In April 1928, a Bell Labs engineer named Harry Nyquist published a paper — and filed a series of supporting patents at AT&T — that would quietly become the mathematical bedrock of every MP3, every Zoom call, every streaming song, and every digital camera shutter that has ever clicked. The paper was called "Certain Topics in Telegraph Transmission Theory," and its companion patent work on pulse transmission included US Patent 1,915,440 (filed 1928, granted 1933) covering methods for transmitting telegraph signals through band-limited channels.

Nyquist's question sounded mundane: how fast can you cram telegraph pulses through a wire of a given bandwidth before they smear into each other and become unreadable? His answer was startling. He proved that a channel of bandwidth B hertz can carry exactly 2B independent pulses per second — no more, no less. Push harder and the pulses interfere; push less and you waste capacity.

This number — 2B — is the Nyquist rate, and it is the secret handshake of the digital age.

What the patent actually described

• A mathematical relationship between bandwidth and maximum signaling speed
• The concept that a continuous waveform contains only a finite amount of information per second
• Pulse-shaping methods to pack symbols densely without intersymbol interference
• The realization that sampling a smooth signal at discrete instants could, in principle, capture it completely

That last point was the bombshell. In 1928, "signals" meant continuous analog wiggles — voices, music, telegraph clicks. Nyquist showed they could be reduced to a stream of numbers with no loss of information, provided you sampled fast enough. Twenty-one years later, Claude Shannon would formalize this into the Nyquist–Shannon sampling theorem: any signal band-limited to frequency f can be perfectly reconstructed from samples taken at rate 2f.

The modern echo

Look at any digital device and you are looking at Nyquist's 1928 theorem in action:

• CD audio at 44.1 kHz: chosen because human hearing tops out near 20 kHz; doubled to 40 kHz, plus margin for the anti-aliasing filter — pure Nyquist.
• Cellular networks: every modulation scheme from GSM to 5G NR computes its symbol rate against the Nyquist limit of its allocated channel.
• Software-defined radio: the entire field exists because Nyquist guarantees a fast enough ADC can replace any analog receiver.
• Medical imaging: MRI k-space sampling, CT reconstruction, and ultrasound all live or die by Nyquist criteria.
• Compressed sensing (2006, Candès & Donoho): a modern extension showing that for sparse signals you can beat Nyquist — but only by encoding the violation back into the math Nyquist gave us.

What makes the patent surprising is the era. In 1928, vacuum tubes were new, television was experimental, and the word "bit" wouldn't exist for another two decades. Nyquist was nominally solving a telegraph billing problem for AT&T — how to sell more characters per second on a leased line. Instead, he wrote down the speed limit of information itself.

Could it be built better now? No — and that's the point. Nyquist's bound is a law of nature, like the speed of light. Every engineer who designs an ADC, a modem, or a codec is still negotiating with a Swedish-American telephone engineer who died in 1976.