Every ultra-massive rocky planet orbits an active star
The study starts from a simple population fact. Among 333 small planets with published masses and radii, the authors identify four worlds whose mass exceeds 10 Earth masses while their radius stays consistent with a volatile-poor rocky composition: WASP-84 c, TOI-1347 b, TOI-2322 c, and HD 63433 b. They add four more planets that sit just outside the strict rocky density regime but are still too dense to carry much hydrogen and helium: GJ 523 b, TOI-815 c, HD 18599 b, and GJ 143 b.
The decisive observation is where those eight planets live. All of them orbit stars with log R'HK above -4.75, the calcium H and K chromospheric index used here as a tracer of magnetic activity, with the Vaughan-Preston gap separating active from inactive stars. In the authors' comparison sample of 37 stars with measured activity, 24 fall in the inactive or quiet regions and 13 in the active region.
That imbalance is the whole argument in miniature. If Mega-Earths were distributed uniformly across activity levels, they should appear around quiet stars too. Instead, the entire set sits on the active side.
A below-1% chance of coincidence
The authors turn the clustering into a probability. Treating the 37 stars as the pool, the odds that a single Mega-Earth orbits an active star are 13/37. For the four strictly rocky cases, the chance that all four land on active hosts is 13/37 × 12/36 × 11/35 × 10/34, which equals 0.01 — about 1%.
Including the four additional volatile-rich candidates drops the figure to 0.0003. The paper reports this as a below-1% likelihood that all literature ultra-massive rocky planets would coincidentally orbit such active stars.
The calculation assumes a uniform underlying distribution across activity levels, and the authors pair it with a physical argument: they find no astrophysical reason for Mega-Earths to prefer active hosts. If anything, unusual formation pathways such as post-disk collisions or envelope stripping should favor older systems that had time to complete them. A statistical excess without a formation mechanism points toward a measurement problem rather than a new planet class.
Starspots can mimic or inflate a planet signal
The mass measurements in question come from radial velocities, which detect a planet through the star's back-and-forth motion. On a magnetically active star, spots and plages suppress convection and distort the observed spectral lines, adding jitter that can mimic a planetary signal as the star rotates. That extra scatter can bury a small planet signal or artificially inflate the semi-amplitude attributed to the planet, and an inflated semi-amplitude translates directly into an inflated mass.
The paper documents this risk planet by planet. WASP-84 shows chromospheric activity at log R'HK = -4.42 ± 0.04 and starspot modulation at a 14.36-day rotation period, with reported jitter of about 13 m/s that exceeds the planetary semi-amplitude of 7.7 m/s. TOI-2322 c has a 20.23-day orbit that coincides with its star's 21.28-day rotation period. HD 63433 b orbits in 7.1 days around a star rotating in 6.45 days, and its own discovery paper cautions that the Gaussian Process may be overfitting the radial velocities.
Gaussian Process regression is the standard tool for separating correlated stellar noise from a planetary signal, and every one of these analyses used it with photometry-informed priors on rotation period and spot evolution. The authors' point is that this best practice did not prevent implausibly large masses.
Signal-to-noise and rotation period set the risk
The paper does not treat all eight cases as equally suspect. It applies a signal-to-noise criterion from prior work, K/N = (K_pl / RV_rms) × √N_obs, with a recommended threshold of 7.5 for reliable recovery of a planetary signal. Three of the five Mega-Earths analyzed in detail fall below it: WASP-84 c at 3.8, TOI-2322 c at 6.5, and HD 63433 b at 5.3. HD 18599 b sits exactly at 7.5, while the remaining cases fall between 11 and 16.
A second risk factor is resonance with the star's rotation. GJ 143 b, HD 63433 b, and TOI-2322 c have orbital periods similar to their host's rotation period, while HD 18599 b and TOI-815 c orbit at harmonics of it. Signals at those periods are exactly where stellar activity is most amplified.
Even the three planets with easily separable rotation and orbital periods are judged likely to have inflated radial-velocity signals. The authors therefore question the reliability of Gaussian Processes for mass measurements of rocky planets around active stars in general, not just for the weakest cases.
The threshold is theoretical, not a measured edge
The 10 Earth-mass line is not an observed boundary. It comes from formation theory and simulations of collisional mantle stripping and post-formation erosion, which suggest that even under extreme conditions a primordial rocky planet is unlikely to exceed roughly 10 Earth masses, because larger cores almost certainly undergo runaway gas accretion. The empirical transition from bare rocky planets to envelope-bearing planets instead appears at 1.5 to 1.75 Earth radii, corresponding to about 4 to 5 Earth masses.
Keeping those two numbers apart matters. The 4 to 5 Earth-mass figure marks where volatile-rich planets become common in the observed population; the 10 Earth-mass figure marks where theory says a rocky core should no longer avoid a gas envelope. The Mega-Earth claim lives in the gap between them, which is why the authors treat it as requiring exceptional formation or evolution rather than ordinary processes.
Their conclusion is correspondingly narrow: there is no compelling evidence for rocky planets forming above the upper mass limit of 10 Earth masses. They do not claim that such planets are impossible, only that the current mass measurements do not establish them.
Better sampling could rescue active-star mass measurements
The paper's skepticism is aimed at current measurement practice, not at the possibility of ever characterizing planets around active stars. It notes that Gaussian Processes should perform better when radial velocities are scheduled to sample both the stellar activity cycle and the planetary signal, so that each can be modeled rather than absorbed into the other.
A specific proposal follows from the sample: high-cadence radial velocities for short and ultra-short period planets, combined with a nightly baseline offset, could reduce the effect of stellar activity on mass determination. The authors also point to the value of photometry-informed priors and of fitting activity indicators alongside the radial velocities, as several of the original analyses did.
Until such data exist, the practical priority is a comparison rule. A rocky-density planet above 10 Earth masses deserves particular scrutiny when its host has log R'HK > -4.75, when its orbital period is close to or a harmonic of the stellar rotation period, and when its radial-velocity signal-to-noise falls below the recommended threshold. Those are the conditions under which the paper expects activity to have manufactured the mass.
Sources used in this answer
"Mega-Earths" are likely an artifact of stellar activity
All eight literature ultra-massive rocky planets orbit stars with log R'HK > -4.75, a clustering with below-1% probability under a uniform distribution, leading the authors to conclude that stellar activity inflates their radial-velocity masses.