I use distance- and mass-limited subsamples of the Spitzer Survey of Stellar Structure in Galaxies ( S ^ { 4 } G ) to investigate how the presence of bars in spiral galaxies depends on mass , colour , and gas content and whether large , SDSS-based investigations of bar frequencies agree with local data .
Bar frequency reaches a maximum of f _ { \mathrm { bar } } \approx 0.70 at M _ { \star } \sim 10 ^ { 9.7 } M _ { \sun } , declining to both lower and higher masses .
It is roughly constant over a wide range of colours ( g - r \approx 0.1 –0.8 ) and atomic gas fractions ( \log ( M _ { \mathrm { H } \textsc { i } } / M _ { \star } ) \approx - 2.5 to 1 ) .
Bars are thus as common in blue , gas-rich galaxies are they are in red , gas-poor galaxies .
This is in sharp contrast to many SDSS-based studies of z \sim 0.01 –0.1 galaxies , which report f _ { \mathrm { bar } } increasing strongly to higher masses ( from M _ { \star } \sim 10 ^ { 10 } to 10 ^ { 11 } M _ { \sun } ) , redder colours , and lower gas fractions .
The contradiction can be explained if SDSS-based studies preferentially miss bars in , and underestimate the bar fraction for , lower-mass ( bluer , gas-rich ) galaxies due to poor spatial resolution and the correlation between bar size and stellar mass .
Simulations of SDSS-style observations using the S ^ { 4 } G galaxies as a parent sample , and assuming that bars below a threshold angular size of twice the PSF FWHM can not be identified , successfully reproduce typical SDSS f _ { \mathrm { bar } } trends for stellar mass and gas mass ratio .
Similar considerations may affect high-redshift studies , especially if bars grow in length over cosmic time ; simulations suggest that high-redshift bar fractions may thus be systematically underestimated .