Re: GSC 540-84

[Arne Henden <aah@nofs.navy.mil>]

Michael Koppelman wrote:
> The displacement I was talking about (and it wasn't clear to me if you 
> were talking about the same displacement) was how the minimum was smooth 
> on some cycles and rough on others. I'm pretty sure, and others have 
> seemed to agree, that this represents an additional fourier component 
> and not errors on my part.
> 
Fourier components are just that: coefficients on periodic terms in the
Fourier expansion.  As such, they need to be present at all points in
the light curve, and in general need to be present on all cycles.
On the other hand, we may have been referring to different features,
and I've now deleted the original email and so cannot check!

> I know that scientists are loath to assume and that amateurs (or more 
> accurately myself) are probably too willing to look for interesting 
> features where they are not. I see the wisdom of this and I am heading 
> that way. At the same time, I think great advances in science are made 
> when wild speculation leads people to study things that more prudent 
> people would dismiss. My point here is only that I intend to continue to 
> wildly speculate because it is creativity that ultimately leads to good 
> science. Lest this seem ungrateful, I am very grateful that Arne and 
> others keep my feet on the ground. There is a big difference between 
> wild speculation when it doesn't matter (like on an email list) as 
> opposed to when it does matter (as in publication).
> 

I have no complaints with "wild speculation", and never recommend someone
remove data when they are discrepant just for that reason:  they may
represent some heretofore unrecognized feature of a particular object.
However, for things like eclipses in an RR Lyr star, you are advanced
enough in your studies to be able to do back-of-the-envelope calculations
to determine whether such a speculation has any chance of occurring.
My rule of thumb is to use Occam's Razor and choose the simplest
theory that fits the data.  In this case, poor photometry, statistical
fluke or else something related to the pulsation, since locking an
eclipse to the exact pulsational period just doesn't seem likely.
As you said, however, the obvious solution is to get more data!
Arne