both of which transform under Lorentz transformations in ways that exactly cancel. So the equation is true in every frame, even though E and p separately depend on which frame you're in.

That is what "four-vector bookkeeping" buys you. You write a physical quantity once, in the language of four-vectors. Every observer can then read off their own slicing of it. The slicings disagree. The four-vector doesn't.

The energy-momentum tensor (≈ 4 minutes)

There is one more object to put on the table, because everything in Block B and Block D of this series is going to come back to it. The four-momentum, p^mu, is what you write for a single particle moving through spacetime. But matter is not usually one particle. Matter is fields, or fluids, or collections — energy and momentum spread out over a volume.

For that case, you need something one step up. Not a four-vector, but a four-by-four table of four-vectors. An object with two indices instead of one. Physicists call it the energy-momentum tensor, or stress-energy tensor, and write it T^{munu} — or with both indices lowered, T_{munu}.

The two indices each run from zero to three, so the table has sixteen entries. By symmetry, only ten of them are independent.

Each entry has a physical meaning. T^{00} is the energy density — how much energy is sitting in a small volume at a given point. T^{0i}, where i runs over the three space directions, is the momentum density — how much momentum is flowing in each spatial direction. T^{ii}, the diagonal space components, are pressure — force per area pushing outward in each direction. T^{ij} off the diagonal, where i ne j, are shear stresses — the same quantities a structural engineer thinks about when they're asking whether a beam will twist.

What the energy-momentum tensor is, in one sentence: it is the complete bookkeeping of how energy and momentum are distributed and flowing through a region of spacetime. Every kind of physical matter — dust, gas, electromagnetic fields, the quantum vacuum, dark energy, anything you can write down a Lagrangian for — has an energy-momentum tensor.

What it does, in this lecture, is set up the next lecture. Because here is the bridge. In 1915, Einstein wrote down a field equation that says, roughly,

$G{munu} = frac{8pi G}{c^4}, T{munu}$

The left side, G{munu}, is the Einstein tensor — a specific combination of the metric g{munu} and its derivatives, which we'll build in Lecture 3. It encodes the local curvature of spacetime. The right side is the energy-momentum tensor, times some constants. The equation, in plain language, says: the curvature of spacetime, at every point, is determined by the energy and momentum present at that point. Geometry on the left. Matter on the right. The equation is the link.

The metric is the geometry. The energy-momentum tensor is what the geometry responds to. Lecture 3 is the lecture about how the response works. Lecture 6 is the lecture about what energy-momentum tensors the universe will and will not let you have. Lecture 8 is the lecture about a region of the quantum vacuum where the T^{ii} entries have been measured to be negative — the Casimir effect, the cleanest experimental window onto whether the source side of the Einstein equation can be engineered. Lecture 10 is the lecture about whether anybody has, as a published claim, manipulated the local geometry by engineering T_{munu}.

Hold the picture. The four-vector p^mu is the bookkeeping for one particle. The tensor T{munu} is the bookkeeping for an entire distribution of energy and momentum. The Minkowski metric is the geometry of empty flat spacetime. The general g{munu} is the geometry once T_{munu} has had its say.

That is the architecture this lecture has built. Before we close, there is the other voice in the room.

The Carroll handoff — the boundary discipline (≈ 4 minutes)

The skeptic-cast voice in this lecture belongs to Sean Carroll. PhD theoretical physics, Harvard, 1993. Currently the Homewood Professor of Natural Philosophy at Johns Hopkins, appointed in 2022. Author of Spacetime and Geometry: An Introduction to General Relativity, published by Addison-Wesley in 2004 — the textbook many physics graduate programs use as the first-year GR text, alongside MTW. Host of the Mindscape podcast since 2018.

Carroll's job in this lecture is to draw a line that special relativity, all by itself, does not draw. The line between what the geometry permits as a mathematical object and what the universe permits as a physical object. It is the same line Hossenfelder drew in Lecture One, in a different register.

We will hear that same discipline applied to one specific kind of overclaim — the move from "four-momentum is a four-vector" to "therefore you can engineer four-momenta arbitrarily." Carroll's published position on the analogous move in propellantless propulsion is on the record, in his 2015 Preposterous Universe post, "Warp Drives and Scientific Reasoning." What follows is the verbatim sentence-grouping from that post, reproduced word-for-word from the published source.

If you want to go forward, you have to push on something or propel something backwards. … What I want from this discussion is highly respected scientists, working under exquisitely controlled conditions, producing refereed publications.

Read that twice, slowly. If you want to go forward, you have to push on something or propel something backwards. That sentence is Newton's third law in one breath. Every motion of a body's center of mass requires that something else's center of mass move the opposite way. Conservation of momentum. The Lorentz-invariant version of that conservation law is exactly that the four-momentum of a closed system is conserved as a four-vector — total p^mu in equals total p^mu out.

What Carroll is pushing back on, in the warp-drive and propellantless-propulsion context — and what carries over into every Block-D claim this series will engage — is the move from "the math admits four-vectors that do interesting things" to "therefore we can engineer four-vectors that do interesting things, without paying the momentum-conservation cost." Four-vector bookkeeping is a language. It does not give you a license to write any four-vector you want and assume the universe will source it. The right-hand side of Einstein's field equations — the T_{munu} side — has to come from somewhere. If you want a particular geometry, you need a particular distribution of energy and momentum.

And if you want momentum to go forward on one side without anything going backward on the other, you are asking the universe to violate the very conservation law that makes the four-vector formalism mean anything in the first place.

