Re: re 23 59 12.0 +55 46 01

[Stupendous Man <richmond@stupendous.cis.rit.edu>]

  Earlier today, Tom mentioned that a recent scan through the
Mark IV database, picking out stars via the Welch-Stetson
technique, produced many objects with larger-than-expected scatter.
He wrote:

> Just for fun I processed the last 10 lines from vizier.list through VizieR.
> Of the 10 three were found using CGCVS and NSV.  Thay are:
> 2 CF Cas
> 3 V0821 Cas
> 7 HU Peg
> 
> Of the remainder about half looked variable to me.  One looks pretty good,
> 23 59 12.0 +55 46 01

  John Greaves pointed out that there is a problem with this
particular star:

> Someone check the field 'round STF 3049 C (this star) and tell him why he
> ought to go check his source image afore assuming variability.

  If you look at an image of this field, you will see a fourth-magnitude
star, BD+54 3082, sitting just 2 arcminutes away from the target,
a tenth-magnitude star.

  Most of the Mark IV measurements record the target as roughly tenth
magnitude.  However, there are a couple which are much brighter:
for example, on JD 2452911.70635, the target star is measured at
V=4.95.  Obviously, on this particular image, light from the 
nearby very bright star was mis-identified as coming from the target.
It is likely that on other occasions, _some_ of the bright star's
light contaminated the target star's measurement.

  Can we quantify this?  I made an ensemble of stars within
10 arcminutes of the target, and looked at their V-band
magnitudes.  If I mark the target star as "not a good member
of the ensemble", and use the others as references, I can find
mean magnitudes and scatter from the mean for each star.
I made a graph showing scatter from the mean ensemble magnitude
as a function of magnitude: look at

         http://spiff.rit.edu/richmond/temp/scatter.png

The star with very big scatter is the target star Tom
noticed.  Yes, it has big scatter, but that is due to contamination,
not real variability.  

  The other bright stars in the ensemble have a scatter
of about 0.02 - 0.04 mag, as you can see in the graph.
Two of these -- the brightest, at differential mag 0, 
and the fourth-brightest, at differential mag 1.3 --
are each about 5 arcminutes from the big bright BD+54 3082.

  John makes a very good suggestion:

> If yer methodology chucks out saturated stars maybes it ought to flag
> everything within a set arcmins radius thereof with a 'problematic' flag?

  Yes indeed.  We learn from this particular, isolated example
that stars within 2 arcmin of a fourth-magnitude star are likely
to have unreliable Mark IV measurements ... but stars at a distance
of 5 arcmin from a fourth-magnitude star are likely to be little
affected.

  SOMEONE could perform a valuable service by doing a little
digging, finding more examples of bright stars and fainter
stars nearby, and putting together a table like this:

      For a star of magnitude           expect contamination within
      -------------------------        ------------------------------
               0                              30 arcmin
               1                              12 arcmin
               4                               3 arcmin
         ....                              .....
     
and so forth.

  Volunteers, anyone?  

  I figure that it would take a day or so
for someone to work out a decent procedure, and then another
day to pick out 10 or 20 sample bright stars of a range
of magnitudes, and check the Mark IV measurements around them.
Well, maybe a couple of more days.  (Of course, since 1 day
of effort requires 10 calendar days, this could possibly
take a month or so of real time; less, perhaps, for someone
who can write scripts to access a database fluently....)

  Thanks to Tom and John for identifying this issue. 

                                      Michael Richmond