Sed.Home | Alph | Intro | BasicClass | QFLClass | Keys | Depo.Envir | Evol | Evol.Model | Sed.Tect | Self Tests
  
Hypothesis Testing, and
Decision Trees
     There are two basic ways to learn to identify minerals and rocks. The Brute Force Method, "I'll just stare at the specimens so long maybe I will recognize one next time I see it." which produces wrong identifications more times than right, and the Hypothesis Testing Method which produces right answers most of the time.
     Many minerals and rocks once identified are easy to remember. A problem you may have is seeing another specimen which while it may be the same as something you already know, differs in some superficial way, like color. Therefore, even though you know a specimen when you see it, it is valuable each time you pick up a specimen to say to yourself, for example:

Hypothesis: This is the mineral galena, therefore, . . . Test: It will feel really heavy, have perfect cubic cleavage, and have a dull gray streak.

     Then, test the mineral for those physical properties. If the specimen has these required physical properties, then the identification is probably correct, even if the specimen is somewhat unusual.
     If, however, the specimen is missing one or more of these important properties, then your hypothesis is wrong. Do not try to identify the specimen as what you first thought it was, that identification will be wrong. You will have to create a new hypothesis for a different mineral or rock to test.

     An example of how the hypothesis-test method may prevent you from making a mistake is the mineral calcite. Calcite has rhombohedral cleavage, but when calcite grows in an open space it forms a six sided crystal with a pointed end looking superficially like a quartz crystal. If you were to see the six-sided crystal and say:

Hypothesis: This mineral is quartz, therefore . . . Test: it will be harder than glass and fracture but not cleave.

     Then, running just one of the tests (hardness vs. fracture, and absence of cleavage) tells you immediately the specimen is not quartz. Then you must observe, again, this time more carefully, to form another hypothesis to test.
     You must admit, it is better to know that you do not know what a mineral or rock is than to believe you know what it is, and get it wrong. To know that you do not know gives you a second chance to get it right.

     The habit of establishing a hypothesis (what mineral/rock you think it is) and then listing the characters you will observe if the hypothesis is correct trains your mind to think systematically about the properties of minerals and rocks. Furthermore, new or unusual specimens will not fool you.
     Sometimes it seems that an expert just simply identifies things, but that is not really true. What happens with an expert is that they are so familiar with what to look for, and what is important, that upon seeing a new specimen their mind automatically checks the specimen against a checklist of properties they have in their mind. The hypothesis and test are done almost instantly, and they may run through several hypothesis-test strategies before they finally tell you what they think it is. And as you become more expert you will do that too.




Decision Trees, and Hypothesis Testing
      What do you do with all those specimens that superficially look alike? For example, how do you distinguish among all the minerals that are clear to white? How do you distinguish among halite, calcite, dolomite, gypsum, colorless fluorite, kaolinite, quartz, sodic plagioclase, feldspar, and white orthoclase? At first they may all look alike.
     Or among igneous rocks, how to you distinguish among diorite, granodiorite, and plagiogranite? Or among sedimentary rocks, among chert, gypsum, micrite and dolomite?
     Or worse yet, how do you cross boundaries and separate coal, from basalt, from black micrite, from black shale, and from metamorphic rocks such as greenstone, slate, amphibolite and black marble?
     Now, some of these you just get to know by experience. But, I still encounter examples of familiar rocks that I have to look at more than once, and carefully test to be sure I have it right. And for someone just learning, all the similar looking minerals and rocks can just be confusing.

     Let's take a look at an example of hypothesis testing for each of the four categories. In each case we have chosen specimens that are frequently confusing to the novice, and shown a strategy for learning to distinguish among them.

MINERALS
     How do you tell the difference between gypsum, talc, and kaolinite, all three of which are whitish, and easily scratched by your fingernail? Kaolinite and talc are similar because they are both soft and easily scratched. However, in hand specimen kaolinite feels powdery and yields a powder when scratched. Talc, on the other hand, yields small, perfect, basal cleavage fragments when you scratch it. Gypsum does not feel slippery or greasy and also yields a powder when you scratch it.
     All this information is useful to you, but it needs to be well organized. Below is a decision tree which uses these characteristics to differentiate among these minerals.



     Want some practice? Take out examples of each of the following sets of minerals and devise a decision tree to identify them. Many variations are possible; the only question is, "Do they work for you?"
     How to proceed? Go back to the Mineral Identification Key and list the physical properties for each of the minerals. From the list of physical properties find ones that will allow you to distinguish first between two minerals. Then look for a physical property that will allow you to separate these two from the third. If there are more specimens, keep following this procedure. For particularly difficult specimens (i.e. very similar in many respects) you may have to make several tries with different properties to get some that work efficiently

A.   Fluorite, calcite, dolomite
B.   Calcic plagioclase, pryoxene, amphibile
C.   Olivine, epidote, apatite, chlorite, beryl
D.   Pyrite, chalcopyrite, magnetitie, specular hematite, earthy (red) hematite.


IGNEOUS ROCKS
     Four igneous rocks, let's look at obsidian, pumice, scoria, and vesicular basalt. All of these tend to be either glassy, or to be vesicular (full of holes). Plus, their colors can sometimes be confusing.
     For example, go to the classifiation and look at each of these, or better yet, get out samples of the specimens.
     Our tree is below:



     Want some practice? Take out examples of each of the following sets of igneous rocks and devise a decision tree to identify them. Many variations are possible; the only question is, "Do they work for you?"
     How to proceed? Go back to the Igneous Rock Identification Key (color/texture, or composition/texture) and list the physical properties for each of the rocks. From the list of properties find ones that will allow you to distinguish first between two rocks. Then look for a property that will allow you to separate these two from the third. If there are more specimens, keep following this procedure. For particularly difficult specimens (i.e. very similar in many respects) you may have to make several tries with different properties to get some that work efficiently

A.   Obsidian, basalt, peridotite
B.   Diorite, rhyolite prophyry, gramite
C.   Plagiogranite, granodiorite, diorite
D.   Syenite, quartz syenite, granite
E.   Granite, Syenite, monzonite
F.   Anorthosite, gabbro, peridotite



SEDIMENTARY ROCKS

     Want some practice? Take out examples of each of the following sets of minerals and devise a decision tree to identify them. Many variations are possible; the only question is, "Do they work for you?"
     How to proceed? Go back to the Mineral Identification Key and list the physical properties for each of the minerals. From the list of physical properties find ones that will allow you to distinguish first between two minerals. Then look for a physical property that will allow you to separate these two from the third. If there are more specimens, keep following this procedure. For particularly difficult specimens (i.e. very similar in many respects) you may have to make several tries with different properties to get some that work efficiently

A.   Fluorite, calcite, dolomite



METAMORPHIC ROCKS
     Want some practice? Take out examples of each of the following sets of minerals and devise a decision tree to identify them. Many variations are possible; the only question is, "Do they work for you?"
     How to proceed? Go back to the Mineral Identification Key and list the physical properties for each of the minerals. From the list of physical properties find ones that will allow you to distinguish first between two minerals. Then look for a physical property that will allow you to separate these two from the third. If there are more specimens, keep following this procedure. For particularly difficult specimens (i.e. very similar in many respects) you may have to make several tries with different properties to get some that work efficiently

A.   Fluorite, calcite, dolomite

    Return to Basic Key
    Pdf version of the Key

  

LSF Home | Geology Web Sites | Courses | Geology Home
Last Update: 10/26/00

e-mail: (Fichtels@jmu.edu)