# Reverberation mapping

Reverberation mapping is an astronomical technique that measures the time delay between variations in an active galactic nucleus's ionizing continuum and its broad emission lines, converting light-travel time into the size and structure of the broad-line region (BLR) and, with line width, a black hole mass. It is a distance-independent mass method: the mass follows from a light-travel time and a line width, without needing the object's distance, so it works beyond the local universe.<sup>[1](https://pos.sissa.it/206/010/pdf)</sup> In nearby Seyfert galaxies it places the BLR at roughly \( 10^{16} \) cm from the central engine<sup>[2](https://ned.ipac.caltech.edu/level5/Sept16/Peterson/Peterson3.html)</sup>, and velocity-resolved responses across a line profile can distinguish rotation, infall, and outflow in the gas.<sup>[3](https://ar5iv.labs.arxiv.org/html/1505.04805)</sup>

| Key fact | Value |
|---|---|
| Measured quantity | Rest-frame lag τ between continuum and line variations (τrest = τobs/(1 + z)); \( R_{\mathrm{BLR}} = c \cdot \tau \)<sup>[1](https://pos.sissa.it/206/010/pdf)</sup> |
| Mass formula | \( M_{\mathrm{BH}} = f \cdot c \cdot \tau \cdot v^{2} / G \), with order-unity virial factor f<sup>[4](https://par.nsf.gov/servlets/purl/10164498)</sup> |
| Mass range covered | ~\( 2 \times 10^{6} \) M⊙ (NGC 4051) to ~\( 1.3 \times 10^{9} \) M⊙ (PG 1426+015)<sup>[5](https://ned.ipac.caltech.edu/level5/March10/Peterson/paper.pdf)</sup> |
| Campaign sample | ~60 AGNs with successful campaigns by 2015, mostly at \( z < 0.1 \), continuum S/N > 50 per exposure<sup>[3](https://ar5iv.labs.arxiv.org/html/1505.04805)</sup> |
| Hβ R–L relation | Newer analyses find a slope of ~0.4–0.42, with rms scatter ~0.3 dex<sup>[6](https://iopscience.iop.org/article/10.3847/1538-4357/ae6109)</sup> |
| Largest survey campaign | SDSS-RM: 849 quasars, \( 0.1 < z < 4.5 \), 11 yr photometric and 7 yr spectroscopic baselines<sup>[7](https://beta.iopscience.iop.org/article/10.3847/1538-4365/ad3936)</sup> |

## How it works

A varying continuum source photoionizes BLR gas, and each parcel of gas re-emits an echo of the continuum light curve delayed by its extra light-travel path. The average lag is of order \( R/c \), where R is the typical radius of the emitting region, so time resolution substitutes for spatial resolution.<sup>[8](https://ar5iv.labs.arxiv.org/html/2105.06926)</sup> The observed line light curve is the continuum convolved with the transfer function Ψ(τ), the impulse response that encodes the geometry of the reprocessing region.<sup>[8](https://ar5iv.labs.arxiv.org/html/2105.06926)</sup> In velocity–delay space, Ψ(v, τ) is the BLR responsivity map; a cloud at position (r, θ) is seen at delay \( \tau = (1 + \cos\theta) \, r / c \), so isodelay surfaces are nested paraboloids.<sup>[2](https://ned.ipac.caltech.edu/level5/Sept16/Peterson/Peterson3.html)</sup>

The technique works because the gas responds almost instantly: at typical BLR densities of \( n_{\mathrm{e}} \sim 10^{10}\ \mathrm{cm}^{-3} \) the recombination time is negligible against light-travel times of days to weeks.<sup>[5](https://ned.ipac.caltech.edu/level5/March10/Peterson/paper.pdf)</sup> The mass follows from the virial product, \( M_{\mathrm{BH}} = f \cdot c \cdot \tau \cdot v^{2} / G \), where v is the broad-line width and f is a dimensionless virial factor set by the unknown BLR geometry and kinematics.<sup>[4](https://par.nsf.gov/servlets/purl/10164498)</sup>

