deletions | additions       diff --git a/reconstructed_wedges.tex b/reconstructed_wedges.tex  index e66c4ec..05f01d2 100644  --- a/reconstructed_wedges.tex  +++ b/reconstructed_wedges.tex 
...
The pile-up of many of the realization modes along the $\epsilon$ cutoff lines in Figure \ref{fig:wizcola_hdaModes_z60_epsilonT15} demonestrate the limited constraining power of an average WiggleZ volume. We do find, however, that the high significance detection threshold is concentrated around the true values of $\alpha_{||},\alpha_\perp=1$. We find $<\alpha_{||}>=0.999\pm 0.154$, $\alpha_{\perp}>=1.001\pm0.081$ (medians, standard deviations). These statistics vary when we change the threshold of the subset.   In Figure \ref{fig:wizcola_hdaUnc_z60_epsilonT15} we notice that the subset are predominantly in the low uncertainties. Their statistics are $\sigma_{\alpha_{||}}$  %The resulting mock modes of $\alpha_{||}$, $\alpha_{\perp}$ obtained without using a prior on $\epsilon$ and $\alpha$ are displayed in Figure REF and their uncertianties in Figure REF. These clearly shows the limitation of constraining power of the WiggleZ volume. Similar results when using flat priors on $\epsilon$ of $15\%$ and $7\%$ are displayed in Figures REF. We clearly see that the results are both less biassed and yield tighter constraints when ruling out the parameter space, pointing out that our mechanism of reconstruction and fitting algorithm do not raise biases that are larger than the statistical uncertainties. ... a few words about $\alpha$ priors....