In a recent article, Bargatze et al. (2005) have identified why the implementation of the minimum variance analysis (MVAB) by Weimer et al. (2003), even though based on an erroneous variance matrix, has been successful in estimating the orientation of the "phase fronts" and the resulting propagation delays of the interplanetary magnetic field (IMF). They recommend further testing of the Weimer analysis as a space weather forecasting tool. In this comment we stress that the Weimer et al. implementation of MVAB closely mimics the results of a well-known version of MVAB that is constrained by the condition that the average field along the phase front normals is zero. This version of MVAB starts from the correct variance matrix, whereas the Weimer analysis is based on an unphysical matrix resulting from a programming error. We recommend that the constrained MVAB, originally developed by Sonnerup and Cahill (1968) and later recast into a more convenient form by A.V. Khrabrov be used instead. The Khrabrov method, which we refer to as MVAB-0, has been tested at the Earth's magnetopause by Sonnerup and Scheible (1998) and more recently by Haaland et al. (2004) and Sonnerup et al. (2004).