Hawaiian Calderas— MACDONALD 
325 
of gas. Local erosional unconformities, stream 
gravels, and soil beds are found between sue- 
cessive flows, and the proportion of pyroclastic 
material increases. The spatter cones built by 
Hawaiian-type eruptions characteristic of the 
shield-building stage are largely replaced by 
cinder cones built by strombolian-type erup- 
tions. A relatively thin cap, a few tens to several 
hundreds of meters thick, is built over the top 
of the shield and the filled caldera. 
4. Extinction of the major volcano is followed 
by a long period of volcanic quiescence, with 
deep erosion and weathering. Sea cliffs several 
hundred meters high and canyons several hun- 
dred meters deep are cut into the volcanic 
mountain. 
5. At some volcanoes (whether eventually 
at all is unknown) volcanic activity returns. 
In this rejuvenated, or post-erosional, stage new 
eruptions take place from rift zones that have 
little or no relationship in position or direction 
to those of the earlier stages. Lava flows partly 
fill valleys and relatively small lava, cinder, and 
tuff cones are built against the dissected sur- 
face of the old mountain, separated from it by 
a profound erosional unconformity and some- 
times by intervening sedimentary deposits such 
as coral reefs. 
It should be noted that not all Hawaiian 
volcanoes have passed through all of the above 
stages. Lanai appears to have stopped activity 
at about the end of the caldera-filling stage. 
Kilauea and Mauna Loa are still In the caldera- 
filling stage and Hualalai appears to have barely 
entered the old-age stage. Neither Hualalai nor 
west Molokai now has any caldera, nor do they 
show any indication of ever having had one, 
though on both a caldera may have been hidden 
by later flows. On Mauna Kea the suggestion of 
a former caldera is very tenuous, consisting only 
in an arcuate arrangement of some of the cin- 
der cones in the summit region that may reflect 
caldera-related ring fractures at depth (Mac- 
donald, 1945). These volcanoes may have 
skipped the caldera-forming and caldera-filling 
stage. 
Recent petrographic studies (Macdonald and 
Katsura, 1964) have demonstrated the relation- 
ship of rock types to the eruptive history out- 
lined above. The lavas of the shield-building 
and most of the caldera-filling stages are tholei- 
itic basalt (including tholeiite, olivine tholeiite,. 
and ocean! te). In the upper part of the caldera- 
filling sequence, and at a corresponding strati- 
graphic level outside the caldera, there occurs 
a change to alkalic lavas (alkalic basalt, olivine 
basalt, ankaramite, hawaiite, mugearite, and 
trachyte), and these rock types persist through 
the old-age stage. The rocks of the post-erosional 
stage are characteristically nephelinites and 
melilite-nephelinites, with associated undersat- 
urated alkalic olivine basalts. 
Thus the formation of the calderas, though 
It may vary somewhat from one volcano to an- 
other and may be lacking from some, takes 
place during the end of the period of shield 
building, while volcanism Is still vigorous, and 
somewhat before the slowing down of volcan- 
ism that accompanies the change of petro- 
graphic types from tholeiitic to alkalic. 
One important bit of evidence that should 
be considered in relation to the origin of the 
Hawaiian calderas is that the caldera floor 
sometimes moves upward as well as downward. 
Not only does the floor swell up as a broad flat 
dome during tumescence of the volcano (Jag- 
gar and Finch, 1929), but it njay move en 
masse in the manner of a huge piston. The 
great collapse of Kilauea caldera In 1840 pro- 
duced a central pit 1.5 km wide, 3 km long, 
and about 150 m deep. About 1845 the floor 
of the pit started to rise bodily, and by 1850 
it had risen about 150 m and was approximately 
level with the "black ledge” that surrounded it. 
The taluses that had accumulated on it at the 
foot of the bounding cliffs had been pushed up 
until they stood as arcuate ridges of angular 
rock fragments that projected as much as 45 m 
above the surrounding caldera floor. Although 
it Is possible that this elevation of the caldera 
floor could have resulted from shallow intru- 
sion of magma beneath it, as did the elevation 
of the floor of Halemaumau crater in 1952 
(Macdonald, 1955^), it appears more probable 
that it was caused by inflation of the underlying 
main magma chamber of the volcano. 
INTERNAL STRUCTURE OF THE VOLCANOES 
In none of the Hawaiian volcanoes has ero- 
sion cut deeply enough to expose the congealed 
