Kilauea Iki, 1959 — Macdonald and Katsura 
DIFFERENTIATION IN KILAUEA IKI LAVA 
In Figure 1 the analyses of samples of Kilauea 
Iki lava are plotted on a standard AFM diagram, 
together with those of 85 other tholeiitic rocks 
of the Hawaiian Islands. The general tholeiitic 
trend of the Kilauea Iki samples is apparent. 
The trends of both the Kilauea Iki rocks and 
of the tholeiitic rocks in general head directly 
toward the position of olivine (Fogo) on the 
FeO-MgO join, and it is clear that compositional 
variations in the suite as a whole, including the 
Kilauea Iki rocks, can be accounted for largely 
by variations in the amount of olivine sub- 
tracted from, or added to, a parent magma with 
a composition lying along the same trend line. 
Although movement of olivine was the main 
control in the differentiation, calculations indi- 
cate that minor amounts of other material also 
are involved. Column 2 of Table 3 shows the 
composition of the least amount of material that 
must be added to magma of the composition of 
the Pele’s hair (Table 1, column 1) to produce 
a rock of the composition of the picrite-basalt 
from a depth of 7.5 ft. in the drill hole (Table 
1, column 4). The material is preponderantly 
olivine, but includes also calcic plagioclase, py- 
roxene, and iron ore. The fact that the plagio- 
clase is wholly anorthite indicates that some 
amount of material greater than the least pos- 
sible actually has been added, since the first 
plagioclase to separate from a magma of the 
composition of the Pele’s hair would have a 
more sodic composition (about An 85 ). The 
pyroxene also would be slightly poorer in Ca 
and increased in amount, but olivine would re- 
main by far the most abundant component. 
Similar conclusions have been reached by Muir 
and Tilley (1957) regarding the 1840 picrite- 
basalt of Kilauea. 
Similar results can be obtained by assuming a 
liquid phase of constant chemical composition 
like that of the Pele’s hair and adding to it 
arbitrary amounts of olivine of the composition 
of that in the picrite-basalt (Table 1, analysis 
11). Table 2 shows the results of these calcula- 
tions compared to the actual compositions de- 
termined by analysis. The lower line indicates 
the amount of olivine added in each case, cal- 
culated on the basis of the amount of MgO and 
FeO in both the olivine and the Pele’s hair. Since 
365 
this method of calculation of necessity results in 
identical values of FeO and MgO in the analyzed 
and calculated materials, the figures are insignif- 
icant in the present connection and are omitted 
from the table. Note that for the most part the 
calculated compositions are quite close to the 
actual ones, again indicating that the principal 
substance added during the differentiation was 
olivine. In the case of sample 9 the agreement 
is somewhat less good, especially in the CaO 
content, suggesting that other factors than the 
addition of olivine have been involved. It is im- 
possible to calculate sample 10 on the same 
basis, because the amount of olivine becomes 
negative— that is, olivine must be removed 
from, not added to, the Pele’s hair to yield sample 
10 . 
The composition of the liquid phase of the 
magma during the beginning of the eruption is 
represented by the analysis of Pele’s hair ( Table 
1, column 1). However, even at that time the 
bulk composition of the magma was somewhat 
more mafic, due to the presence of olivine phe- 
nocrysts and possibly other crystals not con- 
tained in the analyzed sample. This bulk compo- 
sition is probably better represented by analyses 
S-l or S-2. These in turn are less mafic than the 
average magma extruded later in the eruption. 
Richter and Eaton (I960) note a general ten- 
dency for the lava of late stages to be a little 
more mafic than that of the early stages. The 
best approximation to the composition of the 
"parent magma” for the suite of drill-hole 
samples may therefore be an average of analyses 
2 and 3. This average is given in column 1 of 
Table 3. 
Whereas samples 4 to 7 are probably cumula- 
tive types, derived by addition to the parent 
magma of sinking crystals, largely olivine, sam- 
ples 8 to 10 are probably residual magma from 
which crystals have been removed. In Figure 1 
they lie along the same trend, but on the other 
side of the parent magma (samples 2 and 3) 
from the cumulative types. Their position on 
the same trend line suggests that they have been 
derived largely by the same general process as 
the other Kilauea Iki rocks and the tholeiitic 
rocks in general. However, in Figure 2 samples 
9 and 10 lie along a line that deviates markedly 
from the general tholeiitic trend. Sample 10 in 
