Hi Frank:\

I flunked statistics in college.  It was the hardest course I ever took.  So
I've often wondered how a math guy like you would analyze the average or
mean or standard deviation or Root Mean square or whaterever you call it
errors in an EOT graph.

I came up with that 7 minute average a very long time ago.  If I remember
correctly, I just added up all the EOT values for every week of the year
(from an EOT table).  The negative values cancelled out the positive values
and the sum was 7 minutes.

I suppose you could also do it with thinking like this:
If at worst, a sundial reading is off by about 16 minutes,
And at best, it is off by 0 minutes,
Then on average, it is off by 16/2 =  8 minutes

It seems to make sense.

But I'm probably wrong.  Like I said- I flunked statistics!

John




-----Original Message-----
From: Frank King [mailto:[email protected]] 
Sent: Tuesday, February 15, 2011 5:04 AM
To: John Carmichael
Cc: 'Frank King'; [email protected]; [email protected]
Subject: Re: part 2 of longitude correction 

Dear John,

I looked into your interesting assertion about
the Equation of Time.  You say:

  On the average, it is only off [mean time]
  by about seven minutes...

It isn't really appropriate to use the term
average here because we are not dealing with
random variables or errors.  We are just looking
at a conversion scheme for going from one time
system to another.  That didn't stop me taking
a look and, essentially, agreeing with your
figure!

Three pedantic points first:

 1. The average of the Equation of Time is,
    of course, zero.

 2. The average absolute value is just
    over seven minutes and this is your
    figure.

 3. The RMS value [Root Mean Square]
    is probably more useful and that is
    about 8.75 minutes which is more
    than I expected intuitively.

Here are some intriguing figures in which
the left-hand column shows minutes and the
right-hand column show the percentage of the
time that the absolute value of the Equation
of Time is less than the associated number
of minutes:
               mins  %age
                 1    7.7
                 2   16.2
                 3   25.2
                 4   37.0
                 5   42.5
                 6   49.6
                 7   58.4
                 8   61.4
                 9   64.4
                10   67.9
                11   71.2
                12   73.4
                13   79.5
                14   86.3
                15   92.1
                16   95.6
                17  100.0

For example, is you pick a day of the year at
random, there is a 25.2% chance that your local
sun time is within 3 minutes of your local mean
time.

Expressed the other way we have:

                10%   1m 13s
                25%   2m 58s
                50%   6m 02s
                75%  11m 58s
                90%  14m 14s
                95%  15m 52s
               100%  16m 26s

I was surprised that the 75% figure was so
high, close to 12 minutes.

Maybe my calculations are wrong!

All the best

Frank

---------------------------------------------------
https://lists.uni-koeln.de/mailman/listinfo/sundial

Reply via email to