A pair too wide to hear

Why can't anyone hear this pair?

Picture two balls rolling round each other on a trampoline. Each dents the sheet, the dents move as the balls move, and the sheet ripples outward. Space does the same around any two masses in orbit. A black hole is one such dent, and so is a star. Albert Einstein predicted these gravitational waves in 1916, and in 2015 the LIGO detectors caught one for the first time, from two black holes in their last fraction of a second before merging.

Vesper and Ember are a black hole of 12 Suns and an orange dwarf star of 0.70, 1.5 times the Earth-Sun distance apart. They go round each other once every 188 days. After half an orbit the pair looks the same again, so space is shaken twice per lap: one wave crest every 94 days, and a wavelength of 0.26 light-years.

The waves are real, and they carry real power: 2.6 × 10¹⁵ watts, about 128 times all the power humanity used in the 2020s. None of it can be caught. A gravitational wave passes through matter almost untouched, and by the time this one reaches the Sun it stretches and squeezes lengths by 1 × 10⁻²¹ of themselves, each squeeze lasting weeks. No detector of the 2020s, on the ground, in space or made of pulsars, is sensitive at that pitch and that strength.

The waves also take energy from the orbit, so the pair sinks together, but very slowly. The gap closes by 0.0037 millimetres a year. Ember will shine for about 30 billion years, and over that whole life the orbit shrinks by about 111 km out of 2.2 × 10¹¹ m. Left alone, the two would meet after 1.5 × 10¹⁶ years, about 1 × 10⁶ times the present age of the universe. In practice they never meet.

Guess first Swap Ember for a second black hole as heavy as Vesper, at the same distance. Would that pair merge within the age of the universe?

No. Two Vespers 1.5 AU apart would take 5 × 10¹⁴ years. Mass helps, but distance matters far more: the merger time grows as the fourth power of the separation. To merge within the universe's age, Vesper and Ember would have to sit 0.046 AU apart.

So the pair is silent, and it is dark too. Ember sits far from Vesper, so no gas streams off the star into the hole and nothing glows in X-rays. What gives the hole away is the star. Ember circles the pair's balance point at 82 km/s, and seen from the Sun it sways back and forth across the sky every 188 days, around a companion nobody can see. That is how Gaia BH1, the nearest black hole known when it was found in 2022, came to light: a Sun-like star wobbling around nothing. Vesper is a closer and larger version of the same signal, close enough that a wobble that size would already be in the catalogues of the 2020s. No such system is known in 2026, so the Vigil's home is invented fiction.

How faint, how slow, and how big is the wobble?

For two masses m₁ and m₂ on a circle a distance a apart, with M = m₁ + m₂, general relativity gives the time until they touch, if nothing else acts:

time to merge = (5 / 256) × c⁵ × a⁴ / (G³ × m₁ × m₂ × M)

The c⁵ on top makes the time enormous unless the bodies move near light speed, and the a⁴ makes the separation decisive: halve it and the time falls sixteenfold. The waves come out at twice the orbital frequency.

Vesper and Ember in 2540	Value
Orbital period	188 days
Wave frequency	0.12 millionths of a hertz
Wavelength	0.26 light-years
Power in the waves	2.6 × 10¹⁵ W, 4 × 10⁻¹¹ of Ember's light
Strain at the Sun, 28 light-years away	1.2 × 10⁻²¹
Orbit shrinks by	0.0037 mm a year
Time to merge	1.5 × 10¹⁶ years

Strain is the fraction by which a passing wave stretches lengths. The pair's strain at the Sun is close to the 1 × 10⁻²¹ of GW150914 at its loudest. Loudness is not what stops a detection; frequency is. Each kind of detector hears one band:

Detector	Band	Vesper and Ember
Ground interferometers (LIGO, Virgo, KAGRA)	from 10 Hz up	8 × 10⁷ times too low
A space interferometer (LISA)	1 × 10⁻⁴ to 0.1 Hz	813 times below the bottom of the band
Pulsar timing arrays	nanohertz, a few billionths of a hertz	3.9 times their reference frequency, with 5 × 10⁻⁷ of the strain they measured

The pair falls in the gap between the last two, where no detector was being built in the 2020s.

Where the pair's wave sits among known sources, as strain against frequency on log axes. The bars below show where each kind of detector listens; the ground band has no upper edge drawn.

With JavaScript on, this is a calculator: pick a pair (Vesper and Ember, Gaia BH1, two Vespers, or GW150914 as LIGO first heard it) and slide the separation to see the time to merge, the orbital period, the wave frequency and the power in the waves. Without it, the key points are: 1.5 × 10¹⁶ years for Vesper and Ember, 5 × 10¹⁴ years for two Vespers, and 0.046 AU as the separation that would merge within the universe's age.

How does the star give the hole away?

Both bodies circle their shared centre of mass. Vesper holds most of the mass, so it barely moves, while Ember swings round a circle 1.4 AU in radius at 82 km/s. From the Sun that circle spans 165 milliarcseconds each side of centre, and the star's spectrum shifts back and forth with a speed of up to 82 km/s toward and away from the Sun.

