[daip] AIPS calibration ?: J2000 <--> B1950(VLA epoch)

[Samuel Conner]

To Whom it may concern,

   I need to convert J2000 coordinates into the special VLA epoch of the
B1950 system.  Why? you may ask. I'm recalibrating an old imaging survey
observed in B1950 (vla epoch) and I would like to improve the astrometry
(yes, it's that old).

  I have a code fragment extracted from an ancient version of OBSERVE
which purports to perform this transformation, but it gets the RAs wrong
by tens of milliarcseconds. Further it fails on the RA when I try to do a 
"round trip" B1950(epVla) --> J2000 -->B1950(epVla)

  So I am confident that this old code fragment is wrong.

   I have not had success in hunting for the correct transformation on the web. The VLA uses an unusual observation epoch (1979.9, I think) in its equinox B1950 positions, and the formulae I have seem out there (or pulled from the astronomical almanac, for example) fail when I test them with the VLA calibrator database.

So:  Is there anyone at NRAO who could point provide me with a working piece of code to do these transformations? 

 Thanks for your trouble.

Sincerely yours, 

Sam Conner