## How it is done

A campaign monitors an AGN spectroscopically and photometrically, extracting the continuum and integrated line light curves, then estimates the lag. Simulations show that high-fidelity velocity-delay maps require duration \( T_{\mathrm{dur}} \) exceeding the BLR light-crossing time by at least a factor of 3 and delay resolution \( \Delta\tau \approx 2\Delta t \); maps became recognizable only with ≳60 days of data regardless of sampling grid.<sup>[9](https://iopscience.iop.org/article/10.1086/420755/meta)</sup> Successful campaigns have mostly used 1.0–4.0 m telescopes with continuum signal-to-noise >50 per exposure and substantially higher in the emission lines, where variability amplitude is only a few percent.<sup>[3](https://ar5iv.labs.arxiv.org/html/1505.04805)</sup>

Lag estimation methods include the interpolated cross-correlation function (ICCF), the discrete correlation function, z-transformed cross-correlation, regularized linear inversion, the Fourier cross-spectrum, JAVELIN, and CREAM.<sup>[4](https://par.nsf.gov/servlets/purl/10164498)</sup> ICCF takes the CCF peak or centroid as the lag, with uncertainties from flux randomization and random subset selection; JAVELIN models variability as a damped random walk, fits a top-hat transfer function by Markov Chain Monte Carlo, and takes uncertainty from the posterior width.<sup>[4](https://par.nsf.gov/servlets/purl/10164498)</sup> Which is more reliable is unsettled: SDSS-RM adopts JAVELIN lags as fiducial after a comparison found it gave the best lag recovery for its data quality<sup>[10](https://google.iopscience.iop.org/article/10.3847/2041-8213/ab3e0f)</sup>, while a uniform reanalysis of 212 AGNs judged ICCF the most reliable despite larger uncertainties, finding JAVELIN more susceptible to underestimation from misestimated flux uncertainties.<sup>[11](https://google.iopscience.iop.org/article/10.3847/1538-4365/ad74f2)</sup> Host-galaxy starlight contamination must be corrected; the current Hβ R–L calibration required high-resolution HST images and two-dimensional image decompositions.<sup>[3](https://ar5iv.labs.arxiv.org/html/1505.04805)</sup>

## Origin

Reverberation mapping was introduced by [R. D. Blandford](https://www.edgepedia.org/roger-blandford) and [C. F. McKee](https://www.edgepedia.org/christopher-f-mckee), whose paper "Reverberation mapping of the emission line regions of Seyfert galaxies and quasars" was published in 1982 in The Astrophysical Journal.<sup>[12](https://doi.org/10.1086/159843)</sup><sup> • </sup><sup>[13](https://www.osti.gov/biblio/5080299)</sup> They articulated the mathematical formalism and coined the term "reverberation mapping".<sup>[14](https://arxiv.org/pdf/astro-ph/9706039)</sup> The paper builds on earlier work that calculated the response of total line intensity in a spherical gas distribution to continuum variability.<sup>[15](https://adsabs.harvard.edu/pdf/1982ApJ...255..419B)</sup> Blandford and McKee estimated that about 50 error-free observations over roughly 10 transfer-function widths, or biweekly observations for 2 years, would suffice for a luminous Seyfert.<sup>[15](https://adsabs.harvard.edu/pdf/1982ApJ...255..419B)</sup>

UV and optical spectroscopic monitoring of a few Seyferts in the 1980s constituted proof of concept, with the International AGN Watch consortium achieving the greatest success.<sup>[14](https://arxiv.org/pdf/astro-ph/9706039)</sup> The first highly successful campaign ran in 1988–89, combining IUE ultraviolet spectra with ground-based optical data for over 8 months at few-day resolution: He II λ1640 responded with a ~2 day lag, Lyα and C IV with ~10 days, Hβ with ~20 days, and C III] with nearly 30 days, revealing radial ionization stratification.<sup>[5](https://ned.ipac.caltech.edu/level5/March10/Peterson/paper.pdf)</sup>

## Variants

Reverberation mapping now spans four size scales: the X-ray corona (a few to tens of gravitational radii), the UV/optical accretion disk, the BLR, and the dusty torus; X-ray reverberation lags were first discovered in 2009 with [XMM-Newton](https://www.edgepedia.org/xmm-newton), using the cross-spectrum phase φ(f) converted to a time lag τ(f) = φ(f)/2πf, and near-IR dust reverberation measures the torus inner edge where dust sublimates at 1300–2000 K.<sup>[8](https://ar5iv.labs.arxiv.org/html/2105.06926)</sup> Photometric RM uses continuum bands instead of spectra.

