RVSAO is a set of programs to obtain redshifts and radial velocities from digital spectra. RVSAO operates in the IRAF environment. The heart of the system is xcsao, which implements the cross-con-elation method and is a direct descendant of the system built by Tonry & Davis. emsao uses intelligent heuristics to search for emission lines in spectra, and then fits them to obtain a redshift. sumspec shifts and sums spectra to build templates for cross-correlation. linespec builds synthetic spectra given a list of spectral lines. bcvcorr corrects velocities for the motion of the Earth. We discuss in detail the parameters necessary to run xcsao and emsao properly. We discuss the reliability and error associated with xcsao-derived redshifts. We develop an internal error estimator, and we show how large, stable surveys can be used to develop more accurate error estimators. We also develop a new methodology for building spectral templates for galaxy redshifts, using the new templates for the FAST spectrograph as an example. We show how to obtain correlation velocities using emission-line templates. Emission-line correlations ate substantially more efficient than the previous standard technique, automated emission-line fitting. Using this machinery, the blunder rate for redshift measurements can be kept near zero. the automation rate for FAST spectra is similar to 95%. We use emsao to measure the instrumental zero-point offset and instrumental stability of the Z-Machine and FAST spectrographs. We compare the use of RVSAO with new methods, which use singular value decomposition and chi(2) fitting techniques, and conclude that the methods we use are either equal or superior. We show that a two-dimensional spectral classification of galaxy spectra can be developed using our emission- and absorption-line templates as physically orthogonal basis vectors.
