Before quoting Mr. Mulherin I think you should know the stastic he shrouds in mystery: RMS stands for Root Mean Square It is a statistic used very much in electronics, and is calculated by taking the square of each data point and adding it to the square of the previous data point in a running total until all the squares of all the data points have been added up into a single sum. Divide this sum by the total number of data points, to get the Mean Square, and take the square root of that to get the Root Mean Square. Interferometric data points are simply obtained by creating an interferance pattern between the mirror under test and a "very accurate reference mirror".  Now for Mr. Mulherin's definition of RMS Wavefront Error:
"To obtain RMS Wavefront Error, a large number of interferometric data points are measured over the entire area of the test wavefront. The RMS error is a statistic that is calculated from all the data and gives a better indication of the system performance. Due to its statistical nature, professional optical shops consider the RMS Wavefront Error to be the most useful measure of optical quality. By common convention an optic with an RMS Wavefront Error of 0.0712 or less is considered diffraction limited."
Mr. Mulherin then goes on to define Strehl Ratios, and to compare their results.  I shall quote him on that on the next page before making any further comments of my own.