Calculation of stacking or NMO velocity (see velocity) from measurements of normal moveout. In current usage, it generally involves finding the velocity associated with the best-fit hyperbola to common-midpoint data. Most analysis schemes assume a normal moveout, measure the coherency at that normal moveout, and then vary the normal moveout in order to maximize the coherency. However, even in the absence of noise and errors, time-offset data are not hyperbolic except in the constant-velocity case, and the stacking velocity value often depends somewhat on the amount of data included in the analysis. Where all reflectors are horizontal and where velocity varies only with depth, the stacking velocity is approximately the rms velocity. With good data, velocity analysis can yield depths to about 1/2%. Usually analyses are averages over 3–10 CMP and they may involve picking about 50 t-x pairs. Departure from a hyperbolic relationship is especially likely as offset distances become large. See Figures V-5 to V-7 and Sheriff and Geldart (1995, 303–313).