A picture that is secretly an equation
You arrive at this rung already holding the hard-won pieces. From the quantum rung you know particles are excitations of fields; from the QED rung you met the electron, the photon, and the idea that the electromagnetic force is carried by photon exchange; and just before this you faced the honest truth about virtual particles — that the exchanged messenger is a calculational device, never a thing you catch. Now we pick up the tool that turns all of that into something you can actually compute on the back of an envelope. A Feynman diagram is a small cartoon — a handful of lines meeting at dots — invented by Richard Feynman in the late 1940s. Its genius is that it is two things at once: a sketch of *what particles went in and what came out*, and, read by a fixed rulebook, a recipe for *the exact formula* that predicts how likely that process is.
Before Feynman, calculating even a simple collision in quantum field theory meant pages of dense, error-prone algebra. His pictures organized that chaos. Every diagram is built from just three kinds of ingredient, and once you can name them, you can read almost any diagram you will ever meet: lines that come in from outside and leave to the outside (the particles you actually start and end with), lines buried in the middle that connect things up, and the dots where lines meet. The rest of this guide is really just learning to see those three ingredients — and learning what they honestly do, and do not, represent.
The three ingredients: legs, vertices, propagators
Start at the edges of the diagram. The lines that stick out — beginning or ending in empty space rather than connecting to anything inside — are the external legs. These represent the real, observable particles: the two electrons you fired into a collision, say, and whatever flew out and hit a detector. External legs are the honest, measurable part of the picture. Every other line is internal. Together, legs, the dots, and the inner lines are catalogued under the single glossary heading vertices, propagators, and external legs — the grammar of every diagram.
Next, the dots where lines meet: these are the vertices, and they are the heart of the physics. A vertex is a single elementary interaction — the one place where particles actually couple to each other and the world changes. In QED the only allowed vertex is starkly simple: two electron lines and one photon line meeting at a point, the electron-photon vertex. That single junction *is* the statement "an electron can emit or absorb a photon," and remarkably, all of electromagnetism is built by wiring copies of that one vertex together. Crucially, each vertex carries a numerical strength — its coupling — which says how readily that interaction happens. In QED that strength is tied to the electric charge, and it is small, which is the secret to why the whole scheme is calculable at all.
Finally, the internal lines that run from one vertex to another, never reaching the outside world: these are the propagators. A propagator stands for a particle exchanged between two vertices — the messenger in the middle, the very thing you just learned to call a virtual particle. The internal photon trading between two electrons is a propagator. And here is the quiet payoff of the previous guide: a propagator can be *off-shell*, carrying an energy-and-momentum combination no real particle could ever have. The mathematics rewards distant-looking exchanges with a small number and close-fitting ones with a large number, so the propagator is precisely where the diagram encodes how strongly the messenger can bridge a given gap.
Reading one as a picture
Let us read the most famous diagram in physics: two electrons repelling each other. Draw time running from left to right (conventions vary — some draw it upward, so always check the axes). Two electron lines come in from the left, approach, and at one electron's line a vertex sits where it emits a photon; that photon line crosses the gap and is absorbed at a vertex on the other electron's line; then the two electrons leave to the right, now veering apart. The incoming and outgoing electrons are the four external legs; the two dots are the vertices; the photon stretched between them is the propagator. That single image *is* the leading explanation of why like charges repel.
e- ----\ /---- e- (external legs: real electrons in/out)
*=========* (two vertices, joined by the
e- ----/ photon \---- e- photon propagator in the middle)
time --->A few reading habits pay off immediately. Arrows on the fermion lines (the electrons) track the flow of particle versus antiparticle: an arrow running against the time direction is the standard way to draw a positron, the electron's antiparticle you met in the QED rung — antimatter as matter moving "backward" in the diagram's bookkeeping. Wavy or doubled lines usually mean a force carrier like the photon. And the *topology* — which leg connects to which, and whether the exchanged photon links the two beams across the middle or is radiated and reabsorbed along one line — distinguishes physically different processes. Rearranging where the exchange sits gives the s-, t-, and u-channels, a vocabulary you will use constantly when classifying scattering.
Reading the same picture as a calculation
Now the magic. The same diagram is a literal set of instructions for writing down a number. The Feynman rules assign a mathematical factor to each ingredient — a factor for each external leg, a coupling factor for each vertex, and a propagator factor for each internal line — and you simply multiply them together. The product is a contribution to the amplitude, technically the matrix element: a complex number whose squared magnitude gives the probability of the process. A picture you can sketch in three seconds thus translates, term by term, into the formula a computer (or a patient human) evaluates to predict a measurement.
But one diagram is rarely the whole story. Quantum mechanics tells us to add the amplitudes for *every* way a process can occur, and there are always more elaborate diagrams for the same in-and-out particles: ones with extra vertices, where an exchanged photon briefly becomes an electron-positron pair before reforming, or where an electron emits and reabsorbs its own photon. Each extra pair of vertices stitches in another factor of the small QED coupling, so each added complication contributes less. This is the engine of the whole method: a perturbation series, an ordered sum in which simpler diagrams dominate and fancier ones are small corrections. Because the coupling is tiny, the series can be truncated and still be astonishingly accurate.
This sorts every diagram into a hierarchy, the tree level versus loop diagrams distinction. A tree diagram has no closed internal loops — like our simple photon-exchange sketch — and gives the leading, roughly classical answer. A loop diagram has at least one closed cycle of internal lines, capturing quantum corrections. Each higher rung of loops is one more step in the perturbation order and one more small factor of the coupling. The electron's anomalous magnetic moment, that triumph predicted to twelve digits, is precisely a sum over more and more loop diagrams — each tiny picture pinning down one more decimal place of reality.
The honest caveat: not flight paths
Here is the misconception this whole field has to keep correcting, and the reason we built so carefully toward the virtual-particle idea first. It is desperately tempting to read the internal lines as literal trajectories — as if a tiny photon really does shoot across the gap on that exact diagonal, at that moment, along that path. It does not. An internal line is not a path through space; it is a *term in a sum*. The diagram lives in an abstract calculational space, not on a film of the collision. The wiggle's slope, its length, the angle of the dots — none of it is a measured route. What is real is only the external legs (the particles in and out) and the final number you compute; everything inside is bookkeeping.
So hold the two readings together, exactly as you learned to do with virtual particles. As a *picture*, a Feynman diagram is a wonderfully intuitive shorthand for which particles interacted with which — keep that, it is genuinely useful. As a *calculation*, it is one labelled term in a perturbation series, and that is where its predictive power lives. The dishonest move is to collapse the two and believe the cartoon is a snapshot of reality. The honest stance, the one that has carried this subject to its greatest triumphs, is to treat the diagram as an indispensable tool for getting the right number — and to let the *number*, checked against a detector, be the thing you actually believe.
Why this tool is about to matter
You now hold the key to the rest of this rung. Everything ahead — turning a process into a probability, predicting how often a collision yields a given outcome, comparing theory against the data a detector actually records — runs on the amplitude you read off a Feynman diagram. The next guides will take that complex amplitude, square it, fold in the available room for the outgoing particles, and arrive at a cross section: the single number an experiment can directly measure. The diagram is the bridge from "here is a theory" to "here is a prediction you can test against the world."