136 WORZEL AND HARRISON [CHAP. 9 



during April, 1956. Thirty-nine of the gravity-meter measurements lay within 

 3 mgal of the pendulum results and all but three lay within 9 mgal. 



These meters possess several advantages over the pendulum apparatus ; they 

 are less fragile and easier to install and very much less labor is required in 

 the computation of the results ; but most important of all, they are suitable 

 for development for use on surface ships. The Graf meter was the first to be 

 used on a surface ship, in November, 1957 (Worzel, 1959), and tests of the 

 LaCoste meter followed in 1958 (LaCoste, 1959; Harrison, 1959). 



B. Some Theoretical Conclusions 



Surface-ship operation of gravity meters possesses the advantage that the 

 measurements are made effectively at sea-level and that, apart from the wave 

 movements, there can be no changes in vertical velocity of the instrument 

 during a measurement. There is, therefore, no need for the accurate leveling 

 survey which may be more trouble than the actual gravity measurements in 

 land surveys and, provided the wave periods are removed from the gravity 

 record, no account need be taken of accelerations caused by changes in the 

 vertical velocity of the vessel, which is an important factor in a submarine or 

 airplane survey. It is, however, necessary to make a correction (the Eotvos 

 correction which may be large) for the ship's speed, and the gravity measure- 

 ment must be taken in the presence of large vertical and horizontal accelera- 

 tions. The ship may be at sea for several weeks, or even months, between ports, 

 so that the meter must have a small or linear drift rate and be free from sudden 

 jumps in reading (tares). 



The beam of a spring gravity meter must be heavily damped so that the 

 meter can withstand vertical accelerations of amplitude as high as 50,000 or 

 100,000 mgal without the beam hitting its stops. The spring and reading 

 system must be linear or symmetrical so that the mean beam position, deter- 

 mined by averaging the beam record, is independent of the magnitude of the 

 wave accelerations. Alternatively, in a non-linear instrument such as Gilbert's 

 vibrating string meter, it is necessary to apply a correction depending on the 

 magnitude of the vertical accelerations. Apart from these effects of non- 

 linearity, the vertical accelerations can be averaged out during a measurement 

 provided they are not large enough to cause the meter beam to hit its stoj)s. 

 The same is not true for the horizontal accelerations. 



The problem of the horizontal accelerations may be met in two ways : first, 

 the meter may be mounted on a stabilized platform which keeps its axis 

 vertical. In this case, the effect of the horizontal accelerations is eliminated 

 provided the gravity meter is constructed in such a way that it is unaffected by 

 accelerations perpendicular to its measuring axis. Alternatively, the meter may 

 be hung in a gimbal sus]iension and allowed to swing with the wave accelera- 

 tions. In this case a correction must be applied for the horizontal accelerations. 

 The Graf meter contains no method for correcting for horizontal accelerations 

 and must, therefore, be used on a stabilized platform. The LaCoste meter 