The Hβ radius–luminosity relation turns a single spectrum and luminosity into a mass. The initial calibration had a slope of 0.7; correction for host-galaxy starlight using HST images revised it to ~0.5, as photoionization physics predicts.<sup>[16](https://beta.iopscience.iop.org/article/10.3847/1538-4357/ad132f)</sup> Newer analyses find a significantly flatter slope, ~0.4–0.42.<sup>[6](https://iopscience.iop.org/article/10.3847/1538-4357/ae6109)</sup> For C IV, the BLR size–UV luminosity slope was updated from 0.61 ± 0.05 to 0.44 ± 0.02 with additional high-luminosity points.<sup>[17](https://www.frontiersin.org/journals/astronomy-and-space-sciences/articles/10.3389/fspas.2017.00031/full)</sup> Hβ shifts out of the optical bandpass at \( z = 0.7\text{–}0.8 \), so high-redshift masses must be bootstrapped through Mg II and C IV.<sup>[3](https://ar5iv.labs.arxiv.org/html/1505.04805)</sup>

Named campaigns include SDSS-RM, which monitored 849 quasars at \( 0.1 < z < 4.5 \) and measured lags for Hα, Hβ, Mg II, and C IV<sup>[10](https://google.iopscience.iop.org/article/10.3847/2041-8213/ab3e0f)</sup><sup> • </sup><sup>[7](https://beta.iopscience.iop.org/article/10.3847/1538-4365/ad3936)</sup>, and the SEAMBH project on super-Eddington accretors, which found deviations from the R–L relation correlating with accretion rate and Fe II/Hβ ratio.<sup>[16](https://beta.iopscience.iop.org/article/10.3847/1538-4357/ad132f)</sup>

A 2024 framework distinguishes emissivity-weighted Ψ_e(v, τ) and responsivity-weighted Ψ_r(v, τ) transfer functions: the mean spectrum traces Ψ_e and the rms spectrum traces Ψ_r, with ΔF_l(v, t) = ∫ Ψ_r(v, τ) ΔF_c(t − τ) dτ. The rms spectrum is proportional to the responsivity-weighted transfer function only when the continuum variation timescale greatly exceeds the BLR delay extension; otherwise virial products deviate by up to ~0.3 dex at short variation timescales.<sup>[18](https://arxiv.org/pdf/2412.10777)</sup> The same framework explains the long-standing shape difference between mean and rms spectra through radially increasing BLR responsivity, the "geometric dilution" effect, and implies the virial factor should depend on luminosity state.<sup>[18](https://arxiv.org/pdf/2412.10777)</sup> New calibrations add the [Eddington ratio](https://www.edgepedia.org/eddington-ratio) as a third parameter, defining a fundamental plane of Hβ time delay, optical luminosity, and line velocity with 0.21 dex intrinsic scatter and ~0.1 dex scatter in predicted \( \tau_{\mathrm{H}\beta} \); previous two-parameter relations overestimated masses of high-Eddington AGNs by up to 0.5 dex.<sup>[6](https://iopscience.iop.org/article/10.3847/1538-4357/ae6109)</sup>

