Does anyone know how Riblett actually calculated his GA-x mean lines? The GA-4 and GA-6 seem to match their description of being a NACA a=0.5 mean line. For the GA-2 and GA-3, however, he added 0.3% and 0.2% leading edge droop, respectively. Unfortunately, this droop means that interpolating mean lines gives odd results, and extrapolating (in my case to lower camber) gives really odd results. What I'd like to do is scale the NACA a=0.5 mean line directly, and then apply the nose droop myself (targetting ~13° leading edge slope, or equivalently a 0.25% chord mean line y-value of 0.060). But since I've been unable to reverse Riblett's process in this manner (that is, go from the GA-4 to the GA-2 by some transformation), I feel like I'm missing something.

So, what exactly does it mean arithmetically, in Riblett's context, to droop the leading edge x%?