Guide Science 6 min read

A gear pair with a shared factor keeps meeting the same teeth

When two gears' tooth counts share a factor, each tooth only ever meets a few teeth on the other gear, and every exact whole-number ratio works this way unless the driver has a single tooth. One tooth more or fewer on the gear breaks the pattern at a small cost in ratio, and published advice differs on whether that is worth it.

Nadia Rahim
Data & Statistics Analyst
Published 15 Nov 2026, 9:26 AM (SGT)
Share:
Close-up of the small brass gears of an opened watch movement, their teeth meshing. Close-up of the small brass gears of an opened watch movement, their teeth meshing. Photo by Tima Miroshnichenko on Pexels
Advertisement

A gear ratio is the driven gear's tooth count divided by the driver's, and the same two counts decide which teeth ever touch. Twenty teeth driving 80 give an exact 4 to 1, and each pinion tooth meets only 4 of the 80, the same four on every turn. Put 79 or 81 teeth on the gear and every pinion tooth meets every gear tooth. The ratio moves by 1.25%, on our arithmetic. Whether a designer wants that depends on how the gears were finished, and published advice points both ways.

Shared factors pair each pinion tooth with the same few partners on every turn

Each driver tooth meets the driven count divided by the greatest common divisor of the two counts. For 20 driving 80 the divisor is 20, so a pinion tooth meets 4 of the 80. It is the number Euclid's algorithm finds.

Hermann J. Stadtfeld of Gleason describes the pattern in "The Hunting Tooth and its Effect on Break-In", Gear Technology, January/February 2021, pp. 54–57. With no common factor, he writes, "every tooth of the pinion will mesh with every slot of the gear." That is a hunting-tooth pair. In his 10 by 30 example, tooth 1 of the pinion meets slots 1, 11 and 21 only.

No exact whole-number ratio can hunt once the driver has more than one tooth. If the gear has R times the pinion's teeth, the common divisor is the whole pinion. A single-start worm is the exception. Its one thread meets every tooth of the wheel, so a worm driving 40 teeth is an exact 40 to 1 and still hunts.

One tooth more or fewer on the gear makes every tooth change partners

Take a pinion of N teeth and a whole ratio R. A gear of R × N + 1 or R × N − 1 teeth never shares a factor with N, because any divisor of both would also divide 1. The ratio error is 1 ÷ (R × N), and no inexact pair on that pinion comes closer. A 20-tooth pinion near 4 to 1 gives 1.25%. A 30-tooth pinion gives 0.83%, with 119 teeth on the gear, and 0.33% near 10 to 1.

That cost belongs to a single pair. In a two-stage train every mesh can hunt while the overall ratio stays exact: we found 21 driving 50, then 25 driving 42, and (50 ÷ 21) × (42 ÷ 25) is exactly 4.

Of the 4,116 pairs we counted, pinions of 17 to 40 teeth against gears up to 200, 2,509 (60.96%) shared no factor. That agrees with a known result of number theory: the coprime share of large random pairs tends to 6/π², about 60.79%.

The older literature wanted changing partners for lapped gears and fixed ones for sets neither heat-treated nor ground

Stadtfeld does not call hunting better in general. He reports that the older literature found a hunting ratio helpful for lapped gears but preferred integer ratios for gears not heat-treated or ground, because integer-ratio sets are easier to inspect, test and assemble accurately. The integer rule, he adds, is still applied for precision actuation at moderate to low load. His own examples are ground bevel and hypoid sets during break-in.

When a run-in set with an integer ratio comes apart, he advises marking a gear tooth and the pinion slot it rolls with, so the same contacts go back together.

Mesh interference depends on the partner and undercut on the cutter

For standard full-depth involute teeth with no profile shift, two separate limits determine the smallest possible pinion. The first is interference in the mesh itself. If the pinion is too small for its gear, the tips of the gear's teeth will dig into the pinion's flanks below its base circle, where the profile is no longer an involute. We computed each cell from the geometry and checked it against the closed form.

Advertisement
Pressure angleEqual gearRatio 2Ratio 4Ratio 10Rack
14.5°2327293132
20°1315161718
25°910111112

At 20°, a 13-tooth pinion clears gears only up to 16 teeth. A 17-tooth pinion reaches 1,309, and at 18 it clears a rack.

Undercut happens in manufacture and depends on the cutter. A pinion generated by a hob or other rack-type cutter loses part of its flank below 2 ÷ sin²20°, which computes to 17.1 teeth, whatever it later meshes with. KHK's Gear Technical Reference, §3.5, gives 17 as the minimum free of undercutting at 20°, and says smaller gears can still serve if strength and contact ratio allow. A hobbed 13-tooth pinion is therefore undercut even though it would clear an equal gear in mesh.

This calculator shows which teeth each stage pairs off and the smallest pinion that clears

The gear ratio calculator, as revised on its fix branch of 9 November 2026, answers both questions for each stage. Enter one stage of 20 driving 80 at 1750 rpm, 10 N·m and 95%: it returns 437.5 rpm and 38 N·m, and offers 79 or 81 as the nearest hunting pairs. Enter 13 driving 80 at 20° and it flags interference, then warns separately about hob undercut. At 20° a 10 to 1 pair needs a 17-tooth pinion and a 170-tooth gear, while two stages of 16 driving 51 give about 10.2 with every pinion clear.

Before that revision the page said a 60 Hz catalogue under-stated a 50 Hz drive's output speed, the reverse of the arithmetic. It also filled a cleared box with its default unannounced, accepted 20.5 teeth and cited the 2005 AGMA nomenclature standard. It now cites ANSI/AGMA 1012-H23, announced in Gear Technology in June 2023.

Supply frequency sets the motor's tempo before any gear does

An induction motor's synchronous speed is 120 × frequency ÷ poles. A 4-pole motor is 1800 rpm on 60 Hz and 1500 rpm on 50 Hz, so a 60 Hz catalogue speed overstates a 50 Hz drive of the same pole count by 6/5. On 50 Hz the output runs at about 5/6 of the listed figure.

The calculator's three-stage default chains 20 driving 80, 20 driving 60 and 20 driving 50, for 30 to 1. Fed 1750 rpm, a 60 Hz figure, it gives 58.3 rpm out; on the 5/6 basis a 50 Hz drive gives about 48.6. A loaded motor runs a little below synchronous; the nameplate gives its real speed.

Worms, ring gears and strain-wave units learn a different dance

A worm's effective tooth count is its number of starts, so a single-start worm always has a hunting ratio. A ring gear turns the same way as its pinion, and the table above covers external pairs only.

Harmonic Drive describes its strain-wave gearing as a rigid circular spline with two more internal teeth than the flexspline inside it. Each full turn of the wave generator moves the flexspline two teeth the opposite way, so the reduction is the flexspline's tooth count divided by two, on our arithmetic. The maker states that its gearheads have "Zero Backlash". There a difference of two teeth, rather than a ratio of two counts, sets the speed.

Advertisement
Nadia Rahim
Data & Statistics Analyst

Nadia Rahim covers statistics, data literacy, measurement, and how published numbers get misread for RECATOOLS.

View author profile → · Editorial policy

About this byline Nadia Rahim is a RECATOOLS editorial persona for statistics and data-literacy coverage. Articles are produced and reviewed under RECATOOLS editorial supervision.

Corrections policy

Advertisement