Analysis of Gravity Field — STRANGE, Woollard, and Rose 
385 
the fact that water, being incompressible, sup- 
ports a part of the overlying rock load, and 
thus the rock particles are not in as solid a 
contact as they might be if the water pressure 
were not present. In lavas the material is not 
made up of individual particles, but if fine 
cracks were present they might serve effectively 
to break the material up into individual parti- 
cles. The effect of both the presence of glass 
and the interstitial water pressure is to cause 
a lower velocity for a given density than one 
might otherwise expect. Both of these factors, 
therefore, probably influence the density- 
velocity correlation associated with lavas in a 
marine environment. 
The interstitial water could have another ef- 
fect also: It could help to explain the apparent 
retention of rather high porosities to great 
water depths. The effective pressure on the 
solid rock material at any depth would be 
(<r r -ov)h rather than o- r h, where o- r is rock 
density, <r w is water density, and h equals the 
thickness of overlying rock. Consider a point 
near the ocean floor, say 4 km below sea level, 
and material with a primary density of 2.3 g/cc. 
The difference in pressure under different as- 
sumptions would be: 
For a density of 2.3 g/cc and no interstitial 
water: 
Px = <x r h = 2.3 g/cc X 4 X 10 5 cm = 
9.2 X 10 5 g/cm 2 
For a density of 2.3 g/cc with interstitial 
water: 
P 2 = (ov - <r w )h = 1.27 g/cc X 4 X 
1(T 5 = 5.08 X 10 5 g/cm 2 
Or, if the rock is solid and <r r = 2.9 g/cc: 
P 3 = o- r h = 2.9 X 4 X 10 5 = 11.6 X 
10 5 g/cm 2 
The effect of pore water, therefore, would be 
to decrease appreciably the strength necessary 
to support the overlying material. Another fac- 
tor which could produce porosity in the vol- 
canic pile below sea level, and thus lower both 
the velocity and density, would be the presence 
of void spaces between successive flows of 
pillow lava. 
CONSTRUCTION OF A DENSITY MODEL 
FOR THE HAWAIIAN SWELL 
The most recent of previous interpretations 
of the gross gravity field of the Hawaiian Is- 
lands in terms of density models are those of 
Woollard (1954), Talwani et al. (1959), and 
Worzel and Harrison (1963). In the light of 
present knowledge none of these interpreta- 
tions is tenable any longer. The interpretation 
of Talwani et al. does not take into account 
the velocity structure of the Ridge, and assumes 
that materials with velocities which range from 
4.0 to 7.0 km/sec have a density of 2.84 g/cc. 
Moreover, this interpretation would make it im- 
possible to explain the local anomalies associ- 
ated with the volcanic pipes without assuming 
an unreasonably high density for the material 
in the pipes. 
The density model proposed by Woollard 
(1954), while reasonable at the time, must be 
rejected now on two bases. First, the boundary 
proposed in this model between the 2.3 and 
2.9 g/cc layers would lie in the middle of the 
5.0 km/sec seismic layer, and it is difficult to 
imagine a density change of 0.6 gm/cc with- 
out an attendant velocity change. Second, as 
pointed out earlier, in order to maintain a den- 
sity of 2.3 g/cc to the level of seabottom re- 
quires either that the dry density of the mate- 
rial be about 2.0 g/cc or that the material be 
nearly impermeable so that the voids are not 
water-filled. Neither of these conditions seems 
geologically probable. 
Taking into account all the information indi- 
cated in the previous section, a model repre- 
senting crustal structure and composition across 
the Swell was constructed and the two-dimen- 
sional gravity program of Talwani et al. (1959) 
was used to compute the gravitational effect 
of the model. Figure 3 is the free-air gravity 
anomaly map of the Hawaiian Islands. Figure 4 
presents the mass distribution model and a 
comparison of the observed and computed free- 
air gravity anomaly profiles along line A-A' 
of Figure 3. As seen, there is a reasonably 
good fit between observed and computed values. 
The densities of 2.95 g/cc and 3.40 g/cc for 
the main oceanic crustal layer and the mantle 
respectively were chosen on the basis of the 
density velocity studies summarized in Wool- 
lard (1962). The density of 2.6 g/cc for the 
upper layer of the oceanic crust was chosen on 
the assumption that this layer represents a com- 
