A visibility dataset in MIRIAD logically consists of two parts -- a stream of variables and calibration tables. The stream of variables consists of parameters that are known at the time of the observation. They describe the changing state of the telescope. Variables include the observing frequency, the observing centre, the source name, ( u,v ) coordinates, the baseline number, the polarisation parameter being measured, and the actual measured correlations. The calibration tables are parameters that are derived after the observation. They are usually deduced from observations of a calibrator. Generally, MIRIAD \ tasks give a higher level view of the internals of a dataset, but it can be useful sometimes to examine these building blocks of the dataset. A list of all the variables and calibration tables is given in Appendix C.
Because source names, frequencies, observing centre etc., are stored as variables, MIRIAD does not distinguish between a single-source file and a file with multiple sources or frequencies, or a mosaiced observation. For example, a visibility dataset containing multiple sources does not differ structurally from one containing a single source. However, the variables source, ra and dec will change several times within a dataset which contains multiple sources, whereas they would remain constant in a dataset with only a single source.