With the traditional least-squares-calculation procedure, some measurements have great influence on the calculated fractional order number, whereas others contribute nothing to the final value. This is very unfortunate for the experimentalist, who would like to make full use of each measured value. To obtain the desired equal influence, the calculation procedure must be modified to use a fixed value of Δ(R2). This fixed value may come from the average of many adjacent spectral lines. Use of the modified procedure results in much greater precision without increasing the measurement effort.
David W. Steinhaus and John V. Kline, "Calculation of the fractional order number at the center of a Fabry-Perot interferometer-fringe system," J. Opt. Soc. Am. 65, 814-816 (1975)