Ten ASASSN-21qj infrared visits give a constant-colour p value of 0.028 under the stated error model

We test whether one added-emission colour fits ten NEOWISE visits of the blended ASASSN-21qj source from April 2018 to November 2022. We reuse a published reduction with 23 nearby controls and separate seasonal baselines. We keep the two bands paired in 12,000 bootstrap draws and add an empirical error allowance. A joint fit treats both measured fluxes as uncertain and fits one true amplitude per visit. The constant-colour fit gives Q = 18.55 for 9 degrees of freedom. Its p value from 10,000 null simulations is 0.028. For five bright visits and five declining visits, the decline/bright excess-colour ratio is 0.925, with an approximate 95% profile interval of 0.851–1.005. If the bright phase is a 1000 K blackbody, the same ratio allows a decline temperature of 995–1182 K under that spectral model. The constant-colour mismatch is weak and depends on the reduction. Removing transient-flagged exposures gives p = 0.132. Omitting the control correction gives p = 0.102. The test detects 20% phase colour changes in 92–94% of simulations, but 10% changes in only 29–30%. The source is unresolved and has poor point-source fits. The result requires a stable earlier flux, a stable response to the blend, and the stated Gaussian error model. It does not show that the temperature was constant or identify a collision. The package preserves raw data, a fresh archive check, code, covariance, simulations, results, and figures.

ASASSN-21qj · NEOWISE · infrared excess · colour stability · blended photometry · detection power