How this instrument works
Frequency counts how many complete cycles a repeating event fits into one second. Period measures how long a single cycle takes. Neither adds information the other lacks, which is why f = 1/T demands no derivation — it is arithmetic applied to a definition. A pendulum returning to its start every two seconds runs at 0.5 Hz. A guitar string recovering its shape 440 times each second sounds concert A. Halve one quantity and its partner doubles, exactly, with no constant of proportionality anywhere in sight.
Heinrich Hertz spent 1886 through 1889 at Karlsruhe generating and detecting electromagnetic waves with spark gaps and wire loops, confirming Maxwell's prediction while insisting his result had no practical use whatsoever. Four decades later IEC proposed naming a unit after him, and CGPM ratified hertz in 1960, retiring 'cycles per second' from formal usage. Timekeeping then inverted their old relationship: since 1967 a second has been defined as 9,192,631,770 periods of a particular caesium-133 hyperfine transition. Counting cycles now underwrites duration rather than borrowing from it.
Reciprocity assumes genuine repetition, and that assumption fails in three ordinary ways. Take a period off one noisy cycle and the result is an instantaneous estimate, not a stable figure, because real oscillators jitter — honest measurement averages many cycles. Short bursts fail harder: a tone lasting Δt seconds cannot carry a frequency sharper than roughly 1/Δt, a bandwidth limit no cleverer hardware escapes. And any waveform built from several components has no single period at all, only a spectrum, which is where Fourier analysis takes over from this one line of arithmetic.
- Type your measurement into Period (time for one cycle). Its unit menu accepts milliseconds, seconds, or minutes, so a 20 ms interval goes straight in as 20 ms.
- Frequency appears immediately underneath. Hertz is its default; that menu also offers kilohertz, megahertz, and revolutions per minute.
- Audit any result by multiplying: Period times Frequency has to come to 1 whenever both sit in SI units.
- For shaft or motor speeds, switch that output menu to rpm rather than dividing by 60 yourself.
- Keep period strictly above zero. A flat, non-repeating signal has infinite period and no frequency to report.
Worked example — a 20-millisecond mains cycle
Put an oscilloscope across a European wall socket and voltage repeats every 20 milliseconds: one full sine up, down, and back to where it began. Enter 0.02 into Period (time for one cycle) and Frequency returns 1 ⁄ 0.02 = 50 Hz, exactly. Nothing rounds, because 20 ms is one fiftieth of a second, so fifty of them fill it.
Fifty is a legal specification rather than an accident. Grid operators across Europe hold it near ±0.05 Hz during normal running, since generators spinning in synchrony must all agree. North America chose 60 Hz instead, giving a 16.67 ms cycle. Mains-driven wall clocks once kept time purely by counting these repetitions, so utilities historically tracked accumulated cycles overnight and ran marginally fast or slow to settle the difference.
Questions
Why write hertz instead of per second?
Both mean s⁻¹ dimensionally, but BIPM reserves hertz for periodic phenomena specifically — things that genuinely repeat. Radioactive decay is random rather than cyclic, so its rate takes becquerel. A production line stamping 5 parts per second gets neither; plain s⁻¹ covers ordinary counted rates. Reserving Hz for repetition keeps a reader from assuming a waveform exists where none does.
What is 3,000 rpm in hertz?
Fifty. Revolutions per minute tallies turns across sixty seconds rather than one, so divide by 60 going this way and multiply by 60 coming back. A shaft at 3,000 rpm completes one revolution every 20 ms, which is exactly mains cadence. Both unit menus on this sheet accept rpm directly, so that division only matters while auditing somebody else's figure.
Why do instruments measure a period and invert it?
Resolution. A counter that gates for one second and tallies edges carries an unavoidable ±1 count error, which at 2 Hz is a 50% uncertainty — useless. Timing one cycle against a 100 MHz reference clock instead resolves that same slow signal to parts per million within 500 ms. Reciprocal counting, as it is known, is why f = 1/T sits inside almost every low-frequency measurement rather than beside it.
How does error in my period carry into the answer?
Proportionally, one for one. Differentiating f = 1/T gives δf/f = −δT/T in magnitude, so a period known to 2% yields frequency known to 2%. The sign flips — overestimate duration and you underestimate rate — but no error is amplified or suppressed by the inversion. Averaging several cycles before entering a value shrinks both figures together.
How does frequency relate to wavelength?
Through wave speed: v = fλ. For a fixed medium, speed stays roughly constant, so wavelength and frequency trade off inversely. Middle A at 440 Hz travels through 20 °C air at 343 m/s, making its wavelength 0.78 m. Change the medium and frequency survives unchanged while wavelength stretches — sound entering water keeps its pitch but stretches more than fourfold, because speed there climbs to roughly 1,480 m/s.
Can frequency be zero or negative?
Zero corresponds to infinite period: steady DC voltage, a mass sitting still, nothing recurring. Since 1/T never reaches zero, this sheet approaches it without arriving. Negative values appear legitimately in two-sided Fourier spectra, where every real-valued tone is carried by a conjugate pair sitting at +f and −f, their imaginary parts cancelling on reconstruction. That is a bookkeeping convention for complex exponentials, not a cycle counted backwards, and it plays no part here.