Papers I and II moved a signal you supplied — through time was frozen and the world stayed outside. Now the capacitor turns to face both: it stores a value across time (memory), and it lets the world write straight into its field (sensing). And doing so reveals the half of the law the whole series quietly held still.
A confession. Every instrument so far ran on i = C·dv/dt — but that's only true when the capacitance C is constant. The honest, complete current through any capacitor is the rate of change of its charge, Q = C·V, and by the product rule that's two terms:
i = C·dv/dt + V·dC/dt
The first term is the whole story when the plates sit still and only the voltage moves — Papers I and II. The second term is what happens when the capacitance changes: move a plate, slide a finger near it, let a sound wave press on it. That's not a footnote — it's an entire mode of operation, and it's how a capacitor becomes an ear, a skin, a sensor. The world doesn't change the voltage you apply; it changes the geometry, and dC/dt carries the message in.
So this paper is the cap pointed two new directions: inward across time (hold a charge, fight the leak), and outward at the world (let dC/dt speak). Both were always in the law; the series just hadn't needed them yet.
The first turn is almost a betrayal of Paper I. There, the capacitor's gift was that it couldn't hold still — it reported only change. Here we use it to do the opposite: catch a moving voltage at one instant and freeze it. Close a switch, the cap charges to the signal's value right now; open the switch, and that value is trapped — held still for the rest of the circuit to read at its leisure. The change-detector becomes a state-keeper, on command.
This is the bridge from "the channel" to "memory." A held value is a memory one tick long. Make it hold longer and you've built a store — which is the next instrument, and the place the leak stops being a nuisance and becomes the whole drama.
The most numerous capacitor on Earth is the storage cell in dynamic memory — billions of them in the device you're reading this on. Each one is a single tiny capacitor: charged = 1, empty = 0. Store a bit by dumping charge in; read it by checking if charge is there. A whole memory is just a vast field of these wells.
But the ideal capacitor of Papers I and II was a polite fiction. Real capacitors leak. The stored charge drains away through imperfect insulation, and within milliseconds a 1 decays toward the threshold where it can't be told from 0. So dynamic memory must refresh — race around reading every cell and rewriting it, thousands of times a second, forever, just to stand still. The "dynamic" in DRAM is the leak.
The ideal cap would hold the bit forever. The real cap forgets in milliseconds — so memory is not storage, it's a refusal to stop rewriting. The leak we banished is the reason the machine has to keep paying attention.
V = V₀·e^(−t/τ)); real cells vary cell-to-cell, which is why memory has margins, error correction, and refresh to spare. This is also why DRAM loses everything on power-off, and why the leak is a genuine energy cost — billions of cells, refreshed forever.
Now the second term, made physical. Until now you moved the voltage. Here the world moves the capacitance — and the cap turns that into a signal with no voltage source changing at all. A sound wave presses on one plate of a charged capacitor; the plate moves a few nanometres; the gap changes; C changes; and because the charge is held fixed, the voltage swings: V = Q/C. That's a condenser microphone — the cap is literally an ear. A finger near a plate adds its own capacitance; the change is sensed as a touch. The cap is a skin.
You never touched the voltage. The world changed the capacitance, and the cap turned the world into a signal. That is the second term of the law — and it is how a passive part becomes a sense.
the world changes C, and V·dC/dt carries it in — is exact; the front-end electronics around it are the engineering this paper doesn't draw.
Across three papers the same passive component spoke in three directions — and it never stopped being a change-detector. It just kept finding new things whose change it could report.
i = C·dv/dt. the seed.V·dC/dt term.The full law, i = C·dv/dt + V·dC/dt, was always the whole capacitor. The first term moves the signals we make; the second lets the world speak. A capacitor was never a bucket for charge — it was an instrument for noticing that something changed, wherever the change came from: the wire, the gap, the clock, or the air against a plate.
A capacitor cannot hear what holds still — so it spent three papers finding every kind of thing that moves, and reporting each one. That is the whole of what it is to communicate.
If the series goes on: the cap as power (decoupling, the reservoir that answers a sudden demand — change in current this time), or the deep non-idealities (dielectric absorption, the "memory effect" where a cap half-remembers its last charge — a ghost of state in a change-detector), or the quantum end (the transmon's shunt capacitor, where this same part sets a qubit's frequency — which, it turns out, you already have a whole sphere on).