"Those are probably not conics but splines or even better NURBS curves. You can repersent a correct conic curve with a rational bezier curve of degree 2 but you need higher degree to obtain good cuvature continuity and an aesthetically pleasing connection with the straight line. Also I suspect they just dimensioned  points on the curve measured at a regular interval, in a way that every interpolation you could build on them will have a desired error with respect to the actual curve used. It's a common practice in optical design for example."