328 
PACIFIC SCIENCE, Vol. XIX, July 1965 
such an origin. So also is the fact that the grav- 
ity highs that extend laterally as bulges from 
the highs beneath the calderas usually coincide 
closely with the surface trace of the rift zones 
(Kinoshita et ai., 1963: Fig. 1; Strange et 
al., Fig. 1, p. 382 in this issue). Whether 
the high value of gravity over the rift zones 
is due to exceptionally dense material at depth 
or whether it is wholly the result of a large 
number of dense dikes, if the path of the 
rising magma were inclined markedly to one 
side at shallow depth the gravity high should 
not only lie to one side of the rift zone, 
but should be only a fraction of the magnitude 
observed and of much greater width. The close 
coincidence of the "high” and the surface rift 
zone, therefore, strongly suggests continuation 
of the rift zone essentially vertically at depth. 
Furthermore, before any landslide or landslide- 
induced rift could exist, there must have been 
a volcanic mass for the slide to form upon. The 
east rift zone of Kilauea occupies the top of 
a gentle constructional arch that extends all 
the way to the sea floor, more than 100 km east 
of the summit of the shield. The arch has ob- 
viously been built by eruptions from the rift, 
and the shape of the shield is fundamentally 
governed by the position of the rift zone. The 
same is true of the other Hawaiian shields. The 
shields are generally not round or oval, but 
lobate, resembling three-pointed stars in ground 
plan (Wentworth and Macdonald, 1953), as a 
result of building by eruptions predominantly 
along the three rift zones. The close dependence 
in shape of the major shields upon the position 
of the rift zones indicates that the rifts are 
earlier-formed and are more fundamental struc- 
tures than any produced by landsliding. 
Displacement of the ground surface during 
the 1955 eruption on the east rift zone of Ki- 
lauea (Macdonald and Eaton, 1964) involved 
slight elevation of the surface adjacent to the 
eruptive fissures, sinking of a graben along 
the rift, and lateral displacement of the ground 
outside the graben by an amount up to a little 
more than 1.5 m in a direction essentially nor- 
mal to the rift. There was no detectable sink- 
ing of the surface outside the graben on either 
side of the rift in relation to the other side. 
The predominance of earthquake foci south 
of the rift zone noted by Moore is probably 
related to movement on faults of the Hilina 
system, which lies along the south slope of 
Kilauea and converges eastward with the rift 
zone (Stearns and Macdonald, 1946: Pi. 1), 
rather than to movement on the rift zone itself. 
Commonly, groups of earthquakes on the rift 
zone are quite distinct from those originating 
farther south, as is clearly shown on the map 
published by Koyanagi (1964). 
The rift zone pattern is closely similar to 
that resulting from distension of the surface of 
domes pushed up over intrusions (Cloos, 1955); 
and the most probable cause of Hawaiian rift 
zones still appears to be inflation of the volcano 
by intrusion of magma within it. 
CAUSES OF CALDERA COLLAPSE 
Two principal hypotheses have been ad- 
vanced to account for the sinking of the sum- 
mits of the Hawaiian shields to form calderas. 
Both depend upon the presence of a magma 
body of at least moderate size at a shallow depth 
beneath the summit area — a once- hypothetical 
magma body the actual existence of which now 
appears to be demonstrated. Williams (1941: 
246, 286-292) took the Hawaiian calderas as 
examples of his calderas of "Kilauean type,” 
which he believed resulted from removal of 
support caused by drainage of magma from 
beneath them, caused in turn by rapid effusion 
on the flanks of the volcano or intrusion as 
dikes or sills. Stearns and Macdonald (1946:33) 
suggested as an alternative explanation that the 
calderas resulted from weakening of the summit 
area by large-scale magmatic stoping and caul- 
dron subsidence as in the Scottish ring com- 
plexes — calderas classified by Williams (1941: 
246) as the "Glen Coe type.” The mechanism 
of their formation was believed to be the up- 
ward enlargement of the magma body by stop- 
ing and cauldron subsidence until the overlying 
crust became too thin and weak to support 
itself, when ring fractures developed and the 
surficial block sank into the underlying magma 
because of its greater density. The same mech- 
anism should operate if the enlargement of the 
underlying magma chamber was largely by 
melting of the enclosing rocks instead of by 
stoping. 
