Additional to 'noske's post 1915
BEA posted this summary of the sea search operations on Monday already, but I haven't seen it mentioned here before:
http://www.bea.aero/fr/enquetes/vol....16.05.2011.pdf
Most interesting in my opinion are fig.5 and fig. 6. One is a diagram of nine marker buoys and their drift paths, from an experiment made in June 2010. Look at those crazy loops! I guess that is what persuaded them to abandon any predictions about the crash site based on the location of floating debris.
The other is a map used for planning phase 4 of the search, indicating the most likely places to find the wreck. It is based on the assumption that most (but not all) areas already covered by sidescan sonar need not be searched again, and (in contrast to phase 3) that the pingers were probably just not working.
Anyone interested in the detailed analysis used to refine the search area should see:
http://wwz.ifremer.fr/lpo/layout/set...A9mentaire.pdf