The -T switch sets a user-defined tolerance used by many computations of TetGen, default is 1e-8.

The main reason to use the tolerance is that, in principle, the vertices which are used to define a facet of a PLC should be exactly coplanar. But this is very hard to achieve in practice due to the inconsistent nature of the floating-point format used in computers. TetGen accepts facets which their vertices are not exactly coplanar but approximately due to a tolerance. In TetGen, four points a, b, c, and d are assumed be coplanar if the ratio v / l^3 is smaller than the given tolerance, where v is the volume of the tetrahedron abcd, and l is the average edge length of tetrahedron abcd.