Postscript 2004.06.09

There is an on-line forum called the Minor Planet Mailing List, which is used by professionals and non-professionals in the astronomy world to communicate on matters relating to asteroids, comets, lost satellites, and similar topics. Following announcement of the C/2004K3 discovery, the following message was posted (by an individual whose identity will not be listed here):

This individual was critical of the fact that 3 of the 4 observations reported by Francisquito (observatory code G70) showed one decimal point less precision that the rest of the observations used in this particular discovery. I had intentionally reported one less decimal point of precision for the three “manual” coordinate picks described earlier in the article “Anatomy of a Comet Discovery”.

Before I had a chance to respond, Dr. Timothy Spahr of SAO posted the following message:

Thanks for taking enough interest to scrutinize these data. I think there is an interesting lesson to be learned here. As NEOCP object AO19841 (as it was known then) began to linger for a few days on the Confirmation Page, I was concerned that the rapidly brightening (and approaching) Moon combined with the object's track across a fairly dense region of the Milky Way, was going to prevent sufficient observations for the MPC to nail down a good orbit solution before all was lost to Moon glare. I had obtained a single measurement on 2004.05.31 (also published in MPEC 2004-L04) as the object passed a narrow gap between background stars; this measurement was made with the assistance of the astrometry software's "snap-to-centroid" capability, and so was recorded at the full precision offered by the software: 1/100 HA-second in RA, and 1/10 arc-sec for DEC. On 2004.06.01 I had three images of the object, but because it was passing very close to a group of field stars, the astrometry software's "snap-to" functionality couldn't cope with the narrow gap between the object and field stars, and wanted to snap to the field stars instead. I could get a manual readout of the object's position because the CCD images had already been solved, but I also knew that I was going to be "eyeballing" a centroid based on a pixel/sky scale of 2.0 arc-sec/pixel. I did my best to visually select the brightness centroid for each image, carefully recorded the position from the software's info. window, and then intentionally dropped one decimal place of precision for both RA and DEC. Of course, no photometry was obtained. What I'm saying here is that I didn't want to report any more precision than I felt could be justified by the means used to obtain the data.

Again, being concerned about a possible shortage of time and data, I placed all of these measurements in an e-mail message sent directly to Tim Spahr at the MPC, along with an explanation of my procedures and rationale. I heard back from Tim shortly thereafter and he informed me that the entire group of measurements formed a good data set. He also explained that selecting an object's centroid manually is sometimes the only feasible way to get a measurement. One then has to trust the specialists who evaluate these data to determine what is acceptable and useful, and what is not.

07/02/06