Lot of discussions on this...
Cane Pole is right. The linear (straight line) distance between the transducer and GPS receiver can make it harder to pinpoint that bit of structure you marked with a waypoint sometime earlier. You won't notice it if you are travelling the same direction when you are looking for it as when you marked the waypoint. Otherwise that 10 foot distance difference between the transducer and GPS receiver could put your transducer up to 20 feet away from the same structure if you were to looking for it while travelling in the direct opposite direction even though the GPS receiver shows you as being dead on the same coordinates. For the best possible repeatable accuracy the GPS receiver should be positioned over the transducer. This is especially true for any waypoint that was made while referencing sonar data. Now I wouldn't go mounting the GPS receiver in some location that would not allow it to find and lock on to all available satellites as this could cause it to be more inaccurate which could cause a difference of more than 10 or 20 feet.
Not all GPS-Sonar users will notice this difference in accuracy though. Some of this gets back to how we each use our GPS and sonar units, how the actual waypoint was marked, and the true positional accuracy of the GPS unit both when you marked the waypoint and when you were looking for it again.