We explore to which extent stars within Galactic disk open clusters resemble each other in the high-dimensional space of their photospheric element abundances , and contrast this with pairs of field stars .
Our analysis is based on abundances for 20 elements , homogeneously derived from apogee spectra ( with carefully quantified uncertainties , with a median value of \sim 0.03 dex ) .
We consider 90 red giant stars in seven open clusters and find that most stars within a cluster have abundances in most elements that are indistinguishable ( in a \chi ^ { 2 } -sense ) from those of the other members , as expected for stellar birth siblings .
An analogous analysis among pairs of > 1000 field stars shows that highly significant abundance differences in the 20-dimensional space can be established for the vast majority of these pairs , and that the apogee -based abundance measurements have high discriminating power .
However , pairs of field stars whose abundances are indistinguishable even at 0.03 dex precision exist : \sim 0.3 % of all field star pairs , and \sim 1.0 % of field star pairs at the same ( solar ) metallicity [ Fe/H ] = 0 \pm 0.02 .
Most of these pairs are presumably not birth siblings from the same cluster , but rather doppelganger .
Our analysis implies that “chemical tagging” in the strict sense , identifying birth siblings for typical disk stars through their abundance similarity alone , will not work with such data .
However , our approach shows that abundances have extremely valuable information for probabilistic chemo-orbital modeling and combined with velocities , we have identified new cluster members from the field .