Gaia BH1 was found from a sway of about 2.7 milliarcseconds, the size of its star's orbit seen from 1,566 light-years. Ember's is 62 times larger, because Ember is 28 light-years away. Moved to Gaia BH1's distance, Ember would sway 3.0 milliarcseconds: the two systems are close cousins.

And why no X-rays? A star in a binary overflows onto its companion only when it fills its Roche lobe, the region inside which its own gravity wins. Ember's radius is 1.2 % of its lobe's. It would start pouring gas onto Vesper only if the pair were closer than 0.018 AU. A pair close enough to merge within Ember's own life (0.056 AU apart) would pass that line on the way in, and become a bright X-ray source first. Feeding a black hole covers what happens when a hole is fed.

Open everything

Where do the formulas come from?The quadrupole formula, the Peters decay, the strain and the wobble

Why waves need a quadrupole

A gravitational wave has to be sourced by something that changes in a way that isn't spherical, and two conservation laws rule out the simplest options.

Mass doesn't change, so there is no monopole wave.
The centre of mass moves in a straight line, so the dipole moment changes uniformly, and there is no dipole wave.

The first thing left is the mass quadrupole, which measures how mass is spread out, like a dumbbell. The quadrupole formula (Einstein 1918) gives the power radiated:

$P = \frac{G}{5c^5}\,\big\langle \dddot{Q}_{ij}\,\dddot{Q}_{ij}\big\rangle$

where $Q_{ij}$ is the traceless mass quadrupole and the dots are three time derivatives. The factor $G/c^5$ is tiny: a source radiates strongly only when its parts move at a good fraction of light speed. Vesper and Ember move at 2.9 × 10⁻⁴ of it relative to each other.

The circular pair

With reduced mass $\mu = m_1 m_2 / M$ , the pair acts like a single mass $\mu$ at distance $a$ from the centre, turning at angular speed $\omega$ :

$Q_{ij} = \mu\left(x_i x_j - \tfrac13\delta_{ij}a^2\right),\qquad (x,y) = a(\cos\omega t,\ \sin\omega t)$

The components are $Q_{xx} = \mu a^2(\tfrac12\cos2\omega t + \tfrac16)$ , $Q_{yy} = \mu a^2(-\tfrac12\cos2\omega t + \tfrac16)$ and $Q_{xy} = \tfrac12\mu a^2\sin2\omega t$ , plus a constant. They oscillate at $2\omega$ , which is why the wave frequency is twice the orbital one. Three derivatives bring down $(2\omega)^3$ , and

$\dddot Q_{ij}\dddot Q_{ij} = \dddot Q_{xx}^2 + \dddot Q_{yy}^2 + 2\dddot Q_{xy}^2 = 32\,\mu^2a^4\omega^6$

at every instant. With Kepler's law, $\omega^2 = GM/a^3$ , and $\mu^2M^3 = m_1^2m_2^2M$ :

$P = \frac{32}{5}\,\frac{G^4\,m_1^2\,m_2^2\,M}{c^5\,a^5}$

For Vesper and Ember that is 2.6 × 10¹⁵ W. The steep $a^{-5}$ is why a wide pair is so quiet even with a hole in it.

From power to a clock

The orbit holds energy $E = -G m_1 m_2/(2a)$ . Setting $dE/dt = -P$ :

$\frac{G m_1 m_2}{2a^2}\,\dot a = -P \quad\Longrightarrow\quad \dot a = -\frac{64}{5}\,\frac{G^3\,m_1 m_2 M}{c^5\,a^3}$

which is 1.2 × 10⁻¹³ m/s in 2540. Separate the variables, $a^3\,da = -\beta\,dt$ with $\beta = \tfrac{64}{5}G^3m_1m_2M/c^5$ , and integrate from the present separation $a_0$ to zero:

$T = \frac{a_0^4}{4\beta} = \frac{5}{256}\,\frac{c^5\,a_0^4}{G^3\,m_1 m_2 M}$

That is Peters' (1964) result for a circular orbit. Because $T \propto a^4$ , the separation that merges in a chosen time $t$ is $a_0 (t/T)^{1/4}$ : a factor of 1 × 10⁶ in time is only a factor of 32 in distance. The mass enters only through the product $m_1 m_2 M$ , which is why swapping Ember for a second Vesper shortens the time by a factor of 32 and no more.

