The data relate to how the solar heat gain coefficient varies with angle.
A sigmoidal model may fit the data given well but there is little physical
reason to use that function, articles on the subject tend to use linear
transformations of cos(A) eg see (3) in
http://gaia.lbl.gov/btech/papers/37747.pdf

i don't agree there is anything dangerous about either formula. There is
definitely a decreasing relation for 0<A<90 and so we can make use of a
simple formula. Looking through the archives, nearly all interpolation
examples posted have had a 1-to-1 relation either increasing or decreasing -
those that dont are generally better estimated by other means.
(This was discussed in another post i recall).

Extrapolation, however, can be very misleading when based on only two
data points, much better to find a suitable regression model, as here, if
you're
needing to estimate values outside the data range.



"Bernd P" wrote in message
...
Hello,

Lori's formula only works if values in A are increasing and if values
in B are decreasing (if B values are increasing we could drop the
"1-"). And its not extrapolating (ok, that was not asked for).

IMHO Lori's approach as well as Joe's one are outright dangerous,
Herbert's is a perfect 110% solution for this special case but he
documented his approach very well so that we can apply it generally, I
think.

Regards,
Bernd