## Applications

A comprehensive reanalysis of 117 reverberation data sets on 37 AGNs yielded masses from ~\( 2 \times 10^{6} \) M⊙ to ~\( 1.3 \times 10^{9} \) M⊙, with statistical errors around 30%.<sup>[5](https://ned.ipac.caltech.edu/level5/March10/Peterson/paper.pdf)</sup> The virial interpretation is supported by the anticorrelation \( \tau \propto \Delta v^{-2} \) between line width and response time, with high-ionization lines broader and shorter-lagged than low-ionization lines.<sup>[9](https://iopscience.iop.org/article/10.1086/420755/meta)</sup> The virial factor dominates the systematic error: calibrations from the M_BH–σ* relation range from 〈f〉 = 2.8 to 〈f〉 = 5.9 depending on the study, and SDSS-RM adopts f = 4.47.<sup>[7](https://beta.iopscience.iop.org/article/10.3847/1538-4365/ad3936)</sup><sup> • </sup><sup>[10](https://google.iopscience.iop.org/article/10.3847/2041-8213/ab3e0f)</sup> Dynamical BLR modeling can derive individual virial coefficients, removing the systematics of an average f.<sup>[7](https://beta.iopscience.iop.org/article/10.3847/1538-4365/ad3936)</sup> Because the single-epoch virial method has been applied to several hundred thousand objects accounting for over 99% of all measured supermassive black holes, RM measurements at low redshift anchor essentially all of those masses.<sup>[19](https://www.nature.com/articles/s41467-026-69166-w)</sup>

## Limitations and alternatives

The ambiguous size–velocity combination in the virial product carries an uncertainty of about a factor of two.<sup>[14](https://arxiv.org/pdf/astro-ph/9706039)</sup> Luminous objects are costly: an AGN with \( L_{5100} = 10^{46}\ \mathrm{erg\ s}^{-1} \) at \( z = 1 \) is expected to show an Hβ lag of 500–600 observed-frame days, requiring 5–10 years of monitoring<sup>[16](https://beta.iopscience.iop.org/article/10.3847/1538-4357/ad132f)</sup>, and high-luminosity RM suffers from smaller variability amplitude and cosmological time dilation.<sup>[17](https://www.frontiersin.org/journals/astronomy-and-space-sciences/articles/10.3389/fspas.2017.00031/full)</sup> The virial factor is found to be inversely proportional to observed line width, so a single constant f can mis-estimate single-epoch masses by up to a factor of 6, overestimating masses for FWHM(Hα) ≳ 4,000 km/s and underestimating them below that.<sup>[20](https://www.frontiersin.org/journals/astronomy-and-space-sciences/articles/10.3389/fspas.2017.00070/full)</sup> FWHM-based C IV single-epoch masses are susceptible to the largest biases, up to an order of magnitude, because the line peak is contaminated by a non-variable C IV component; RM-based C IV masses avoid this by measuring velocities from the rms spectrum.<sup>[1](https://pos.sissa.it/206/010/pdf)</sup> RM and single-epoch methods also probe different BLR components: RM isolates the variable component, which "breathes" between campaigns, and [NGC 5548](https://www.edgepedia.org/ngc-5548), monitored with nearly 1600 spectra over 43 years, has exhibited BLR "holidays" in which line and continuum variations were decorrelated.<sup>[19](https://www.nature.com/articles/s41467-026-69166-w)</sup> Long-term variability also limits single-epoch precision: a 20-year Hβ luminosity change standard deviation of 0.26 dex implies scatter in single-epoch R estimates for single snapshots at an R–L slope of 0.45.<sup>[19](https://www.nature.com/articles/s41467-026-69166-w)</sup> As an alternative, accretion-disk SED-fitting mass estimates are independent of BLR properties and of comparable accuracy to single-epoch virial masses<sup>[20](https://www.frontiersin.org/journals/astronomy-and-space-sciences/articles/10.3389/fspas.2017.00070/full)</sup>; near-IR dust reverberation probes the torus rather than the BLR.<sup>[8](https://ar5iv.labs.arxiv.org/html/2105.06926)</sup>

