The user provides SCHED with source coordinates in one of the J2000, B1950, or DATE coordinate systems. SCHED then determines the source coordinates in the other systems. The J2000 and B1950 systems are not stationary with respect to each other. To do a proper conversion, it is necessary to specify for which date to do the conversion. This date is specified using PRECDATE. When SCHED converts from, for example, B1950 to J2000 coordinates, it uses the B1950 equations to precess to PRECDATE, and then the J2000 equations to precess from there to equinox 2000. If the wrong date, by many years, is used, the errors in the conversions will be a few tenths of an arc second. This is enough to harm phasing of the VLA at high frequencies and enough to be rather undesirable for positions used on the correlator.
The correct PRECDATE to use depends on various factors. If the target source coordinates are being provided in the J2000 system, as they really should be, then it doesn't matter much. The correlator and the VLA will be given the J2000 positions so no conversion is involved. Some antennas will be given B1950 coordinates for pointing, but for single dishes, the few tenths of an arcsecond differences that can be caused by different PRECDATEs are not important.
The more important case is when the user has a B1950 coordinate for the target source and wants SCHED to take care of the conversion to J2000. Then the correct PRECDATE should be used. The correct date depends on how the B1950 coordinates were measured. If they were determined based on absolute measurements made on some date, that is the date that should be used. However, most target source positions will be based on measurements of offsets from some calibrator. Then the date that should be used depends on the nature of the calibrator positions. For example, the B1950 coordinates of the VLA calibrators are either based on absolute measurements made in 1979.9, or are converted from more modern J2000 coordinates using an assumed ``observe'' date of 1979.9. Therefore, any B1950 coordinates determined on the VLA using the B1950 coordinates of sources in the VLA calibrator list should be precessed with a PRECDATE of 1979.9, which is now the default in SCHED. Note that the 1979.9 date is hard-wired into the VLA scheduling program OBSERVE and is not visible to the user.
Other systems may treat B1950 coordinates in a manner similer to the VLA, but with a different reference date. An obvious choice is 1950.0, which is used by MERLIN and is the default in the STARLINK COCO precession routine. On MERLIN, The calibrator positions presumably are all based on measured positions originally determined in J2000 (by VLBI etc), then converted to the B1950 coordinates that are required by the on-line system. Those conversions have all been done assuming an ``observe'' date of 1950.0. Thus, if you have B1950 source positions determined by offsets from a calibrator using MERLIN, PRECDATE=1950.0 should be used.
If anyone knows what other systems of general interest do, let Craig Walker know and a section can be added here.
The default PRECDATE in SCHED has changed a few times over the years. Prior to 5 Aug 1997, a date of 1985.0 was hardwired into the program. After that, the user parameter was introduced with a default of 1997.0. On 6 May 1998, the default in the development version of the code, and hence the next release (unknown date at this writing) was changed to 1979.9, the VLA value.