ON THE DETERMIXATION OF GRAVITY AT SEA. 557 



The methods will only be briefly referred to here. From Table II. 

 curves were drawn connecting the readings of the aneroid at land stations 

 with the corresponding atmospheric pressure in millibars obtained from 

 the reduced readings of the marine barometer (fig. 2) ; the graphs are 

 nearly linear, the slopes varying with time. It was found simplest to 

 correct separately for (a) alteration of scale division value with time, and 

 (fe) creeping of the zero. Details of the method are given in the full 

 Report. The calculated pressures corresponding to a given aneroid reading 

 are designated pa in the Tables. 



A. second method lay in drawing fig. 3 from the data of Table II. and 

 fig. 2, relating the aneroid reading with the data of observation for 

 particular values of the pressure. The degree of imperfection of tiie 

 aneroid method of determining ' g ' is shown by the deviations of indi- 

 vidual readings from the graphs. Some of the discrepancies may be due 

 to the transport and re-setting up of the instrument between the Meteoro- 

 logical Office, s.s. Ascanius, R.M.S. Morea, and Reading. When the ob- 

 servations were continuous the readings are more consistent— hence more 

 reliance is to be placed on the Morea observations During the Ascanius' 

 voyage the suspension and levelling were altered several times. 



The reduction factor being known at any date, it was simple to find the 

 atmospheric pressure corresponding to any aneroid reading at a given date. 

 These pressures are designated pt in the Tables. 



A further value for an aneroid division was calculated on the assumption 

 that the Morea readings could be treated quite separately from the rest, 

 and that the graph was a straight line ; this appeared to be an extreme 

 assumption providing a useful check upon the other methods. These 

 values are designated Pe. "When corrected for station errors there is very 

 little difference between the results of the different methods of treatment. 

 The aneroid was not suitable for the investigation of its properties in 

 an experimental chamber, consequently other means were employed for 

 investigating the effects of (1) temperature ; (2) rate of change of tempera- 

 ture ; (3) rate of change of pressure (see O.R.). The pumping of the 

 aneroid was lessened but not obviated by the mounting ; though the boxes 

 were on a horizontal axis placed parallel to the keel, rolling aft'ected the 

 reading. Except in harbour its pumping was less than that of the 

 mercury barometer. The effect of different ships is shown in fig. i. 

 Though the aneroid was mounted on springs on R.M.S. Morea, the vibra- 

 tion of the ship had a greater effect upon the pumping of the aneroid. 



But the most troublesome feature of the pumping is that contact is 

 registered too early, as already explained. The amount depends upon 

 the relative frequencies of the pumping and of testing. When the head 

 is turned very slowly contact occurs only at one end of the travel of the 

 boxes; great rapidity would be required to make it equally probable that 

 the other end of the travel is recorded. The error is indeterminate, but 

 the value of gravity is systematically too low by a small amount. 



In a subsequent section this point is further considered. For inclusion 

 in the final diagrams the criterion has been an amount of pumping half 

 that permitted for the mercury barometer. For future work an aneroid 

 recording photographically by a reflected spot of light method is recom- 

 mended. The essential tb'Jig is to measure the extremes of the pumping 

 on each side of the mean. 



