SECT. 1] THE CRUSTAL ROCKS 99 



Other evidence suggests an igneous origin. For example, it appears that 

 Layer 2 is thicker near some groups of Pacific islands and seamounts (Raitt, 

 1957), giving support to a volcanic origin at least in some areas. The high 

 values, 5.99 km/sec, east of Hawaii are characteristic of a granitic continental 

 crust and may be intrusive, plutonic rocks. It has been suggested by Ewing 

 and Ewing (1959) that Layer 2 is the upper part of Layer 3, forming a transition 

 layer of increasing velocity with depth, presumably the result of the sealing of 

 holes and fissures by the increasing pressure of the overburden. 



It would be surprising, in view of the great variety of rocks forming similar 

 layers beneath the continent, to find that Layer 2 is monotonously uniform 

 under the oceans. It seems likely that it is a variety of rocks. The wide range of 

 velocities supports this conclusion. 



3. Layer 3 



Observation of Layer 3 is much more satisfactory than that of Layer 2. It 

 is nearly universally observed by all investigators at truly oceanic stations. It 

 customarily occurs as a long linear segment of the travel-time plot extending 

 from about 12 km shot distance to about 35 km (see Figs. 1-8). Complications 

 sometimes occur in regions of rough bottom topography, partly because of 

 large corrections for water-depth changes, but also because of complications in 

 the crustal structure associated with large topographic features. 



Generally the stations chosen for Tables I to VI were placed as much as 

 possible in regions of flat or subdued topography in order to reduce these 

 complications. The success of this procedure is testified by the internal con- 

 sistency of the velocities. For example, in Table III the standard deviation of 

 the Layer 3 velocities in the eastern North Pacific is only 0.17 km/sec; and 

 much of this was contributed by a low value, CR8, which is well to the north 

 of the region covered by these stations. If this station is not included in the 

 average and standard deviation, the average is 6.79 + 0.11 km/sec for the 

 remaining stations of this area. This standard deviation contains the real 

 variation of the velocity and hence can be regarded as an upper limit to the 

 random standard error of measurement for this group of stations. 



It is probably too optimistic to assume that this estimate of the upjjer limit 

 to the random error of measurement also applies to the other groups tabulated 

 in Tables I, II, IV, V and VI. If this assumption is valid, then a large part 

 of the variability observed for these stations represents a real variation of 

 velocity. Unfortunately, the quality of the travel-time data is also variable ; 

 and it is not possible to say with confidence that a given variation is real even 

 though it may be several times the standard deviation determined from the 

 homogeneous group of Table III. 



In addition to the random error of measurement there can be systematic 

 errors. One of these arises from failure to recognize Layer 2 arrivals at short 

 range so that one includes them with the Layer 3 solution. This will cause 

 the Layer 3 velocity to be systematically low. A test for this was made by 



