· Pr and Ri suffer from self correlation, and some means to avoid such false correlation is required.
· At low stability, both Pr and Ri are sensitive to errors in measured air temperature gradient.
· At moderate stability, waves and sporadic turbulence lead to uncertainty in flux estimates.
· At high stability, data are sparse, and suitable averaging requires large data sets, of the order of months
Three months of data from the austral winter at Halley Station (76oS, 26oW) are presented. The large volume of data allows harsh quality control whilst retaining sufficient records for subsequent analysis of the Pr(Ri) relationship. An empirical Butterworth function is fitted to Pr-1(Ri), using a component variable technique which avoids self-correlation issues.