Gyrochronology

The idea: spin as a clock

Most ways of measuring a star's age are indirect and imprecise. You can fit its position on the Hertzsprung–Russell diagram against theoretical isochrones — but for a low-mass dwarf the main sequence is nearly vertical, so a star barely moves over billions of years and the age is uncertain by a factor of two or more. You can measure lithium depletion, chromospheric activity, or galactic kinematics, each with its own degeneracies. Gyrochronology offers something cleaner: cool main-sequence stars are observed to spin down in a smooth, reproducible way, so the rotation period itself becomes a clock.

The intuition is the same as a figure skater run in reverse. A young sun-like star arrives on the main sequence spinning quickly, with a period of hours to a few days, fast enough to be visibly flattened. Then it begins to brake. A magnetised wind carries angular momentum off into space, the spin slows, and the period lengthens — past ten days, twenty days, a month — in a way that depends almost entirely on two things: the star's mass (read off as a colour) and its age. Pin down the mass and measure the period, and the age falls out. The Sun, turning once every 24.5 days at 4.567 billion years old, is the single most precisely anchored point on the curve.

Why cool stars brake: the magnetised wind

The engine behind gyrochronology is magnetic braking, worked out in its essentials by Evry Schatzman (1962) and quantified by Mestel (1968). A star cool enough to have an outer convective envelope runs a magnetic dynamo, threading its surface with a global magnetic field. That field channels the ionised stellar wind: as gas flows outward, the magnetic field lines force it to keep corotating with the star — not out to the surface, but all the way to the Alfvén radius r_A, the distance at which the wind's kinetic energy density finally overpowers the magnetic energy density. For the Sun r_A is roughly 10–30 solar radii.

That long lever arm is the whole trick. Angular momentum is the product of mass, velocity, and lever arm, so gas that is forced to corotate out to r_A leaves carrying the angular momentum it would have at r_A, vastly more than the modest amount it had at the photosphere. A trickle of mass loss therefore drains spin angular momentum with brutal efficiency. The torque per unit mass lost scales as the square of the Alfvén radius:

dJ/dt = (2/3) (dM/dt) Ω r_A²        (magnetised-wind torque)

Crucially, this only works for stars with a convective envelope and a dynamo. Stars hotter than spectral type roughly F5 — above the Kraft break, named for Robert Kraft's 1967 discovery that rotation velocities drop sharply at this temperature — have radiative envelopes, no strong dynamo, weak winds, and never spin down. They remain fast rotators their whole lives, and gyrochronology simply does not apply to them.

Skumanich's law and the t⁻¹ᐟ² spin-down

The empirical cornerstone came in 1972, when Andrew Skumanich plotted mean rotation velocities and chromospheric calcium H&K emission for sun-like stars across clusters of increasing age. Both fell off as the inverse square root of age:

v_rot ∝ t^(-1/2)      →      P_rot ∝ t^(+1/2)

You can recover this from the braking torque. If the field strength scales with rotation (a dynamo runs faster when the star spins faster) and the mass-loss rate is roughly steady, the torque takes the form dΩ/dt = −k Ω³ for a constant k. For a star whose moment of inertia I is nearly fixed on the main sequence, integrating that differential equation gives

dΩ/dt = -k Ω³     ⟹     Ω(t) = Ω₀ / √(1 + 2 k Ω₀² t)
                            →   Ω ∝ t^(-1/2)   for large t

So after the star "forgets" its initial spin Ω₀ — which happens within a few hundred million years — its rotation rate depends only on age, not on how fast it started. This convergence is what makes the clock reliable: by an age of a few hundred Myr, fast and slow starters have merged onto a single tight relation, so you no longer need to know the star's birth spin.

