On a traditional rotary telephone, the digit 9 took the longest to dial. It sat furthest counterclockwise on the dial, so the finger traveled the longest arc and the spring returned the dial through the most pulses. Rotaries transmitted digits as trains of electrical interruptions, and 9 required nine of them; 0, sitting beside 9, gave up only ten pulses but traveled almost the same distance.

The difference matters because dialing time split into two phases with different physics. Finger travel ran while the user pushed the wheel; pulse return ran while a governor-controlled spring spun the dial back. Both phases scaled with digit position, which is why high digits on the dial were slow digits on the line. This explainer unpacks the mechanism, the measured timings, and the counterintuitive corner cases that thin summaries miss.

The pulse mechanism inside the dial

A rotary phone dial is a mechanical encoder. Behind the finger wheel sits a cam and a set of spring contacts; as the dial returns, a shunt switch opens and closes the line loop once per unit of travel. Each open-close cycle is a pulse. The central office counts those pulses to identify the digit.

The finger stop sets the reference point. Dialing 0 means rotating the wheel fully against the stop; dialing 1 means rotating it barely at all. The return spring then unwinds the wheel back to the stop, and the pulse contacts fire along the way. Higher digits therefore produce longer travel and more pulses, and the two effects compound.

A centrifugal governor regulates return speed. Early designs used a friction governor; later ones used a brass cup with spring-loaded weights that fling outward as speed rises. The governor holds the return near a designed rate — roughly ten pulses per second on North American exchanges — regardless of digit, so pulse trains stayed within the central office's detection window.

The mechanism explains why 9 was slowest in human terms. It owns the longest finger arc and the longest pulse train among digits a caller dialed routinely. Zero traveled nearly as far but pulsed once more in a slightly different timing window.

Quantified timing at ten pulses per second

At the standard North American rate of about ten pulses per second, each pulse occupies roughly 100 milliseconds, split between a break interval and a make interval. A 9 therefore generated about 900 milliseconds of pulsing, plus the finger-travel time to reach it.

Finger travel depends on the dial's geometry. The finger stop sat roughly 130 to 140 degrees from the rest position on many Western Electric-style dials. Each digit hole sat about 36 degrees apart around that arc, so 9 sat near 144 degrees from rest and 0 sat just past it.

The return takes time proportional to pulse count at governed speed. A digit near 1 returns in roughly 100 to 200 milliseconds; a 9 returns in about 0.9 to 1 second of pulsing plus governor settling. Add finger travel of a few hundred milliseconds, and a single 9 could consume well over a second of a caller's time before the next digit began.

Area codes and exchanges amplified the effect. A number like 999-9999 stacked seven 9s, each costing nearly a second of pulse return. The cumulative penalty made high-digit numbers measurably slower to dial than numbers dominated by 1s and 2s.

The rate was a design constant, not a universal law. European systems pulsed at different rates near ten pulses per second with different break ratios, so exact timings varied by country and exchange. The ranking of digits by dialing time, however, held across rotary systems.

Timeline of a single digit

Phase What happens Approximate duration
Finger insertion Caller places finger in the hole Variable, user-dependent
Wind-up travel Wheel rotates clockwise to the finger stop 200–500 ms for high digits
Release and return start Spring engages, governor spins up 50–150 ms
Pulse train Contacts open and close once per digit unit About 100 ms per pulse
Governor settling Return completes at governed speed Included in return time
Next digit Caller moves to the next hole User-dependent

The table shows why the pulse train, not the finger push, dominated total time for high digits. At ten pulses per second, the return phase alone for 9 ran roughly nine-tenths of a second. The finger travel added a comparable fraction.

Edge cases that reshuffle the ranking

The simple answer — 9 is slowest — assumes a standard dial and a caller who dials from rest each time. Several edge cases complicate it.

Dialing 0 required the longest physical rotation, since the hole sat just past 9. In finger-travel time alone, 0 could edge out 9 by a hair. But 0 produced ten pulses against nine for 9, and the pulse count mattered more in total line time. The net result generally kept 9 and 0 within a fraction of a second, with 9 slower in most practical measurements.

Some exchanges used letters and digits together, as in early alphanumeric dialing. Letters mapped to digits, so a letter high in the alphabet could route through 9 and inherit its timing. The mapping, not the physical key, set the cost.

Party lines and shared lines changed perceived speed, not mechanical speed. Waiting for a free line added time unrelated to the digit, which could swamp the 9-versus-0 difference. The dialing physics stayed fixed while the user experience varied.

Rotary dials on private branch exchanges sometimes ran at different govern rates. A PBX tuned faster could compress pulse trains, shortening every digit including 9. The ranking persisted because every digit scaled together.

A counterintuitive detail: the digit 1 was fastest to dial but risked being misread. A single pulse sat close to the central office's minimum detectable break, so line noise or a slow dial could blur it. High digits were slow but robust; low digits were quick but fragile.

Misconceptions worth correcting

One common claim says 0 took longest because it sat at the end of the dial. That conflates finger travel with total dialing time. Zero did require the longest rotation, but 9 combined a nearly equal rotation with a heavy pulse train, and pulse time dominated the total.

Another claim says pulse speed varied by digit. It did not, in a healthy dial. The governor held return speed near constant, so every pulse took about the same 100 milliseconds regardless of which digit generated it.

A third misconception treats the dial as a digital keypad with equal timing. Rotary dialing was analog and mechanical, and timing scaled with position. Keypads with dual-tone multifrequency signaling replaced pulse trains entirely, which is why push-button dialing felt instant by comparison.

Some sources say rotary phones sent digits as clicks the caller could hear. Callers heard the governor's whir and the contacts, but the central office counted electrical breaks, not sound. The audible clicks were a side effect, not the signal.

Practical actions for collectors and restorers

If you own or restore a rotary phone, these steps keep dialing timing within spec.

Measure pulse rate with a line simulator or a phone tester; target about ten pulses per second on North American sets. Clean the pulse contacts with a burnishing tool, not abrasive paper, to avoid changing break ratios. Check the governor for sticky pivots or worn weights; sluggish governors slow every digit. Verify finger-stop alignment so high digits reach full rotation without overshoot. Test 9 and 1 first, since they bracket the timing extremes and expose faults fastest. Confirm the shunt switch opens and closes cleanly, because a stuck shunt drops pulses entirely.

Red flags that signal a timing fault:

A 9 that pulses fewer than nine times means a skipped contact or a shorted shunt. A 1 that registers as 2 or 0 indicates contact bounce or a slow governor. Audible speed changes within one return suggest governor wear. Digits that work on one exchange but fail on another point to rate mismatch, not dial damage.

Limits of the comparison and what cannot be concluded

Exact dialing times cannot be pinned to a single number across all phones. Finger-travel time depends on dial diameter, hole spacing, and the caller's hand. Governor rate varied by manufacturer, country, and exchange standard, so the ten-pulses-per-second figure is a North American reference, not a universal constant.

Central office detection windows also varied. Some offices tolerated wider break-ratio ranges than others, so a dial that felt slow might still register correctly while a faster dial dropped pulses. Without the specific exchange standard and dial model, only the ranking — 9 slowest, 1 fastest, 0 close behind 9 — can be stated with confidence.

Measurements from restored phones should be treated as condition-dependent. A worn governor or dirty contacts can shift timing by tens of milliseconds per pulse, which compounds across seven-digit numbers. The mechanism is stable; the precise milliseconds are not.