[Rockhounds] Fw: SOFTWARE FOR INDENTIFYING MINERALS

[Rik Dillen]

I agree with Pete.
For relatively easy cases you don't need artificial intelligence... you need experience.
For difficult cases the type of expert systems we are talking about wouldn't work. Imagine that you would have to
differentiate between all types of greenish minerals from the vicinity of Schwaz, Austria ? It's even sometimes very
hard to identify them positively in the scanning electron microscope w/ EDX.
And what about all kinds of (sub-microscopic) intergrowths, solid solution cases, pseudomorphs etc. ?

Proper identification of minerals is a profession. Of course, some part of the mineralogical species can be recognized
with the naked eye (if well crystallized), but a large part can only be identified with some approximation and a limited
degree of certainty, if no heavy weapons (XRD, SEM/EDX/WDX...) are allowed :>)
It is e.g. quite impossible to recognize small grains of all kinds of sulphosalts without $$$$$ instruments.
It is also very difficult to put some data in an expert system that have to do with complex experience, such as knowing
that in dolomite of the Lengenbach Quarry (Switzerland), if your material comes from the southern part of the quarry,
the gray sulphosalts on specimens that contain realgar are almost always sartorite (skleroclase) or baumbauerite, almost
never dufrenoysite... this not being true for recently mined material from the central part of the quarry. How would you
put that in an expert system ?

One important criterion is the locality and the literature on it, but even that is not always enough.
During many decades zeolite crystals of a famous Swiss locality, Gibels Bach, near Fiesch, Wallis, have been described
as stilbite, even in professional literature and books (Parker's "Die Mineralien der Schweiz"), but a few years ago it
was proven that those crystals are in fact stellerite. There is no way to tell that by any expert system that only
contains the amateuristic means, without a petrographic microscope and/or XRD and/or SEM/EDX/WDX et al.

This is my biggest frustration in my own hobby. Collectors of postage stamps have a much more easy life. They can
recognize their specimens with knowledge and some very simple means...

But don't give up : if that is our only frustration we are lucky people.
Greetings,

Rik DILLEN 
Doornstraat 15,  B-9170 Sint-Gillis-Waas 
Belgium 
E-mail rik.dillen at skynet.be 
Homepage : http://users.skynet.be/rik.dillen 
 
MINERANT 2008  -  26-27 April 2008
Bouwcentrum (Antwerp Expo)
Jan Van Rijswijcklaan 191 Antwerpen 
http://www.minerant.org/mka/minerantnl.html 
Mineral collector's page http://www.minerant.org/ 


-----Original Message-----
From: rockhounds-bounces at lists.drizzle.com [mailto:rockhounds-bounces at lists.drizzle.com] On Behalf Of Pmodreski at aol.com
Sent: Wednesday, December 12, 2007 5:44 AM
To: rockhounds at lists.drizzle.com
Subject: Re: [Rockhounds] Fw: SOFTWARE FOR INDENTIFYING MINERALS

I hate to be a wet blanket, but I really think the ideas for this kind of  
software are doomed to failure, or at any event, would be of minimal use.
 
There are just too many minerals with very similar properties; too few  tests 
one can readily make, too many distinguishing features that require  advanced 
equipment to determine, and too many cases that will yield ambiguous  results 
or have a such large overlapping ranges of characteristics (variable  color, 
crystal shape, and so on).
 
One could have a chance at doing this if you limited the mineral base to a  
relatively small number of common minerals (100 or so?), but nature doesn't  
confine itself to those few minerals; or included the kind of sophisticated  
tests that an amateur could not normally do (optical properties; chemical  
composition; X-ray diffraction).
 
It's fine to talk about this, but in practice, I think it wouldn't have  much 
utility.  I don't think it would ever be a replacement for the  individual 
person learning about the properties and nature of minerals, and  using his 
educated experience to guide him to the most likely candidates, and  then refer to 
books or databases to narrow down the choices.  And then of  course, in many 
cases it will never be possible to make a positive i.d. without  using those 
advanced methods like XRD.  There are just too many, for  example, hard, 
refractory silicate minerals that aren't ammenable to any  simple tests for chemical 
composition; or, similar-appearing black metallic  sulfide and sulfosalt 
minerals.
 
But it's a still good intellectual exercise to plan out how one would  devise 
these databases and identification keys!  Or to dream about how  simple life 
would be, if they really worked to identify very many of those  "mystery 
minerals".
 
Just "IMHO"...   Pete