We describe an algorithm to detect voids in galaxy redshift surveys .
The method is based on the void finder algorithm of El-Ad & Piran .
We apply a series of tests to determine how accurately we are able to recover the volumes of voids using our detection method .
We simulate voids of different ellipticity and find that if voids are approximately spherical , our algorithm will recover 100 % of the volume of the void .
The more elliptical the void , the smaller the fraction of the volume we can recover .
We insist that voids lie completely within the survey .
Voids close to the edge of the survey will therefore be underestimated in volume .
By considering a deeper sample , we estimate the maximal sphere diameters are correct to within 30 % .

We apply the algorithm to the Point Source Catalogue Survey ( PSCz ) and the Updated Zwicky Catalog ( UZC ) .
The PSCz survey is an almost all-sky survey with objects selected from the IRAS catalog .
The UZC covers a smaller area of sky but is optically selected and samples the structures more densely .
We detect 35 voids in the PSCz and 19 voids in the UZC with diameter larger than 20 h ^ { -1 } Mpc .
Using this minimum size threshold , voids have an average effective diameter of 29.8 \pm 3.5 h ^ { -1 } Mpc ( PSCz ) and 29.2 \pm 2.7 h ^ { -1 } Mpc ( UZC ) and that they are underdense regions with \delta \rho / \rho values of -0.92 \pm 0.03 ( PSCz ) and -0.96 \pm 0.01 ( UZC ) respectively .
Using this quite stringent threshold for void definition , voids fill up to 40 % of the volume of the universe .