VOunits draft

[Anita M. S. Richards]

>
>> The document doesn't seem to address how magnitudes are to be handled. Are 
>> they unitless?  Do we have a 'mag' unit a la the FITS standard? What about 
>> magnitudes per square arcsecond? If they are unitless how do we find out 
>> that the column is a magnitude?  UCD? UType?  Do we allow SI prefixes?
>
> Or instrumental magnitudes versus absolute magnitudes.
>
> Perhaps our interpretation of the concept of "unit" has been warped by all 
> those junior high school science teachers who told us that every number has 
> to be accompanied by a unit.  So we tend to regard units as essential 
> metadata, when they are really just shorthand for the far more complex set of 
> data and metadata (filter transmission curves, light time corrections, etc.) 
> whose interpretation is required to pose proper scientific inferences based 
> on empirical observations.
>
> Note that we haven't been talking about very complex data the last few 
> messages.  A list of magnitudes vs. MJD is the simplest time series the VO 
> will ever see.

Thanks Rob,

On the MJD point, maybe we shoud stop pushing it, but yes there ar 
examples where that is all you get.  But actually the reason for it as an 
example in the first place was not because I really love it as a unit (I 
don't, because it is too easy to get the origin wrong, it was only 
invented to save numerical overflows before modern computers, I think). 
It was to show that we have to find a way of distinguishing between M (or 
any other SI prefix) as an SI prefix, and M as some other acronym - I 
could have chosen crab v. centi-rab as the example if there were rabs 
around!  Sorry!

On magnitudes, there is a separate working froup looking at photometry, 
who will hopefully come up with a standard for the minimum metadata needed 
to identify a magnitude system and so on, which Units should support. With 
detauls for old data which is just in 'B' with no explicit system, I 
guess.

So finally - actually magnitudes v. MJD is a nightmare because of the 
origin problems, I'd prefer Jy/arcsec**2 per JD any JD!

thanks

Anita