## References

1. [Reverberation Mapping: Masses and Distance (Denney, Proceedings of Science)](https://pos.sissa.it/206/010/pdf)
2. [The Broad-Line Region in Active Galactic Nuclei (Peterson, NED Level 5)](https://ned.ipac.caltech.edu/level5/Sept16/Peterson/Peterson3.html)
3. [AGN Reverberation Mapping (Bentz & Katz 2015 review)](https://ar5iv.labs.arxiv.org/html/1505.04805)
4. [On reverberation mapping lag uncertainties (Yu et al., MNRAS)](https://par.nsf.gov/servlets/purl/10164498)
5. [Reverberation mapping of active galactic nuclei (Peterson & Horne, NED Level 5)](https://ned.ipac.caltech.edu/level5/March10/Peterson/paper.pdf)
6. [New Black Hole Mass Calibrations and the Fundamental Plane of the Broad-line Region Size, Luminosity, and Velocity (ApJ)](https://iopscience.iop.org/article/10.3847/1538-4357/ae6109)
7. [The Sloan Digital Sky Survey Reverberation Mapping Project: Key Results (ApJS)](https://beta.iopscience.iop.org/article/10.3847/1538-4365/ad3936)
8. [Reverberation mapping of Active Galactic Nuclei: from X-ray corona to dusty torus (Cackett et al. review)](https://ar5iv.labs.arxiv.org/html/2105.06926)
9. [Observational Requirements for High-Fidelity Reverberation Mapping (Horne et al. 2004, ApJ)](https://iopscience.iop.org/article/10.1086/420755/meta)
10. [The Sloan Digital Sky Survey Reverberation Mapping Project: Improving Lag Detection with an Extended Multiyear Baseline (ApJL 2019)](https://google.iopscience.iop.org/article/10.3847/2041-8213/ab3e0f)
11. [Revisiting the Hβ Size–Luminosity Relation Using a Uniform Reverberation-mapping Analysis (ApJS)](https://google.iopscience.iop.org/article/10.3847/1538-4365/ad74f2)
12. [R. D. Blandford, C. F. McKee (1982). Reverberation mapping of the emission line regions of Seyfert galaxies and quasars. The Astrophysical Journal.](https://doi.org/10.1086/159843)
13. [Reverberation mapping of the emission line regions of Seyfert galaxies and quasars (OSTI record)](https://www.osti.gov/biblio/5080299)
14. [Reverberation Mapping of the Broad Line Region in Active Galactic Nuclei (Peterson, AGN Watch review, 1997)](https://arxiv.org/pdf/astro-ph/9706039)
15. [Reverberation Mapping of the Emission Line Regions of Seyfert Galaxies and Quasars (Blandford & McKee 1982, ApJ 255, 419)](https://adsabs.harvard.edu/pdf/1982ApJ...255..419B)
16. [The Seoul National University AGN Monitoring Project. III. Hβ Lag Measurements of 32 Luminous AGNs (ApJ)](https://beta.iopscience.iop.org/article/10.3847/1538-4357/ad132f)
17. [Reverberation Mapping of High-Luminosity Quasars (Kaspi et al. 2017, Frontiers)](https://www.frontiersin.org/journals/astronomy-and-space-sciences/articles/10.3389/fspas.2017.00031/full)
18. [A rigorous framework for BLR reverberation mapping with locally dependent responsivity (2024)](https://arxiv.org/pdf/2412.10777)
19. [Strong long-term variability in active galactic nuclei affects virial black hole mass measurements | Nature Communications](https://www.nature.com/articles/s41467-026-69166-w)
20. [The Virial Factor and Biases in Single Epoch Black Hole Mass Determinations (Mejía-Restrepo et al. 2017, Frontiers)](https://www.frontiersin.org/journals/astronomy-and-space-sciences/articles/10.3389/fspas.2017.00070/full)

---
*Topic: Encyclopedia › Physical world and mathematics › Astronomy › Cosmology and observation › Observational techniques: astrometry, photometry, spectroscopy*

*Initially written Sep 29, 2026 · Reviewed: Sep 30, 2026 · Edited: Sep 30, 2026 · Last review: Sep 30, 2026*

*Copyright 2026 EdgeChat AI, a subsidiary of Biostate AI.*

License: Edgepedia Community License 1.0, https://www.edgepedia.org/edgepedia/license