One of the big misunderstandings about icons is that they are 
	supposed to mean something. But this is a very tough area, and hasn't 
	worked out well, especially for actions. Look at "Blissymbolics" for 
	an interesting attempt from Germany in the 60s.

Blissymbolics has found a niche in "Bliss boards" for people with
cerebral palsy.  With a board of ~ 100 symbols you can "say" a lot
rather faster than you could by spelling.  That's a subset of the
full Blissymbolics system, of course.  Of course, full Blissymbolics
is NOT just a collection of icons (carefully designed to be easy to
draw with a limited set of templates and easy to distinguish); it is
a language.

As a matter of fact, I would rather LIKE icons that were Blissymbols.
They'd be A LOT easier to read than the cutesy-cutesy shaded "icons"
foisted on us these days.

	      We put them in because, for most people, they are more memorable 
	and searchable. Take a look at classic experiments by Haber, et. al. 
	about most people's visual memories, e.g., how long does it take when 
	clicking channels into the middle of a movie to recognize that you've 
	seen it before some time (maybe 20 years ago). How long does it take 
	to find the elephant from a wall of 100 randomly placed animals as 
	compared to words in the same places? (Ans. Most people can find the 
	picture 4 times faster.) A very interesting side factlet is that if 
	you simple draw a rectangular boundary around the 100 words, the 
	search is speeded up by a factor of 2 (you have "iconized" them).
	
The problem is that this only works for someone who knows what an elephant
looks like.  If you try the same experiment, this time with Mayan glyphs
(drawn at postage-stamp size), you'll find that people are **much** faster
with words in their own language.  And the icons we see on our screens
resemble Mayan glyphs *far* more than they resemble elephants.

Typical example:  the printer icon on my Mac is a sheet of paper with
ampersands on it, atop a low pedestal.  I suppose it is supposed to be
iconic for the ACT of printing.  But no part of it is even remotely visually
similar to my printer or what I see when printing is happening.  It is, in
short, NOT pictorial at all, but an arbitrary glyph in a collection of
arbitrary glyphs, and the association between it and printing is entirely
an arbitrary collection.  We're talking about semi-idiographic written
languages here.  If it didn't have "1st floor printer" in its label, I'd
have no idea what it was.  In fact, whenever I print I have to search for
the text.  Similarly, the icon on this UNIX box that *LOOKS* like a CD player
(or possibly some kind of oven) is in fact the printer icon.  It looks nothing
like any printer I've ever seen in the last 25 years.  It is an ARBITRARY
symbol in a sort of latter-day Chinese, without the multilingual usability
of Chinese.

	So the basic idea was to have them be, not meaningful per se, but 
	memorable and findable.

The problem is that you CAN'T find an icon memorable UNTIL YOU ALREADY KNOW IT.
As a Mac user, when you sit me in front of a Windows box I am confronted with
hundreds of visual symbols WHICH I CANNOT YET INTERPRET, and the "dictionary"
is constantly growing.

Let me give one more example.  If you wanted to produce high quality
documents, would you KNOW to look for a mountain and a lion?  If you
didn't know that, in what sense would the Alpha and OzTeX icons be
"findable" or "searchable"?

I don't know how I ever managed without DragThing.

	Bach particularly loved to show (mostly the 
	player, it's harder to pick up as a listener) the equivalent of 
	congruences and other similarities of structure. As an organist, it 
	is amazing the lengths he was able to go to here without compromising 
	the musical quality of the works to the outside listeners....
	
Some "Early Music" is full of numerical patterns, apparently.