The size of the wave

At distance $r$ , the strain is $h_{ij} = \dfrac{2G}{c^4 r}\,\ddot Q_{ij}$ (its transverse, traceless part). For a circular pair seen face-on this becomes

$h = \frac{4}{r}\left(\frac{G\mathcal M}{c^2}\right)^{5/3}\left(\frac{\pi f}{c}\right)^{2/3},\qquad \mathcal M = \frac{(m_1 m_2)^{3/5}}{M^{1/5}}$

where $\mathcal M$ is the chirp mass, 2.2 Suns here, and $f$ the wave frequency. With $r$ = 28 light-years this gives the 1.2 × 10⁻²¹ in the table. It holds in the wave zone, many wavelengths out. Inside about $\lambda/2\pi$ the pair's ordinary, swinging Newtonian pull dominates, and there is no wave to speak of. The whole swarm lies inside that distance, so what it feels from the pair is a slowly turning tide, not a wave.

The wobble

The centre of mass stays put, so $m_1 a_1 = m_2 a_2$ with $a_1 + a_2 = a$ . Ember's orbit around it has radius

$a_2 = a\,\frac{m_1}{M} = 1.4\ \text{AU},\qquad v_2 = \frac{2\pi a_2}{P} = 82\ \text{km/s}$

Seen from distance $d$ , the orbit subtends an angle $a_2/d$ in radians. The astrometry measures $a_2/d$ and the period; the star's spectrum measures $v_2$ and the star's own mass. Kepler's third law then fixes the unseen mass. A companion of 12 Suns that gives off no light at all can only be a black hole, the argument El-Badry and colleagues made for Gaia BH1.

Who has heard pairs like this, and who has seen one?Pulsars, LIGO, the detector bands, and the Gaia black holes

The first proof: a pulsar that spirals

The waves were inferred before they were seen. In 1975 Russell Hulse and Joseph Taylor reported a pulsar in orbit with an unseen compact companion. In 1982 Taylor and Joel Weisberg used seven years of its pulses to measure the masses, computed how fast the orbital period should decay as gravitational waves carry energy away, and found the observed decay in agreement with general relativity. By 2016 Weisberg and Yuping Huang had the ratio of observed to predicted decay at 0.9983 ± 0.0016. That pulsar's orbit is a few hours long; Vesper and Ember's is 188 days, which is why theirs decays too slowly to measure.

The first wave: GW150914

On 14 September 2015 the two LIGO detectors recorded a signal that swept from 35 to 250 hertz, with a peak strain of 1.0 × 10⁻²¹. It matched two black holes of 36 and 29 Suns spiralling in and ringing down as one hole of 62, 1.3 × 10⁹ light-years away. The missing 3 Suns of mass left as gravitational waves, 4.6 % of the total. By the end of the third observing run, in 2020, the catalogue held 90 candidate mergers. Pick GW150914 in the calculator to see how little time such a pair has left once it is loud enough for the ground.

The bands, and the gap

Ground interferometers hear from about 10 Hz; below that, the shaking of the Earth swamps them. The LISA proposal (2017) covers below 1 × 10⁻⁴ Hz to above 0.1 Hz, using spacecraft millions of kilometres apart as the arms. Pulsar timing arrays use the ticks of millisecond pulsars across the Galaxy: in 2023 NANOGrav reported evidence, from 15 years of timing 67 pulsars, for a background of waves with a strain of about 2.4 × 10⁻¹⁵ at a frequency of one cycle a year (32 nanohertz). Alberto Sesana and colleagues pointed out in 2021 that by the 2040s the microhertz band between pulsar timing and LISA would be the largest gap in coverage, and proposed a space detector for it. Vesper and Ember's 0.12 microhertz sits in that gap, and its strain is 5 × 10⁻⁷ of the background the pulsars measure. A detector to hear a single pair like this one would have to be far more sensitive than anything proposed in the 2020s.

Black holes found by wobble

Gaia, the European astrometry satellite, measures the positions and motions of stars across the Galaxy. Its data led to the first dormant black holes found this way:

Gaia BH1 (El-Badry and colleagues, 2023): a hole of 9.62 Suns and a Sun-like star of 0.93, every 185.6 days on an orbit of eccentricity 0.45, 1,566 light-years away. Kepler's law puts the pair 1.4 AU apart. It is the model for Vesper and Ember.
Gaia BH2 (El-Badry and colleagues, 2023): a dark companion of 8.9 Suns and a red giant, every 1,277 days.
Gaia BH3 (Gaia Collaboration, 2024): a hole of 32.7 Suns found in pre-release data, with an orbit of 11.6 years.

None of these pairs is close to merging, and none is bright in X-rays. They show that black holes paired quietly with ordinary stars on wide orbits exist; how such pairs form is open, and El-Badry and colleagues found that common envelope evolution can make Gaia BH1's wide orbit only under extreme assumptions.

What this page leaves out

Eccentric orbits. The formulas assume a circle. Vesper and Ember's orbit is circular by design; Gaia BH1's is not, and an eccentric pair radiates more strongly near its closest approach and merges sooner. Peters worked out how much.
Everything else that moves an orbit. The waves shrink the orbit by 111 km over Ember's life. Moving mass between the two bodies changes it far more, and the Vigil's far-future plan is to lift Ember's gas into Vesper (Thinking in a cooling universe). How the orbit evolves under that has not been worked out frontier.