We present an analysis of voids in the 2dF Galaxy Redshift Survey ( 2dFGRS ) .
This analysis includes identification of void regions and measurement of void statistics .
The 2dFGRS is the largest completed redshift survey to date , including a total of 245,591 galaxies covering 1500 deg ^ { 2 } to a median depth of z _ { med } \sim 0.11 .
We use the voidfinder algorithm to identify a total of 289 voids in the 2dFGRS with radius larger than 10 h ^ { -1 } Mpc .
These voids have an average effective radius of 14.89 \pm 2.65 h ^ { -1 } Mpc in the North Galactic Pole region ( NGP ) and 15.61 \pm 2.84 h ^ { -1 } Mpc in the South Galactic Pole region ( SGP ) .
These voids are extremely underdense , with average density contrast of \delta \rho / \rho = -0.94 \pm 0.02 .
The centers of voids are even emptier , because the few galaxies within the voids typically lie close to the edges .
The total volume of the universe filled by these void regions is approximately 40 % .
These results are very similar to results found from our analysis of the PSCz survey and the Updated Zwicky Catalog ; here we detect almost a factor of 10 more voids .
We measure the Void Probability Function ( VPF ) of the 2dFGRS for volume-limited samples with limiting absolute magnitudes , M _ { lim } -5log ( h ) , from -16 to -21 in b _ { J } .
We measure the Underdensity Probability Function ( with density contrast threshold \delta \rho / \rho = -0.8 ) for samples with limiting absolute magnitudes , M _ { lim } -5log ( h ) , from -18 to -21 .
We find that the SGP is more underdense than the NGP for all but the brightest sample under consideration .
There is good agreement between the VPF ’ s of the Center for Astrophysics Survey and the 2dFGRS .
Comparison of VPF ’ s measured for the 2dFGRS with the distribution of simulated dark-matter halos of similar number density indicates that voids in the matter distribution in \Lambda CDM simulations are not empty enough .
However , semi-analytic models of galaxy formation that include feedback effects yield VPF ’ s that show excellent agreement with the data .