Following engineering Apr 01-04, I had a 'pointing' shift on the evening of 04 Apr. data was taken despite the persistence of non-zero paz/pel (servo closure) errors. I attempted to account for this in the production of a new model, but data the following night were disastrously bad and I reverted to the old model on the 07th.

I had naively subtracted the paz/pel offsets from all recorded (uaz,uel) pairs, before solving for the new model, although the sense of the paz/pel errors cf uaz/uel is unknown to me.

Additionally, following the restoration of some order to the modelling, observed (collimation) offsets on UT20020408 : SCUBA(0,0), RxA(10,-3), RxB(53,-6) : revealed a sign convention error in the way azimuth collimations were being handled. Since 18 June 2001, and as annotated in /jac_sw/itsroot/src/tcs/params/collimation.dat I have corrected the collimations thus :

       az_coll_new = az_coll_old - uaz_observed
       el_coll_new = el_coll_old + uel_observed

The former sign is clearly incorrect and I have changed it and annotated accordingly.