Page 6 - i1052-5173-29-2
P. 6

A                      Q                              B                   M


                                                   50
                                                   40
                                                   30 Color
                                                   20 index
                                                   10
                                                   0




                                                                              Q


                                                                                                             P
           A                                            P           A

          Figure 4. Modal data for 403 quartz-bearing plutonic rocks from the Sierra Nevada batholith, plotted on the quartz-alkali feldspar-plagioclase (QAP)
          triangle and in a tetrahedron whose base is QAP and whose apex M represents mafic minerals. It is clear that the proportion of mafic minerals (color
          index) increases dramatically away from the center of the QAP triangle, toward the QP sideline. This trend is neglected by the International Union of
          Geological Sciences classification. Data from Bateman et al. (1984) and Bateman et al. (1988).
          separation-by-name of closely related   of classification. Noting an analogy   described in the literature are covered by
          rocks is an unfortunate consequence of   between mineral assemblages and life   just one or two dozen. We compiled the
          fitting observations into arbitrarily   assemblages, he said:         number of citations in GeoRef mentioning
          defined boxes.                     How artificial a classification of faunas   19 of the most common IUGS names over
            Figure 3 also shows that the color indi-  would be which was based on the ratio of   the period 1970–2018. “Granite” made up
          ces of rocks called El Capitan Granite   foxes to hares, of hares to moles and so on!   just under half of the total number of cita-
          range from near zero to more than 20, an   To be sure it is by no means accidental that   tions (~223,000) in this list, and the names
          important point to which the name “gran-  the ratio of hares to foxes is 10 in a certain   were distributed as in Figure 5. The first
                                             area and only 2 in another, but as compared
          ite,” as defined by the IUGS system, gives   with the broad factors controlling life in the   nine names account for 90% of the cita-
          no indication. In general, color index   two areas it is relatively accidental… So it is   tions (allowing for multiple counting of
          increases moving away from the center of   not accidental that a rock is nearly pure oliv-  citations listing more than one name).
          the quartz-alkali feldspar-plagioclase   ine here and only 75 per cent olivine a few   As is common with textual data,
                                             feet distant, but it is relatively accidental and
          (QAP) triangle, toward plagioclase (Fig.   should not be made a fundamental factor    rock names roughly obey Zipf’s Law
          4). The abundances and identities of the   in classification.         (Aitchison et al., 2016), which states that
          mafic phases (e.g., biotite, hornblende,                              the frequency of a given word is inversely
          titanite) are typically where the geochem-  Bowen clearly recognized the problems   proportional to its frequency rank. There
          ical action is, and yet the IUGS classifica-  inherent in classifications based on min-  are ~50,000 rock names in the North
          tion has no provision for this.    eral proportions, and yet that is the sys-  American Volcanic and Intrusive Rock
            Thus, the current system of igneous rock   tem with which we live. We contend that
          nomenclature suffers from several short-  an improved system could address the
                                                                                             troctolite
          comings: (1) it conceals the variability of   shortcomings of the IUGS system by:  norite  clinopyroxenite,
                                                                                       monzonite
          igneous rocks at the map unit scale with   • using a more restricted set of names,   harzburgite  }  wehrlite,
                                                                                                       websterite,
                                                                                      syenite
          restrictive names; (2) it lacks important   because restricting the subdivision of   dunite    gabbronorite,
                                                                                   lherzolite
          information about the mineralogy of the   names allows each name to encompass   anorthosite     orthopyroxenite
          rocks at scales ranging from pluton to hand   greater modal variation;  tonalite
          sample; (3) it is burdened by unnecessary   • allowing for overlap between rock
          names that, by themselves, tell nothing;   categories; and          granodiorite                   granite
          and (4) the quantitative data used to derive   • including quantitative information about   gabbro
          a name are largely discarded once the   modal mineralogy at scales for which
          name is applied. This last point must be   quantification is appropriate.  diorite
          addressed if we are to move field observa-
          tions from “dark data” (Heidorn, 2008),   ROCK NAMES, NECESSARY               peridotite
          buried in field notebooks, into sharable   AND SUPERFLUOUS
          digital form (Walker et al., in press).  There are thousands of igneous rock   Figure 5. Frequency distribution of documents
            Bowen (1928) foresaw the problems   names (e.g., Johannsen, 1932, and accom-  using the given rock names in the GeoRef data-
          with using mineral abundances as a basis   panying volumes), but most samples   base, 1970–2018. The ten most common names
                                                                                account for more than 90% of the citations.
       6  GSA Today  |  February 2019
   1   2   3   4   5   6   7   8   9   10   11