274 
must the rock material chosen satisfy the seis- 
mic velocity values noted, but it must also pro- 
vide a density contrast with the crust that will 
yield the observed change in crustal thickness 
with surface elevation and also satisfy the ob- 
served change in gravity values with elevation 
and isostasy. 
When one examines the relation between 
seismic velocity values and density values for 
different rocks, as determined in the laboratory 
under surface conditions and high confining 
pressures, the only rock material that appears 
to satisfy the above restrictions is a rock com- 
posed predominantly of olivine having a density 
of about 333 gm/cc. Eclogites have too high a 
density and too low a seismic velocity, and no 
other rock types having the requisite density, 
such as pyroxenite, appear to have the required 
high velocity. The companion velocity-density 
relations for the crust suggest a continental 
crust having a surficial layer with an observed 
density of 274 gm/cc (from 1158 samples of 
crystalline rock distributed over North Amer- 
ica) and an observed seismic velocity of 5.5 
km/sec at the surface increasing to 6.15 km/sec 
at a depth of 3 km because of the compressibil- 
ity effect on the modulus of rigidity with no 
appreciable change in density, and varying lin- 
earity through the relations for gabbroic rocks 
under pressure to mantle relations defined by a 
velocity of 8.15 km/sec and a density of 3.33 
gm/cc which correspond closely to the proper- 
ties of dunite under high pressure. 
This conclusion concerning the mantle agrees 
with that of Hess (1955, 1962) who, in a 
recent report (1964) on the serpentinite in the 
AMSOC core hole in Puerto Rico, concludes that 
the worldwide similarity of olivines, and of re- 
constituted serpentines to yield a similar chem- 
ical composition, provides a universal rock type 
that will satisfy not only the requirements of 
the mantle, but also that of a source rock pro- 
viding the structure and composition of the 
oceanic crust. Hess postulates that "layer 2” of 
the crust is basalt derived by magmatic differ- 
entiation from an olivine-rich mantle rock 
through volcanism, and that "layer 3” is ser- 
pentinite formed by the hydration of dunitic 
mantle material. 
There is, however, one problem connected 
PACIFIC SCIENCE, Vol. XIX, July 1965 
with Hess’ model; namely, how to get the re- 
quired crustal density stratification and observed 
seismic structure without having the "layer 2” 
as full of holes as Swiss cheese. The work of 
Moore of the U. S. Geological Survey (in press) 
shows that "fresh” submarine basalts increase 
in bulk density from 2.2 gm/cc at the surface 
to 2.9 gm/cc when emplaced under a hydrostatic 
head of 3000 ft, and that at oceanic depths of 
5 km the density is 3.0 gm/cc. This change in 
density with depth of water is due to the de- 
crease in vesicle porosity with confining pres- 
sure. Laboratory studies of the seismic veloci- 
ties associated with these whole rock basalts 
likewise indicate an incompatible velocity of 
about 6.6 km/sec. If the "layer 2” crustal layer 
is basalt, its low seismic velocity of 4.0-4. 5 
km/sec would require it to be now mostly ser- 
pentine or having a high porosity. The latter 
could be affected by having it emplaced under 
subaerial conditions or made up of pillows hav- 
ing sufficient inter-pillow voids to give the 
required low velocity. Normal basalts having a 
velocity of about 4.5 km/sec have a density of 
around 2.35 gm/cc. 
The basal crustal layer postulated by Hess to 
be serpentinite could well exist. Although only 
one serpentinite tested to date has the requisite 
velocity of about 6.5 km/sec and density of 2.8 
gm/cc under a confining pressure equivalent to 
about 1.5 kilobars, the fact that such material 
does occur (Birch, 1964) Is sufficient argument 
to prove that the hypothesis is not an unreason- 
able one. 
Using the velocity-density relations defined 
earlier (Woodard, 1962), the mean density of 
the crust on the continents based on seismic 
measurements is 2.86 gm/cc, giving a density 
contrast between the crust and mantle of 0.47 
gm/cc. In the Pacific Ocean the mean density 
of the crust is calculated to be 2.90 gm/cc. The 
Atlantic Ocean crustal data show two group- 
ings, 2.80 gm/cc and 2.90 gm/cc, with an 
average value similar to that found on the 
continents. 
The fact that these regional differences in 
derived crustal and mantle density will affect 
the thickness and depth of the mantle has al- 
ready been remarked in connection with the 
difference in crustal thickness observed in the 