The Barnes colour–period relation

Skumanich's law captures the age dependence but ignores mass. A red K dwarf and a yellow G dwarf of the same age spin at different rates because their convection zones, dynamos and winds differ. Sydney Barnes (2003, refined 2007) separated the two effects by writing the period as a product of a colour function and an age function:

P(B−V, t) = f(B−V) · g(t)

with   g(t) = t^n      (n ≈ 0.5, the Skumanich exponent)
       f(B−V) = a [(B−V) − c]^b     (a colour term, redder = longer P)

In Barnes's 2007 calibration the constants are roughly a ≈ 0.77, b ≈ 0.60, c ≈ 0.40, with the age entering as t^0.52 (t in Myr, P in days). The colour term means that at a fixed age the slow rotators trace a single smooth curve in the period–colour plane — a "gyrochrone." Stars that obey it are said to lie on the interface (I) sequence; the young fast rotators that have not yet converged form a separate convective (C) sequence, and stars are observed to hop from C to I over their first few hundred Myr. Only stars on the I sequence can be dated.

Later calibrations (Mamajek & Hillenbrand 2008; Angus et al. 2015, 2019 using Kepler; Spada & Lanzafame 2020 with a two-zone angular-momentum model) refine the exact functional form, but every one is the same idea: a measured period and a colour, run through a calibrated surface, return an age.

Calibrating against open clusters

Because the braking torque cannot be computed from first principles, gyrochronology is anchored empirically to star clusters, where every member shares a single age set independently by main-sequence-turnoff fitting. Each cluster supplies a snapshot of the period–colour relation frozen at one age; stacking them reconstructs how the gyrochrone evolves.

Cluster	Age	Distance	Sun-like P_rot	Role
ONC / NGC 2264	~1–3 Myr	~400 / 720 pc	1–10 d (disk-locked)	Initial conditions
Pleiades (M45)	~125 Myr	136 pc	~3–10 d	Young, near convergence
Hyades	~625 Myr	47 pc	~8–12 d	Intermediate anchor
Praesepe (M44)	~625 Myr	186 pc	~8–12 d	Confirms Hyades gyrochrone
NGC 6811	~1 Gyr	1100 pc	~11–14 d	Kepler field, extends curve
NGC 6819	~2.5 Gyr	2400 pc	~18 d	Old, tests stalling
M67	~4 Gyr	850 pc	~23–26 d	Solar-age anchor
Sun	4.567 Gyr	—	24.5 d	Gold-standard fixed point

The young clusters are messy: at the Orion Nebula Cluster age, accretion disks magnetically lock stars to a fixed period regardless of contraction, scattering rotation across more than a decade in period. By the Pleiades, disks are gone and the fast rotators are converging; by the Hyades and Praesepe the I-sequence is tight and clean. NGC 6811 and NGC 6819, both in the Kepler field, pushed the calibration past 1 Gyr and were the first clusters precise enough to reveal that the spin-down stalls in old, sun-like stars.

Worked example: dating a K dwarf

Suppose Kepler photometry of a single K-type dwarf with colour B−V = 0.90 shows a clear 17-day rotational modulation from starspots crossing the disk. What is its age? Using the Barnes 2007 form P = a[(B−V) − c]^b · t^n with a = 0.77, b = 0.60, c = 0.40, n = 0.52 (P in days, t in Myr):

P = 0.77 · (0.90 − 0.40)^0.60 · t^0.52
17 = 0.77 · (0.50)^0.60 · t^0.52

(0.50)^0.60 = 0.660
17 = 0.77 · 0.660 · t^0.52 = 0.508 · t^0.52
t^0.52 = 17 / 0.508 = 33.5
t = 33.5^(1/0.52) = 33.5^1.923 ≈ 860 Myr

So this star is roughly 0.9 Gyr old, with a formal uncertainty around ±100–150 Myr once colour error, the intrinsic scatter of the relation, and the calibration spread are folded in. Run the same arithmetic for the Sun (B−V = 0.65, and the ~26-day rotation period at sunspot latitudes that Barnes calibrated to) and the relation returns about 4.5 Gyr — the consistency check the method must pass. Note how steeply age depends on period: because t ∝ P^(1/n) ≈ P^1.9, a star spinning at 24 days is roughly four times older than one spinning at 12 days, even though its period is only twice as long.

