In the course of any of the operations, it is often necessary to transpose datasets. In particular, data which are represented in a multi-column ASCII format, will likely need to be transposed after being converted from ASCII to binary with a2b. The reason for this is that it is convenient to have the fast dimension of the data be the time (or depth) dimension for seismic purposes.
Our example above was a 5 column dataset, which you might think of as 5 seismic traces, with the same number of samples in each trace, side by side in the file. The processing sequence
% a2b < data.ascii n1=5 > data.binarywill result in a file with the fast dimension being in the trace number direction, rather than the number of samples direction The additional processing sequence
% transp < data.binary n1=5 > data.transpwill put the data in the desirable form of the fast dimension being in the number of samples direction, so that each trace is accessed successively.