VirgoA is a double source at 850um

The above is a composite 850um contour map of Virgo A comprising 7 POINTINGs (9-point jiggle maps) taken between 08 Dec & 17 Jan. Noisy bolometers have created a lot of confusing structure outside the central part of the image.

A secondary peak is seen 16" East and 13" South (21" total) from the main peak, and is ~20% of the peak in strength. A centroiding scheme that assigns equal weights to all components of such a source would put the centroid some 3" SE of the main source. This is an error we could certainly ill-afford.

However, the KAPPA CENTROID program clearly does much better than this, essentially locating just the main peak :

------------
INIT - Initial position /'0,0'/ > 12:30:49.5,12:23:25

Title: Feature positions in virgoa

     Centroid positions:  RA(hh:mm:ss.s) Dec(ddd:mm:ss)

1:   12:30:49.4 12:23:25
------------

Provided the online centroiding scheme is similarly discriminating, the usefulness of VirgoA as a pointing source appears uncompromised.

Tim Jenness adds :

"The online centroiding algorithm is an unknown quantity. . . . The kappa centroid seems to be fairly robust and that is the one used by oracdr. All I can say is that the observer monitors the offsets generated by oracdr and compares them to the online. Be very suspicious if they differ by more than 0.5 arcsec."

From the data of UT 20001208 and 20010108 :

20001208 #  sou   online (uaz,uel)  
        70  3c273   -4.5, 1.2  
        71  virgoA  -1.2, 4.4  
                   ------------
   online change     3.3, 3,2  
   orac change       3.7, 2.0    i.e. 'good' agreement
  

20010108 #  sou   online (uaz,uel)  
        86  3c273   -6.9, 4.5
        90  VirgoA  -5.0, 7.2
                   ------------
   online change     1.9, 2.7  
   orac change       1.7, 3.1   i.e. 'good' agreement

20010108 #  sou   online (uaz,uel)  
       118  Mars    -4.1, 4.8
       120  VirgoA  -6.7,-0.8
                   ------------
   online change    -2.6,-5.6
   orac change      -1.5,-1.5    i.e. 'poor' agreement

where 'good' means <1" or so (cruder than Tim's 0.5").

2 'good' agreements from 3 is enough to allow me to conclude that there's no evidence of consistent error in pointing due to the secondary peak in VirgoA. Whether this is due to good fortune or the robustness of the on-line centroiding is, I guess, unclear.