(my reply was originally posted to sci.astro on Jan 29, 1998)

  Gene Nygaard <gnygaard@crosby.ndak.net> asks angrily:

> ... Why
> don't you just use the SI units, metres, especially now that you have the
> prefixes up through yotta- which are sufficient to express any distance
> you could possibly measure with no more than 3 digits to the left of the
> decimal point.
>
> Before you give me any b.s. about the convenience of inverting the
> parallax angle in seconds to approximate this measure, remember that this
> only works in a limited range.        Yet astronomers use units such as
> megaparsecs, which are never obtained by inverting a measurement of
> parallax angle.

  I'll give you 4 reasons.

  A big reason which I'm sure has already been mentioned is:

      1. because the technical literature has been using parsecs for
         many decades, and we want to continue to refer to it

  Gene Nygaard notes, correctly, that trigonometric parallaxes are
measured only for a limited range of distances -- currently, out to
about 300 parsecs.  There are several proposed space interferometers
which (if successful) will increase this by several orders of magnitude,
but we still won't be measuring distances to the Virgo cluster via
direct parallax.  Nonetheless,

      2. since the parsec _is_ a convenient unit for _some_ distance
         measurements, and we don't want to use two different units,
         we prefer to stick with it even at distances well beyond
         the reach of direct trigonometric parallax

  But the new twist I'll introduce is the following: astronomers measure
the intrinsic luminosity of objects in "absolute magnitudes", which are
defined as "the magnitude the object would have if it were at a distance
of 10 parsecs."  

      3. if you replace the parsec, you must replace all "absolute
         magnitude" values, too

  Moreover, many observers use the unit "distance modulus" when discussing
distances to extragalactic objects.   If a star of KNOWN absolute
magnitude M is _observed_ to have an apparent magnitude m, then
the "distance modulus" is just that difference in apparent magnitude:

       (distance modulus) = (apparent mag) - (absolute mag B)

  For example, suppose you observe two Cepheid variable stars,
each pulsing with a period of 100 days.  From this identical
period, you deduce that both stars have the same intrinsic luminosity,
and hence the same absolute magnitude.  Let's say it's M=-5.

  However, you observe the _apparent_ magnitudes of those stars
to be m=15 and m=20.  How far away is each?

      absolute mag    apparent mag     distance modulus      distance
 ----------------------------------------------------------------------
        -5                15                20                 10 kpc
        -5                20                25                100 kpc

  Because of the way absolute magnitudes are defined, and the way
distance modulus is defined, there is a simple, easy-to-remember 
relationship between the two *if the distance is expressed in parsecs*:

                                  (distance modulus/5)
          distance in parsecs = 10

  Thus

distance modulus (mag)     0       5     10    15    20    25    30    35
----------------------------------------------------------------------------
distance in parsecs       10     100   1000   10^4  10^5  10^6  10^7  10^8 

 
  So my final rationale for keeping the parsec is because it meshes so
well with absolute magnitudes:

      4. if you replace the parsec, you lose the simple relationship
         between distance modulus and distance


  I maintain that astronomers should spend their time and energy
on tasks other than replacing parsecs by light years, or by meters,
or any other distance unit.