A candle that needs no ladder
Astronomers usually measure large distances with standard candles: objects whose true brightness is known, so their apparent brightness says how far away they are. Each kind of candle is calibrated against a nearer kind, rung by rung, from parallaxes of nearby stars up to supernovae in distant galaxies. An error in a lower rung climbs up the whole ladder.
A merging binary does not need that ladder. General relativity fixes how strongly a binary of given masses radiates, and the same masses fix how quickly its frequency sweeps upward, the "chirp". Measure the sweep and you know the true strength; measure the strain at the detector and you know the arriving strength. The distance follows from their ratio. Detectors record the signal much as a microphone records sound, so these sources are called standard sirens. Holz and Hughes, Using gravitational-wave standard sirens
The tilt of the orbit gets in the way
The wave is not equally strong in every direction. A binary seen face-on sends its strongest signal toward us; seen edge-on, its signal is about half as strong in one polarisation and absent in the other. A face-on binary far away and a slightly tilted binary somewhat nearer can therefore arrive with almost the same strength.
This is the main source of distance uncertainty for a single event. For the neutron-star merger GW170817 the gravitational-wave distance was 43.8−6.9+2.9Mpc, and the lopsided interval comes largely from this trade between distance and tilt. Abbott et al., A gravitational-wave standard siren measurement of the Hubble constant
Where the redshift comes from
The wave gives distance but not redshift. Expansion stretches the signal, and a stretched signal from a binary looks exactly like the signal from a heavier binary nearer by. The redshift must come from somewhere else.
For GW170817 it came from light. Telescopes found a kilonova, the glow of freshly made heavy elements, in the galaxy NGC 4993. That galaxy's group recedes at about 3000kms−1 once its own local motion is removed. Dividing velocity by distance gave a Hubble constant of 70.0−8.0+12.0kms−1Mpc−1, independent of every rung of the distance ladder. A merger with a visible counterpart is called a bright siren. Abbott et al., A gravitational-wave standard siren measurement of the Hubble constant
When nothing lights up
Most mergers are black holes, and black holes are not expected to glow. Then every galaxy inside the patch of sky and range of distance allowed by the signal is a candidate host. Each candidate implies its own Hubble constant. One event gives a blurred, many-peaked answer; with many events, the true value is the one that keeps coming back while the wrong galaxies average away. This is a dark siren.
The 2021 analysis of 47 events from the third detector catalog combined dark sirens with GW170817 and found 68−6+8kms−1Mpc−1. The authors also showed that this dark-siren result depends strongly on what they assumed about the masses of black holes in the population. Only one event, GW190814, was well enough located to be informative on its own. LIGO, Virgo and KAGRA, Constraints on the cosmic expansion history from GWTC-3
Why loud events are not a fair sample
Detectors catch only signals above a threshold. Sources whose noise happened to push them up are over-represented, and so are sources that happen to face us. If the analysis forgets this, it reads distances as systematically too short and the Hubble constant as too large. The fix is to divide by the fraction of all possible sources that would have been detected in a universe with each trial value of the Hubble constant. That fraction grows as the cube of the Hubble constant for nearby sources, so it matters a great deal.
What a siren cannot do yet
A siren is only as good as its redshift. Galaxies have their own motions of a few hundred kilometres per second, which matter most for nearby events. Galaxy catalogs miss faint galaxies, so the true host may not be listed. For very distant sources, matter along the way can magnify or dim the wave, adding a scatter of about 5 to 10 percent to each distance at high redshift. Holz and Hughes, Using gravitational-wave standard sirens
None of this removes the central idea. The distance comes from the physics of the source, so sirens test the expansion rate without borrowing any calibration from other methods. Whether they will settle the disagreement between local and early-universe measurements of the Hubble constant is still an open question.
Continue with the derivations in waveform amplitude, orbit tilt and luminosity distance.
Try it in the Lab
Start from the kilonova host preset, then drag the binary outward and watch the waveform shrink while its chirp stays the same. Open the standard sirens Lab