Above I am subtracting the summed length of the center boxes used to create the voids from the overall length of the curve, calculating the residual length of wall and insetting the ends of the curves used to place the center boxes. I suspect the slight variation from ends to interior walls (end walls are slightly longer than interiors) arises from the substitution linear opening length for actual curvilinear length.
So I tried to use the curvilinear lengths of the openings to calculate the residual wall length. As I was kind of expecting, this created a circular data stream and broke the distribution.
Can anybody please suggest a way I might be able to accomplish this? Thank you!
Your method with subtracting boxes suggests A (boxes subtract whatever they subtract, but leave the same gap when measured along the wall centreline. Nothing is perpendicular to the wall faces except by chance). Personally, I would be trying for B (windows and gaps are all measured along the centreline, with cuts perpendicular to wall faces). but I suppose C might make sense too (inside and outside face of openings are the same width, with spacing to suit)