That is the discipline. The mathematics admits four-vectors with any components you can write down on paper. The physics requires that those four-vectors come from real sources, conserved consistently, in configurations the universe has been observed to deliver. The second criterion is much, much stricter than the first.

Carroll's second sentence — what I want from this discussion is highly respected scientists, working under exquisitely controlled conditions, producing refereed publications — is the editorial test. It is the test this lecture series will hold every load-bearing claim against. The mathematical scaffolding we have built today is impeccable: special relativity is the most-tested theory in the history of physics, and four-vector bookkeeping is how every working particle physicist writes down a calculation. But "the formalism is impeccable" does not buy you "any particular engineering proposal is licit." The formalism is the language.

The licit proposals are the ones that pay the conservation and energy-condition costs in coin the universe accepts.

Hold that line. Special relativity tells you what the geometry of flat spacetime is. The energy-momentum tensor tells you what is sourcing the geometry. Newton's third law, in its Lorentz-invariant form, tells you that the source has to come from somewhere. None of this is overthrown by the next lecture, when we let the metric curve. The discipline carries.

What this lecture buys you (≈ 4 minutes)

What this lecture has bought you is the language. Four pieces of language, each of which is going to do work in a specific later lecture.

The Minkowski line element — ds^2 = -c^2 dt^2 + dx^2 + dy^2 + dz^2 — and the three kinds of interval it admits. Spacelike. Timelike. Lightlike. That trichotomy is going to do load-bearing work in Lecture 5, when we meet the Morris-Thorne traversable wormhole — Michael Morris and Kip Thorne, American Journal of Physics volume 56, page 395, 1988. The "flare-out" condition that Morris and Thorne wrote down for the wormhole throat is, at heart, a statement about which directions through the throat are spacelike and which are timelike, and what stress-energy you need at the throat to keep the spacelike directions from collapsing the geometry.

You can't read that condition without the vocabulary you just learned.

The four-velocity and four-momentum — u^mu and p^mu = (E/c, px, py, p_z). That bookkeeping is going to come back in Lecture 7, when we meet the quantum vacuum and the mode expansion of a quantum field. Every mode in that expansion carries a four-momentum. The vacuum is the state in which every mode is in its lowest-energy configuration, and the zero-point energy of each mode contributes to the total T^{00} of the vacuum. The Casimir effect, which we meet in Lecture 8, is what happens when you reshape the boundaries that select which modes are allowed — and the T^{ii} between two parallel plates is negative.

The energy-momentum tensor we just defined is the object that says so.

The energy-momentum tensor T{munu} itself. That tensor is the bridge from this lecture into Lecture 3, when Einstein's field equations show up and the left side, G{munu}, gets built out of the curvature of g{munu}. T{munu} is also the central object of Lecture 6, on energy conditions — the weak, null, strong, and dominant energy conditions are constraints on T{munu}, statements about which energy-momentum distributions the universe is willing to deliver.

And T{munu} is the central object of Lecture 10, on the Pais patent series — the engineering claim of those patents, read as physics, is that a particular kind of electromagnetic boundary condition can drive T_{munu} into a regime that sources a particular kind of geometry. The patents and the accompanying IEEE Transactions on Plasma Science paper exist in the public record. Whether the proposed mechanism does what the patents claim is the question Lecture 10 will engage, with the discipline Carroll just articulated.

Lorentz invariance — the statement that the Minkowski metric is the same in every inertial frame. That is the local symmetry of every patch of every spacetime in this series. In general relativity, the metric varies from point to point, but in a small enough neighborhood of any point it looks like Minkowski. The equivalence principle is the formal statement of that fact. Lecture 3 is where it earns its keep.

Newton's third law, in its Lorentz-invariant form — conservation of four-momentum for a closed system. That is the conservation law you cannot escape by going relativistic. Carroll's sentence about pushing on something carries through every subsequent lecture in this series. Whenever you see a propulsion or geometry-manipulation claim, the question to ask is: where is the four-momentum going? What four-vector is being pushed against to produce the four-vector being pushed forward? If the claim does not answer that question, the claim has not yet earned the formalism it is invoking.

Four pieces. The metric. The bookkeeping. The tensor that sources the metric. The conservation law that disciplines the tensor. The next lecture turns those four pieces over to general relativity and lets the metric curve.

Closing (≈ 90 seconds)

Go back to the platform. Same listener. Same two figures. One stationary, one having rolled past on the train. The platform observer's stopwatch read one number for the flash. The train observer's stopwatch read another. They are both correct.

The question you can now ask is sharper than the question you started with.

You can ask what their measurements agree on. The answer is the squared spacetime interval ds^2 = -c^2 dt^2 + dx^2 + dy^2 + dz^2 between the flash event and the next event of their choosing. You can ask what each of them is measuring, in terms of the underlying geometry. The answer is the projection of a four-dimensional reality onto a particular slicing — their own. You can ask what kind of separation lies between any two events they care about. Spacelike, timelike, or lightlike. You can ask what kind of momentum-conservation law their disagreement still respects. The Lorentz-invariant version of Newton's third law.

None of this was in their disagreement when the lecture started. All of it is in the language now.

Lecture 3 picks up where this one stops. General relativity. Christoffel symbols, geodesics, the Riemann tensor, the Einstein field equations. The metric stops being flat. The geometry starts responding to what's in it.

Same listener. Same platform. New geometry.