---
title: "Black Hole Ringdown and the No-Hair Test Interactive Lab | Sandbox Physics"
description: "Why does a ringdown fit near the peak need the overtone?"
canonical_url: "https://sandboxphysics.com/labs/black-hole-ringdown-spectroscopy/"
language: "en"
page_type: "lab"
alternate_languages:
  zh-CN: "https://sandboxphysics.com/zh/labs/black-hole-ringdown-spectroscopy/"
last_modified: "2026-11-17"
---

# Black Hole Ringdown and the No-Hair Test

**Why does a ringdown fit near the peak need the overtone?** A black hole left by a merger rings at a few set tones, and in general relativity their pitch and fade time depend only on its mass and spin. Two measured tones therefore test that rule, but close to the loudest moment a fast-fading overtone must be included, or the inferred mass and spin come out wrong. A single fit that drifts away from the prediction is not, by itself, new physics.

Drag the remnant mass, spin and fit start. Two damped tones with $f(M,\chi)$ and $\tau(M,\chi)$ pin down the black hole; dropping the overtone biases the result.

> Interactive simulation: open this page in a web browser to change the parameters and run the model.

Model assumptions

The record is synthetic, not detector data. It holds two Kerr modes with $\ell=m=2$: the fundamental $n=0$ and the first overtone $n=1$, both starting at the peak, with $A_1/A_0=4$ by default (close to the fitted ratio in Giesler et al. 2019, Table I) and white Gaussian noise scaled to the chosen signal-to-noise ratio over a $60\,M$ record. Mode frequencies come from a Leaver continued-fraction table computed for this site, which matches the published Berti tables to about $10^{-9}$. Each fit maximises the mode amplitudes and phases and puts flat priors on mass (0.6 to 1.6 times the injected value) and spin (0 to 0.99); the outlines are 90 % highest-density regions on that grid. They are not calibrated Bayesian posteriors: over 100 noise draws the 90 % region held the injected values 89 % of the time, the 50 % region only 40 %. Not modelled: inspiral and merger, nonlinear and other angular modes, counter-rotating modes, coloured detector noise and antenna response, precession, and uncertainty in the time of the peak. The scene is schematic: a flat-space delay without $1/r$ fall-off, the horizon drawn 2.5 times larger than the wave, compressed brightness and slowed time. A fitted shift of the overtone frequency on its own is not evidence against general relativity.

## Record, fits and what they infer

Synthetic record computed in this browser; no detector data.

From this start time the fundamental-only fit misses the injected remnant; adding the overtone recovers it.

Synthetic strain and the two fits

- Noisy samples
- Injected fundamental
- Injected overtone
- Fit, fundamental + overtone
- Fit, fundamental only

Axes and priors

Horizontal: time after the peak in ms. Vertical: whitened strain in units of the injected fundamental amplitude $A_0$. Grey samples before $t_0$ are left out of both fits.

Remnant mass and spin, 90 % regions $(M,\chi)$

- Fundamental + overtone
- Fundamental only
- Injected remnant

Axes and priors

Horizontal: $M/\mathrm{M}_\odot$ over the whole prior. Vertical: $\chi$. Flat priors; amplitudes and phases fitted.

Overtone frequency test $\delta f_{221}$

- Posterior, shaded 90 %
- General relativity

Axes and priors

Horizontal: $\delta f_{221}$ with $f_{221}=f_{221}^{\mathrm{Kerr}}(M,\chi)\,(1+\delta f_{221})$, flat prior on $[-0.5,\,0.5]$. Mass and spin are marginalised.

More readouts

- **Credible level of the injected point, fundamental + overtone**: $49.7\,\%$
- **Credible level of the injected point, fundamental only**: $100\,\%$
- **Injected fundamental frequency**: $f_{220}=249\,\mathrm{Hz}$
- **Injected fundamental damping time**: $\tau_{220}=4.16\,\mathrm{ms}$
- **Injected overtone frequency**: $f_{221}=244\,\mathrm{Hz}$
- **Injected overtone damping time**: $\tau_{221}=1.37\,\mathrm{ms}$
- **Overtone share of the amplitude at the fit start**: $80\,\%$
- **Samples in the fit window**: $83$

A credible level near 100 % means the injected values sit outside the inferred region. Over many noise draws the 90 % region should miss about one time in ten.

## Hearing a black hole’s mass and spin

**Background** After two black holes merge, the remnant settles down by radiating damped gravitational waves. In general relativity each tone, labelled by $(\ell,m,n)$, has a frequency and a damping time fixed by the remnant mass $M$ and spin $\chi$ alone. The overtone $n=1$ has almost the same pitch as the fundamental $n=0$ but dies about three times faster.

**Why it matters** Move the fit start and watch whether a model with one tone, or with two, can find the remnant that was put into the record.

- **One-sentence intuition**: Near the peak the overtone is louder than the fundamental. A one-tone model then has to bend $M$ and $\chi$ to imitate it; with both tones tied to the same $(M,\chi)$ the fit lands on the injected remnant.

### Core mathematical model

#### A sum of damped tones

$$
h(t)=\sum_{n}A_n\,e^{-(t-t_0)/\tau_n}\cos\!\left[2\pi f_n\,(t-t_0)-\phi_n\right],\qquad t\ge t_0
$$

The Lab fits this model from the start time $t_0$ with one tone ($n=0$) or two ($n=0,1$). Amplitudes and phases are free; frequencies and damping times are not.

