We use H \alpha and far-ultraviolet ( FUV , 1539 Ã ) GALEX data for a large sample of nearby objects to study the high mass ( m \geq 2 M \odot ) star formation activity of normal late-type galaxies .
The data are corrected for dust attenuation using the most accurate techniques at present available , namely the Balmer decrement for H \alpha data and the total far-infrared to FUV flux ratio for GALEX data .
The sample shows a highly dispersed distribution in the H \alpha to FUV flux ratio ( Log f ( H \alpha ) / f ( FUV ) = 1.10 \pm 0.34 Ã ) indicating that two of the most commonly used star formation tracers give star formation rates with uncertainties up to a factor of 2-3 .
The high dispersion is partly due to the presence of AGN , where the UV and the H \alpha emission can be contaminated by nuclear activity , highly inclined galaxies , for which the applied extinction corrections are probably inaccurate , or starburst galaxies , where the stationarity in the star formation history required for transforming H \alpha and UV luminosities into star formation rates is not satisfied .
Excluding these objects , normal late-type galaxies have Log f ( H \alpha ) / f ( FUV ) = 0.94 \pm 0.16 Ã , which corresponds to an uncertainty of \sim 50 % on the SFR .
The H \alpha to FUV flux ratio of the observed galaxies increases with their total stellar mass .
If limited to normal star forming galaxies , however , this relationship reduces to a weak trend that might be totally removed using different extinction correction recipes .
In these objects the H \alpha to FUV flux ratio seems also barely related with the FUV-H colour , the H band effective surface brightness , the total star formation activity and the gas fraction .
The data are consistent with a Kroupa ( 2001 ) and Salpeter initial mass function in the high mass stellar range ( m > 2 M \odot ) and imply , for a Salpeter IMF , that the variations of the slope \gamma can not exceed 0.25 , from \gamma = 2.35 for massive galaxies to \gamma = 2.60 in low luminosity systems .
We show however that these observed trends , if real , can be due to the different micro history of star formation in massive galaxies with respect to dwarf systems .