In building the Zplane program, several interesting practical aspects arose. First, the program allows us to put a box on the data, and the Burg spectrum of the data in that box is computed and displayed. Thus the Burg computation of the reflection coefficients is a ratio of a numerator to a denominator, each of which is averaged in your selected box. Second, some traditional literature suggests that the only parameter you choose with the Burg spectrum is the filter length. After experimenting a while, I decided to keep the filter length at a constant 25, and instead let the variable be the corners of the estimation box that we draw on the data plane. Third, I found it necessary to bias the reflection coefficients downward as the lag approaches the data length.