Jökull - 01.12.1989, Qupperneq 73

Sphericity
Pig- 11. Regression plots for elongation versus
sphericity and OP-index. The pararrieter formulas
are given in the text. All 22 samples are included in
the plots.
Mynd 11. Línurit yfir tengsl ílengdar (sem er tvívíð-
ur þáttur, hlutfallið miðásllangás) annars vegar og
kýlni og OP - þáttar (þrívíðir þœttir) hins vegar. Öll
sýnin eru með á þessari mynd.
°n a separate regression plot (Fig. 11). The
regression was calculated with the MGLH module
of the SYSTAT package and the linear regression
plots were drawn with the GRAPH module. Table III
lists the results of the regression analysis.
CONCLUSIONS
A three dimensional analysis shows signiíicant
variation in the overall shape of tephra grains from
various eruptions. The variance or sorting of shape
parameters is also variable and may be used to
differentiate samples. There is a clear relationship
between the elongation of tephra grains and the
amount of Si02. This is probably related to magma
viscosity. Cluster analysis shows a grouping of his-
torical eruptions partly in agreement with the char-
acter of the eruptions (magmatic, phreatomagmatic).
The homogeneity of phreatomagmatically produced
tephra pointed out by Heiken ( 1972) and by Walker
and Croasdale (1972) may be tested by looking at
the variance values for elongation and OP-index.
The highest five elongation variance values are for
the samples from Kúðafljót, Öræfajökull 1362,
Skógar, H5 (light colour), and Askja 1875. The
lowest values were obtained for samples from
Heimaey, Surtsey, H4 (final phase), Snæfellsjökull,
Katla 1625, and Hekla 1693. Although our selection
of samples does not warrant environmental conclu-
sions to be made with any confidence, there are indi-
cations that the phreatomagmatic eruptions do pro-
duce measurably more homogeneous forms. How-
ever, the interesting fact that tephra of intermediate
composition (H5-dark, H4-dark, H1693) groups with
tephra from phreatomagmatic eruptions (Surtsey,
Katla) on Fig. 7 indicates that equant grains are also
very common in some plinian eruptions.
Several samples have to be analyzed from each
tephra layer or subunit after careful mapping to
obtain control on vertical and lateral variations
before environmental conclusions based on form
can be considered reliable. We suggest that a cluster
analysis may be a useful tool in identifying varia-
tions within eruptions as well as between eruptions
although the limited field control on our set of sam-
ples does not allow definite conclusions in this
respect. The clustering in Figures 6-8 is based on
very small differences that are not statistically
significant in some of the parameter values. In addi-
tion, atypical samples or "outliers" may interrupt the
cluster analyses. Special statistical tests are avail-
able to detect such anomalous "outliers", but we do
not feel that they would improve our results because
of the limitations of the data. Our results indicate
that a two dimensional analysis may reflect the vari-
ations in shape effectively although the presence of
particularly platy grains would not be registered.
This is very important because the analytical
JÖKULL, No. 39, 1989 71