I think the issue that  Frank King raises below is true but the effect is 
trivial for most applications.  The elliptical distortion is imperceptable 
exept 
when the sun is at the extrememly low altitudes of sunrise and sunset.  Nodus 
dials rarely function at these times anyway because 1) the projected image is 
thrown far off of the dial surface and 2) John Shepard's observations from his 
work on the (University of Wisconsin's) Swenson Dial about blurring of the 
nodus shadow when projected over large distances.

I still believe that a flat disc (with or without a central aperture), 
parallel to the dial surface, is in most cases an excellent design, superior to 
the 
more commonly used sphere nodus or a point nodus.  

-Bill Gottesman

---------
Subj:    Re: Turtle Bay Sundial Bridge opens 
Date:   7/13/2004 4:09:51 AM Eastern Daylight Time
From:   [EMAIL PROTECTED] (Frank King)
Sender: [EMAIL PROTECTED]
Reply-to:   <A HREF="mailto:[EMAIL PROTECTED]">[EMAIL PROTECTED]
</A> (Frank King)
To: [EMAIL PROTECTED] (john shepherd)
CC: [email protected] (Sundial List), [EMAIL PROTECTED]

Hi John,

At last signs of the truth...

> Theoretically it's correct that the projection of a circular
> disc on to a flat surface parallel to the disc will be a circle. 
> Unfortunately the sun's apparent size results in the disc becoming 
> very blurred when you get a couple of hours off of local noon.

The theory suggesting that a circular disc casts a circular shadow
depends on the sun being a point source of light which is isn't!

If the model on which the theory is based is refined to take
the angular diameter of the sun into account, you find that
the shadow is generally degraded into an approximate ellipse
whose major axis is less than the diameter of the disc and
whose minor axis is smaller still.

The actual shape at a given time can be determined by noting
the shape that the image of the sun that would form if the sun
were projected through a pin-hole at the centre of the nodus
(this image really is a true ellipse).

You then draw the circular shadow that the simple theory
suggests and at each point on the rim you draw this ellipse,
being careful to preserve its orientation.  You then get
two envelopes, the inner of which is a fair approximation
to the true shape of the shadow.

The inverse of this effect occurs with an aperture nodus.
The anti-shadow of a circular hole likewise distorts into
an approximate ellipse but its major axis coincides with
the minor axis of the shadow of the surrounding disc and
vice versa.  Of course the anti-shadow is bigger than the
original hole.  You have to use the outer envelope.

Unless a nodus designer understands all this, it is ever so
easy for the anti-shadow from the hole to exceed the size of
the shadow of the surrounding disc.  The result is indeed
pretty useless! 

Frank King
Cambridge University
England
-

Reply via email to