Date: Wed, 15 Mar 2006 12:56:01 -0600
From: Richard Huziak <huziak
	rv:1.0.2) Gecko/20030208 Netscape/7.02
References: <000801c647cd$3c90ace0$6401a8c0
	<44176F42.7090004

Thank you for saying this, Mike.  Consistency of CCD measurement needs 
solid datapoint needs to be statistical.  Minimum number is usually 
stated as statistically relevent is 5 measurements (images).  This 
allows single hit error sources (cosmic rays, poor flatting, glare, 
scintillation, hot pixels, etc) to be weeded out or averaged out, and as 
Mike states, a relevent error to be calculated using a simple formula.   
 If you look at CCD light curves made from single hits, often the 
observations, somewhat misleading in that everyone 'assumes' CCD to be 
more accurate.  

averaged datapoint (unless you are specifically asked to stack and 
average).  Ones that are obviously bad (cosmic ray hits, etc) can be 
manually rejected before submittal (though you shouldn't bias you data 
by 'making' it look good by excluding points if you cannot identify the 

rick (HUZ)

Michael Koppelman wrote:

> We probably don't want to start another monstrous debate, but we can  
> simplify this pretty far:
>
> 1. Take at least 3 observations.
> 2. Take the standard deviation of a comp star of about the same  
> magnitude as the variable star and report that as the error.
>
> I know some people will take a single observation with a CCD. I've  
> never done this. I always take at least 3 and usually around 8  
> observations. If you do this, calculating the uncertainty is as easy  
> as the two step process above and is more inclusive of various  
>
> Michael
>
>
> On Mar 14, 2006, at 8:37 PM, Aaron Price wrote:
>
>> ...which scare away the average observer out of fear that error
>> reporting will take the fun out of it.
>
>
> _______________________________________________
>