but I'm not comfortable with the statement that it cannot be proved by measuring physical triangles, which is what your original comment implied to me.

It's just the problem of induction.