A second possible source for the spread in the CL data is the procedure by which we obtain
the ball’s coordinates during the time camera 1 films. As figure 4 shows, we use a circle’s
centre to locate the ball’s centre of mass in each frame. The procedure for extracting the lift
coefficient is sensitive to the centre-of-mass coordinates obtained by camera 1. By ‘sensitive’,
we mean that the value of CL could easily change by a factor of 2 if one is not careful to align
our in-house software circle’s centre with the centre of the ball. Given that level of sensitivity,
we analyse a given camera 1 movie several times, each time using our in-house software to
obtain the ball’s centre-of-mass coordinates. We then average those sets of coordinates and
determine CL from the average set. The number of times we need to analyse a given movie
depends on how quickly the computed CL settles to a stable value. We usually need to average
three to eight sets of coordinates to get a stable CL. We feel this procedure helps eliminate
errors we introduced while trying to determine the location of the ball’s centre of mass in each
frame of the camera 1 movie.