[Enchilada]

Originally Posted by bojan

Apparently, Hipparcos did not very well with double stars (spectroscopic and close binaries).
This is became pretty obvious for me when I was playing with Celestia.. I wanted to see what Pleiades look like when viewed from side.. so I went to the same distance from Sol and looked at them and the cluster was grossly elongated along the line of sight (towards Sol), something I did not expect but I immediately concluded something went very wrong with measurement accuracy..
Hope Gaia will do better...

Actually this is only partly true. Hipparcos design involved measures by interferometry. The problem is separations was worst at around 0.7 arcsec, and poor up to 2 arcsec. Anything above 2.5 arcsec has no difficulties in accuracies averaged at about 0.003 arcsec. Several doubles below 0.3 arcsec have similar accuracies - with error between 0.5% to 1% - better than visual observations. These errors are also complicated by difference in magnitude (Delta-m) between the components. Also the proximity of the star made defining the photometric centres problematic, hence the errors in the parallaxes.
The reasons for the cluster is because of significant systematic errors in the observations. In case of the Pleiades , if you look at the error bars. We know the RA and Dec very well, parallaxes are much worst. Still better accuracy than what we had before it by a factor of ten or more at least.