Back to journal

Avenyora reflection

The 2026 M87 Black Hole Images Use Two Different Data Sets

An independent reconstruction recovered M87*’s ring from older telescope records. A separate April 2026 scan starts a campaign to track changes around the black hole.

Cosmic illustration for The 2026 M87 Black Hole Images Use Two Different Data Sets

Two reports about the M87 black hole image circulated in November 2026. Samuel Lai and colleagues recovered the bright ring using a different reconstruction method on Event Horizon Telescope recordings from 2017, 2018 and 2021. The Event Horizon Telescope collaboration also presented a first image from a newly collected scan on April 3, 2026. The first report tests whether the ring survives a change in how the data are analyzed. The second starts a closer look at how the black hole’s surroundings change over time. The same target and a similar ring connect the two stories, but the evidence comes from different observing sessions.

The original image was released two years after the measurements

M87* is the supermassive black hole at the center of the galaxy Messier 87. The European Southern Observatory’s account of the original release came out in April 2019 using radio recordings made in April 2017. The collaboration synchronized telescopes separated by thousands of kilometers. Pairs of telescopes sampled different aspects of the arriving radio signal. Combining the observations let the team infer structure on extremely small angular scales.

Each pair of stations provides a partial measurement. It records how the radio waves received at two different sites line up with each other. The different separations and orientations between stations carry different information about the source. The array never had a measurement for every possible separation and direction, so the image had to be reconstructed. Multiple teams in the original collaboration already compared reconstructions to see which features survived their choices. Publication in 2019 made the result public. The light in that image was received in 2017.

What the independent team reanalyzed

Lai and colleagues’ research paper returns to public telescope records at a wavelength of 1.3 millimeters. The researchers applied their method to four observing dates from April 2017, two from April 2018 and two from April 2021. The team used combinations of measurements that cancel certain errors associated with individual stations. A change in the signal introduced at one telescope can affect several pairs. Combining those pairs in carefully chosen ways can remove the common contribution from that station and preserve information about the source.

One way to picture the idea is to follow three stations around a triangle. Each pair contributes a phase, which describes how the waves line up. When the phases are added around the triangle, a shift associated with a single station appears with opposite signs in two of the terms and cancels. Related combinations involving four stations can cancel certain multiplicative errors in the strength of the recorded signal. These quantities are called closure invariants. They make the reconstruction less dependent on a particular set of station corrections. They leave the need to account for noise and the gaps in what the network measured.

The CSIRO announcement also explains the role of machine learning. The authors’ training material was mainly ordinary non-astronomical images. The model brought general expectations about how brightness can be distributed in an image. The researchers then used the radio measurements to constrain the reconstruction. The relevant test is whether the new approach recovers the ring from the measured signal. It still uses assumptions about images and shares the underlying radio observations with the earlier work.

What the ring measurement supports

The independent team reports an angular diameter of 42.2 ± 1.5 microarcseconds for the bright ring. A microarcsecond is one millionth of an arcsecond, a very small unit of angle. The number describes how wide the ring appears from Earth. It is a measurement of the ring in the reconstructed image. Converting that angle into a black hole mass takes further assumptions, including a distance and a model for how the emission relates to the black hole. Under their stated scaling assumptions the researchers infer about 6.5 billion solar masses, consistent with the original result.

The bright ring traces emission from hot material near the black hole, shaped by the bending of light in strong gravity. The central depression is associated with the black hole’s shadow. The event horizon is the boundary from inside which light cannot escape, and the image does not map that boundary as a glowing edge. The color scale represents radio brightness. It helps show where the signal is stronger. For a separate example of how astronomers describe received light and what a measurement can support, see our guide to apparent magnitude and brightness.

What the April 2026 observation added

The first-look report from the EHT collaboration uses a single four-minute scan on April 3, 2026. It belongs to a campaign of 18 observations taken at a pace of about twice a week over a roughly two-month period. IRAM’s November 19 account describes how the institute’s NOEMA observatory in France and its 30-meter telescope in Spain contributed to the campaign. It also explains how improvements in operations, data delivery and processing helped the team turn observations into a scientifically checked image in months.

A first frame tells researchers whether they can recover the structure from one short look with the new observing setup. A sequence is needed to follow changes. In a single image, a bright patch may sit on one part of the ring. A series can reveal whether brighter emission shifts, how quickly that happens, and what persists across the observations. The first-look release presents one image. A description of the observing program is distinct from a release of all its reconstructed frames.

Read the dates as part of the evidence

An observation date tells you when the instruments collected the signal. A release date tells you when a team made its result available. The data dates in the independent reconstruction identify familiar records seen through a different analysis. The April 2026 date marks new measurements from the array. Looking for both dates is useful in other astronomy stories, too. The Bao stellar merger report combines observations from different instruments and phases of an event. Our supernova confirmation article separates a first detection from later observations that established what it was. With M87*, changing the analysis and collecting a new scan provide complementary checks.

Frequently Asked Questions

Was the first black hole image taken in 2019?

The Event Horizon Telescope released the first M87* image in 2019. It reconstructed that image from synchronized radio observations made in April 2017.

Did the independent team gather new 2026 data?

The reconstruction by Lai and colleagues used public observations taken in 2017, 2018 and 2021. The April 2026 scan came from the separate monitoring campaign run by the Event Horizon Telescope collaboration.

Is the bright ring the event horizon?

The ring shows radio emission from material close to the black hole, shaped by strong gravity. The associated shadow spans the dim central area. The event horizon is the boundary from which light cannot escape.

Did the first-look release show the full time-lapse?

The report presented one image made from four minutes of observations on April 3. The campaign collected 18 observations for a broader study of how the black hole’s surroundings change.

Privacy on this device

Clear tool inputs, results and pending saves from this browser, then reload this page. Unsaved numerology, basic compatibility and guest tarot results disappear on reload. Saved account data, downloads and other devices are separate. Your sign-in and analytics choice stay as they are.