How the period is actually measured

Gyrochronology only became a precision tool when space photometry made rotation periods easy to harvest in bulk. Three observational routes feed it:

• Starspot modulation. Dark magnetic starspots rotate in and out of view, dimming the star by 0.1–3 percent on the rotation period. Kepler, K2, and TESS have measured periods this way for tens of thousands of stars; McQuillan et al. (2014) catalogued rotation periods for over 34,000 Kepler dwarfs. This is the dominant source today.
• Chromospheric activity cycling. The Mount Wilson HK survey tracked Ca II H&K emission, which is modulated by active regions and traces both rotation and long-term activity cycles. Activity-based ages (gyrochronology's chromospheric cousin) calibrate against the same clusters.
• Spectroscopic v sin i. Rotational broadening of spectral lines gives the projected equatorial velocity v sin i. It is geometry-limited (you only see the line-of-sight component) but works where photometric modulation is absent.

The cleanest signal is the photometric one, which is why the relation exploded after 2009. A caveat: starspots also produce differential rotation signatures, so the measured period can drift by a few percent depending on the latitude of the dominant spot group — a small but real noise floor on the clock.

Gyrochronology versus other age methods

No single technique dates every star. Gyrochronology fills a specific niche — single, cool, slowly rotating dwarfs — where the classical methods are weakest.

Method	Best for	Typical accuracy	Main limitation
Gyrochronology	Cool MS dwarfs, 0.2–4.5 Gyr	~10–20 %	Fails above Kraft break; stalls past solar age
Isochrone fitting (HR diagram)	Subgiants, turnoff stars	Factor ~2 for dwarfs	Main sequence near-degenerate in age for low mass
Asteroseismology	Bright stars with detectable oscillations	~10 %	Needs high-cadence photometry; sparse for faint K/M
Lithium depletion	Pre-MS and young stars	Good < few hundred Myr	Lithium exhausted in older stars
Chromospheric activity	Cool active dwarfs	~factor 1.5–2	Activity-cycle scatter; same Kraft limit
Galactic kinematics	Statistical populations	Statistical only	No individual-star age
Radiometric (meteorites)	The Sun only	< 0.1 %	Solar System only

The current best practice is to combine them: asteroseismology and gyrochronology agree well for sun-like stars younger than the Sun and together pinned down the weakened-braking effect, while isochrones take over for evolved subgiants where rotation has lost its diagnostic power.

The twist: weakened magnetic braking

For three decades it was assumed the Skumanich clock kept ticking smoothly to the end of the main sequence. Then Kepler measured precise ages (from asteroseismology) and rotation periods for the same old, sun-like stars — and they disagreed. Stars older than the Sun were rotating faster than the extrapolated relation predicted. Jennifer van Saders and collaborators (2016) proposed weakened magnetic braking: once a star's Rossby number — the rotation period divided by the convective turnover time — climbs to roughly the solar value, the large-scale magnetic field that the wind grips onto becomes too tangled and small-scale to brake efficiently. Spin-down nearly stalls.

If correct, this caps useful gyrochronology near solar age, about 4–5 Gyr, for G dwarfs. The Sun itself may sit right at the threshold, which is partly why its activity cycle and rotation look the way they do. The effect is still debated — some argue it is a selection or detection bias in faint, slowly modulating old stars — but it is the most important open question in the field, and it is why a gyrochronological age for a 6-Gyr star should be treated with suspicion.

Common misconceptions and edge cases

• It works on every star. No. Only single, cool dwarfs with a convective envelope below the Kraft break. Hot A and early-F stars never brake and stay fast forever; gyrochronology is simply silent about their ages.
• Young stars can be dated this way. Below ~100–200 Myr, fast rotators have not yet converged onto the I-sequence — the period still depends on the unknown birth spin and on disk-locking history. The clock has not "started" reading reliably until convergence.
• Binaries behave like single stars. Tidally interacting close binaries are spun up or synchronised by the companion, overwriting the wind-braking signal entirely. A short period in a tight binary means tides, not youth, and yields a nonsense age.
• A faster spin always means a younger star. Usually true below the Kraft break — but a blue straggler, a merger remnant, or a tidally spun-up star can spin fast while being old. Gyrochronology assumes the star has had an undisturbed angular-momentum history.
• The measured period is the rotation period. Differential rotation means spots at different latitudes report slightly different periods; the surveyed period is a spot-weighted average that can scatter by a few percent. It is a small effect but sets a floor on the precision.
• Period scales linearly with age. It scales as the square root, so age scales as roughly the period squared. Small period errors translate into amplified age errors for fast rotators and compressed ones for slow rotators.