Robin, I just came up with a thought experiment that lends support to the idea that a reactionless drive is not likely to exist. Take 2 different observers, one that is moving beside the ship at the same velocity as it has prior to activating the drive. The second one is moving at a velocity that allows him to observe the ship decelerate first until it reaches a velocity of zero relative to him and then to accelerate in the reverse direction until it reaches the exact same original velocity in the opposite direction.
The first observer sees the velocity of the ship go from for this example 0 meters per second to 2 meters per second. He determines that the ship now has 2*2*Mass/2 units of kinetic energy. The amount of internal mass that the ship burns up to achieve this acceleration is extremely small and can almost be neglected. The second observer sees the ship moving at the same speed before and then after the application of the drive. The only difference he measures is that the direction of the motion of the ship is reversed by the drive. So he sees the ship begin the motion moving 1 meter per second relative to him initially and then after the drive shuts down the ship is moving 1 meter per second in the opposite direction. This observer determines that the kinetic energy of the ship has not changed measurably due to the application of the drive. At this low velocity the second observer determines that the mass converted into drive power is essentially the same as that determined by the first observer. Both guys have a very hard time figuring out exactly how much mass is converted, and they agree that any difference is hidden in the noise. In this experiment we have two independent observers seeing the same ship being subject to the same drive. The amount of kinetic energy being deposited to the ship by essentially the same loss of internal mass varies remarkably according to each. This does not add up. As I have mentioned before, with a normal drive this case can be handled without a problem. The exhaust material supplies the kinetic energy and momentum needed to balance the equation. Apparently every observer moving at a different initial velocity relative to the driven ship arrives at a significantly different calculation regarding the ships energy balance after the application of the drive. How could something this radical be possible? Let me say it again, there is no problem of this sort to deal with when a standard drive is applied. Dave