PCA doesn't remove mean

I noticed in the documentation that in the PCA analyzer, the variance is dominated by a single component which is constant across nodes. This is a rookie mistake (mine) in handling the analysis, as the node- mean should be subtracted before computing the PCA.