Hi Jayne,
Do you mean an ordinary cluster (as a spatial cluster) after a threshold has been applied? The cluster as a whole is meant to be considered significant or not, but you can get a measure of spatial variability if you compute the standard deviation of the effect size across voxels of that cluster.
Regarding variability across subjects that entered into the analysis, the variability isn't defined for the clusters, but for the individual voxels. This variability (standard error) is in the VARCOPE file (from FEAT). You can also use the command fsl_glm with the option "--out_varcb".