#### Mass and spin set every tone

$$
2\pi f_{\ell mn}-\frac{\mathrm{i}}{\tau_{\ell mn}}=\frac{c^3}{GM}\,\hat\omega_{\ell mn}(\chi)
$$

The dimensionless number $\hat\omega_{\ell mn}(\chi)$ comes from black-hole perturbation theory. The Lab reads it from a table computed with Leaver’s continued-fraction method and checked against the published Berti tables. Doubling $M$ halves every frequency.

#### Fitting with the amplitudes solved for

$$
\ln\mathcal{L}(M,\chi)=\frac{1}{2\sigma^2}\,\mathbf{d}^{\mathsf T}X\left(X^{\mathsf T}X\right)^{-1}X^{\mathsf T}\mathbf{d}
$$

For each trial $(M,\chi)$ the columns of $X$ are the damped cosines and sines of the tones. Least squares fixes the amplitudes; $\sigma$ is the white-noise level. The outlines enclose 90 % of the resulting weight on a flat grid.

#### Letting the overtone drift

$$
f_{221}=f_{221}^{\mathrm{Kerr}}(M,\chi)\,\left(1+\delta f_{221}\right)
$$

Isi et al. (2019) used this form on GW150914. A value of $\delta f_{221}$ far from zero, repeated across events and robust to the start time and noise model, would be a problem for general relativity; one wide or offset curve is not.

### Common difficulties

#### Waiting longer is not free

**Typical misconception** Starting the fit later always gives a cleaner answer, because the overtone has gone.

**Better mental model** Later start times avoid the overtone but throw away the loudest part of the signal, so the regions grow. With the default record the fundamental-only region already contains the injected values a couple of M after the peak, mostly because it has grown, not because the bias has gone. By 10 M it is wider than the two-tone region from the peak.

#### An offset is not a discovery

**Typical misconception** If the fitted overtone frequency is not exactly the Kerr value, general relativity has failed.

**Better mental model** Noise alone moves the curve: try other noise draws. A real test also has to rule out a wrong start time, missing modes, nonlinear effects and detector noise that is not white. Published GW150914 analyses disagree on how strongly the overtone is detected at all.

### Run the experiment

1. 01

   #### See the bias

   Start from the default and compare the amber dashed outline with the white cross in the mass-spin chart.

   **What to observe:** The fundamental-only fit from the peak points to a lighter, slower-spinning remnant; the two-tone fit around it contains the injected $(M,\chi)$.

2. 02

   #### Move the fit start

   Drag along the strain chart, or choose “Wait for the fundamental”.

   **What to observe:** By $t_0\approx 10\,M$ the overtone is already weaker than the fundamental and the amber outline contains the cross, but both regions are much larger than the two-tone region from the peak.

3. 03

   #### Turn up the signal

   Choose “A loud event” and look at the overtone frequency test.

   **What to observe:** At $\rho=80$ the curve for $\delta f_{221}$ narrows and still contains zero, while the fundamental-only outline shrinks to a small region well away from the cross.

4. 04

   #### Change the remnant

   Drag the ergoregion edge and the spin dial in the scene.

   **What to observe:** A heavier remnant rings at a lower pitch with longer waves; a faster spin raises the pitch and makes the fundamental ring longer.

Physical conclusion

Two tones from one remnant give two measurements of the same mass and spin, which is what makes a no-hair test possible. Near the peak the overtone must be in the model, and a single offset fit, on synthetic or real data, needs noise, start-time and model checks before it means anything.

## Related experiments and articles

Black hole ringdown and the no-hair test · Reading path

### From a ringing black hole to a no-hair test

Articles on this path that use this experiment, followed by the other experiments they suggest.

1. [Article **Why does a newly merged black hole ring like a bell?**](https://sandboxphysics.com/stories/black-hole-ringdown/why-a-black-hole-rings/)
2. [Article **Damped tones and the Kerr spectrum**](https://sandboxphysics.com/stories/black-hole-ringdown/damped-tones-and-the-kerr-spectrum/)
3. [Article **Why a ringdown fit needs the overtone near the peak**](https://sandboxphysics.com/stories/black-hole-ringdown/fitting-a-ringdown-from-the-peak/)
4. [Article **From two tones to a no-hair test**](https://sandboxphysics.com/stories/black-hole-ringdown/from-two-tones-to-a-no-hair-test/)
5. [Article **Recomputing ringdown tones, a synthetic signal and a slice of GW150914**](https://sandboxphysics.com/stories/black-hole-ringdown/recomputing-ringdown-tones-and-a-public-event/)
6. [Article **How to read the 2019 GW150914 overtone paper and the debate that followed**](https://sandboxphysics.com/stories/black-hole-ringdown/reading-the-gw150914-overtone-paper/)

Waves & optics · Starter path

## From one ray to interference measurement

Start with refraction, find where the ray picture stops working, then use interference to measure tiny distances.

[Start here → **Optical Boundary Explorer** Turn the incident angle and watch reflected and transmitted power trade off. Then switch to the critical-angle scene. Observe · 11 min](https://sandboxphysics.com/labs/optical-boundary-explorer/)

1. [1 Optical Boundary Explorer](https://sandboxphysics.com/labs/optical-boundary-explorer/)
2. [2 Rainbow Caustic Observatory](https://sandboxphysics.com/labs/rainbow-laboratory/)
3. [3 Camera Obscura](https://sandboxphysics.com/labs/camera-obscura/)
4. [4 Coherent Wave Composer](https://sandboxphysics.com/labs/coherent-wave-composer/)
5. [5 Young Double-Slit Bench](https://sandboxphysics.com/labs/young-double-slit-bench/)
6. [6 Michelson Precision Lab](https://sandboxphysics.com/labs/michelson-precision-lab/)

[View the full path →](https://sandboxphysics.com/topics/waves-and-optics/#starter-path)
