Most of the seismic utility software at the Stanford Exploration Project handles seismic data as a rectangular lattice or ``cube'' of numbers. Each cube processing program appends to the history file for the cube. Preprocessors extend fortran to enable it to allocate memory at run time, to facilitate input and output of data cubes, and to facilitate self-documenting programs.
At Stanford Exploration Project (SEP) a library of subroutines known as ``seplib'' evolved for routine operations. These mostly handle data in the form of cubes, planes, and vectors. A cube is defined by 14 parameters with standard names and two files: one the data cube itself, and the other containing the 14 parameters and a history of the life of the cube as it passed through a sequence of cube processing programs. These cube processing programs are mostly written by researchers but several nonscientific cube programs have become highly developed and are widely shared. Altogether there is (1) a library of subroutines, (2) a library of main programs, (3) some naming conventions, and (4) a graphics library called vplot described by Dellinger . The subroutine library has good manual pages. The main programs rarely have manual pages, their documentation being supplied by the on-line self documentation that is extracted from the comments at the beginning of the source file. Below is a list of the names of popular main programs.
To use cube processing programs, read this document, and then for each command, read its on-line self documentation. To write cube processing programs, read the manual page for seplib and the subroutines mentioned there and here. To write vplot programs, see Dellinger's  documentation.