The study examined 225 targets drawn from Rowan et al. (2023), each described as having two different eclipsing periods. It reports 76 confirmed blends, with corrected source identifications and respective periods. For the candidate set it examined, the reported false-positive fraction was 33.8%.
Here, the 76 figure refers to systems, not individual eclipse signals. A blend is a source-identification problem: signals assigned to a doubly eclipsing candidate were linked instead to separate eclipsing binaries. The analysis reports the corrected source identifications and periods for those cases.
The central question was whether each periodic signal had been assigned to the correct source. The reported output covers both source identifications and periods, rather than source labels alone.
The item is a preprint dated 27 August 2026.
Following each signal to its source
To make that check, the author inspected a 200-arcsecond-radius area around each candidate. The search looked for resolved nearby variable sources and measured their periods. Its stated purpose was to verify the source of each periodic signal.
The cross-check relied mainly on the VSX catalogue. It was supplemented by Gaia DR3, ASAS-SN, ZTF, ATLAS and ASAS, using both catalogue records and photometric information.
The analysis compared nearby sources with the periods attached to the candidates. Its reported corrections cover the source identifications and, in some cases, the periods themselves.
The reclassification changed more than a label
The paired sources were typically separated by dozens of arcseconds, with some more than 100 arcseconds apart. The inspection around each candidate used a radius of 200 arcseconds.
Only 13 of the 76 detected systems were also listed in Rowan et al. (2023) Table 2. The remaining systems were presented as confirmed blends for the first time in this analysis.
The changes could affect the period as well as the source attached to a candidate. Some corrected periods were half or double the values given by Rowan et al. (2023).
Possible explanations in the data
The author presents relatively large TESS pixels and insufficient cross-checking of available catalogues and photometric data as probable explanations for the blends.
In the introduction, TESS's susceptibility to source misidentification is linked to pixels measuring 21 arcseconds. In its conclusion, the paper recommends rigorous detection using all available means.
That recommendation matches the method described in the paper: nearby variable sources were sought, their periods measured and multiple catalogues and photometric sources consulted as part of source verification.
A result tied to one candidate list
The 33.8% figure is a descriptive result for the 225 targets drawn from Rowan et al. (2023). It is reported alongside the 76 confirmed blends identified in that set, with corrected source identifications and periods. That keeps the number tied to the candidate list examined in the paper.
Publication and data
The author's research was supported by the COOPERATIO - PHYSICS project of Charles University in Prague. The manuscript says data not already included in its tables will be shared on reasonable request to the corresponding author.
Paper data and sources
Original title: Large fraction of blends from Rowan et al. (2023)
Authors: P Zasche
Journal/Repository: 2025MNRAS.544L.174Z
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text