Howard and Simnett (HS) employed a new technique for associating LASCO CMEs to SMEI ICMEs. In order to extrapolate the SMEI data back to the LASCO field of view they used nonlinear trajectories, dependent on a speed and direction, what is more realistic than the linear extrapolation with only one parameter (a speed). However, there are two errors and one mistake in their procedure: (1) HS used two free parameters of the direction, whereas only one can be freely selected, because the second is provided by SMEI data. As a result, the directions determined by HS are incorrect. (2) HS overlooked that, since the trajectory depends on more than one parameter, there is a broad set of trajectories, for various speeds and directions, matching the event, and thus a broad range of the onset times. HS select only one trajectory for each SMEI event. Therefore the associations made by them are incomplete, and they should be reexamined. As long as it is not done any conclusion about CMEs undetected by solar coronagraphs are premature. (3) HS made some mistake in determination of the SMEI speeds. The speeds given in their Table 1 are about twice as high as those demanded to obtain the onset times given in the table. It explains why the SMEI speed distribution is excessively shifted toward high speeds ; hence, there is no reason to search for a physical explanation.