The Telegraph Relay

Wire a battery, a switch and a small electromagnet in series, then insert increasing lengths of thin wire — or a chain of resistors standing in for line resistance.

Measure the current with a multimeter at each length and note where the electromagnet stops pulling its armature.

Expect current to fall roughly in proportion to added resistance, and the mechanical action to fail abruptly well before the current reaches zero.

That gap matters: there is a wide band where a signal is clearly still present but can no longer do useful work. Everything that follows lives in that band.

Wind two electromagnets on identical iron cores: one with a few dozen turns of thicker wire, one with many hundreds of turns of the finest wire you have.

Find the smallest current that just moves the armature of each.

Expect the many-turns coil to respond to a far smaller current, because the pull follows ampere-turns — current multiplied by the number of turns — not current alone.

You have not amplified anything. You have built a detector sensitive enough to notice what survives the line, which is a different and cheaper thing.

Mount the sensitive coil so its armature closes a second, entirely separate circuit with its own battery and a loud sounder or lamp.

Now operate the far end of your long line and listen to the local circuit.

Expect a full-strength click from a signal that could not make a sound by itself.

Trace where the energy comes from and say it out loud: the line supplies the decision, the local battery supplies the power. Separating those two is the entire invention, and it is worth more than any component in it.

Deliberately degrade the incoming signal — long line, weak battery, a poor connection that makes it ragged — and compare the input and the relay's output on an oscilloscope, or by ear.

Expect the input to be rounded, drooping and noisy and the output to be square and full-amplitude.

Then push further until the relay starts to miss or chatter, and note where the cliff is.

This is the whole difference between a signal that survives a hundred stages and one that does not. Degradation that is thrown away at every stage cannot accumulate; degradation that is merely amplified always does.

Put two or three relays in series, each driving the next through a length of line, and send a message through the chain.

Expect the amplitude at the end to be as good as at the start, however many stages you add.

Now measure what does not recover: the delay. Time a single stage, then the whole chain.

Expect delay to add up stage by stage, and expect the crispness of the timing — how faithfully a short pulse stays short — to decay slowly even though amplitude does not.

Amplitude is restored; time is not. Every long digital link since has the same asymmetry.

Joseph Henry demonstrated the principle and did not patent it. He showed a small current working an electromagnet that closed a second circuit, and regarded it — explicitly — as a contribution to science rather than property. Samuel Morse and Alfred Vail built it into the working telegraph, and the resulting arguments about who invented what ran for decades. Henry's choice is why the idea spread as fast as it did.

It is what made the telegraph a NETWORK rather than a line. Without relays, range is set by wire resistance and battery voltage, and every route is a separate engineering problem. With them, distance stops being interesting — you add stages. The relay is the reason a message could cross a continent, and the reason the Morse Telegraph blueprint's sounder is worth building at all.

Then it became a way to think, not just a way to reach. A relay whose contacts feed other relays' coils computes: series contacts are AND, parallel contacts are OR, a normally-closed contact is NOT, and a relay wired to hold itself in remembers. Whole telephone exchanges and early computers were built from nothing else. The vacuum tube and then the transistor replaced the hardware and kept the model exactly.

The one idea to carry away. Regenerate, do not amplify. An amplifier magnifies the noise along with the signal, so errors compound down a chain; a device that makes a fresh clean decision at each stage throws the noise away. That single distinction is the reason digital signals can cross an ocean and analogue ones cannot — and it was discovered by people trying to make a click audible in the next county.

Honest limits. It is mechanical, so it is slow — milliseconds, not nanoseconds — and its contacts wear, pit and bounce. It needs a local power source at every station, which means a person to maintain every one. It restores amplitude but adds delay. And it makes a hard decision at a threshold, so a signal that arrives right at the threshold produces chatter rather than an honest failure, which is the most annoying fault in the system to find.