We have compared direct measurements of the rate of dissipation of turbulent kinetic energy, epsilon, to estimates using Batchelor fitting techniques in controlled laboratory experiments. Turbulence was generated uniformly throughout a linearly salt stratified fluid by the continual horizontal oscillation of a rigid vertical grid. Batchelor estimates of epsilon were obtained by processing data acquired from traversing fast response thermistors through the fluid in three orthogonal directions. It was found that the Batchelor estimates from the two probes traversing in the plane of the grid were similar to each other, and systematically larger than those from probes traversed perpendicular to the grid plane. We show that this is related to the spatial inhomogeneity of the high wavenumber content of the temperature gradient field generated by the grid. Hence, our experiments demonstrate that care needs to be taken when using these techniques to estimate epsilon in grid-generated turbulence with zero mean flow. Turbulent lengthscales were also measured from the same traverse data, and it was found that the estimates of this quantity from all three traverse directions were similar.