The 2024 ASASSN-21js code makes g-band flux errors 1.723 times too large for its stated scale

We tested the public light-curve code and the 2024 data archive for ASASSN-21js. The code divides g-band flux errors by a mean magnitude of 13.02280. It divides the g-band flux by a mean flux of 22.43688 mJy. With that flux scale held fixed, the errors are 1.722893 times the correct values. The corrected weights are 2.968359 times the code weights. Our independent calculation matches all 245 time bins from the original script. The correction changes flux in eight bins that contain both filters, by at most 0.008213 in relative flux. It does not change flux in bins that contain only g data. The archive also has only one g observing night before the reported shallow event at HJD minus 2400000.5 = 58250. Camera pairs differ by up to 0.0315 mag in their median measurements on the same nights before the deep event. Five cameras with at least five nights near the deep minimum each show a flux loss of 17.5 to 19.1 percent from their earlier level. These results establish a code correction and limited data coverage before the shallow event. They do not reject the ring model or test the 2027 end date. We have not established that the published fit used the same error column. We supply source files, corrected time bins, camera-night data, figures, and code.

ASASSN-21js · ASAS-SN · photometry · flux errors · camera calibration · occultation · reproducibility