88 REPORTS ON THE STATE OF SCIENCE. — 1919. 



co-operate with him in this research. The films were measured up in the 

 Physics Laboratory of University College, Reading, by Mr. Whittall, 

 and the calculations carried, out by Mr. F. S. Hayhoe and Mr. Harrenden 

 Harker. 



The first point to be decided was whether the variations in temperature 

 during the course of each experiment, due probably in part at least to 

 the lamps used for illuminating the mercury surfaces, occasioned a change 

 which ruled the whole investigation out of account. It was, however, 

 found that in general the mercury did not rise as the run proceeded, in 

 spite of the usual increase in temperature registered on the stem. It 

 was decided, therefore, to take the first reading of the stem thermometer 

 as correct, and it was presumed that the lag of the barometer temperature 

 was greater than that of the indicating thermometer on the stem within 

 the casing. 



The readings taken as standards were any four observations made at 

 the following ports — there was usually a choice of two at each port : — Fre- 

 mantle, Colombo, Bombay, Aden, Malta, Gibraltar. 



The following are the constants in the above equation obtained by 

 taking two Colombo and two Bombay observations as standards : — 



A„ = -1*2395, B„ =2*0364, C„ = 0'131* 



Whereas, taking one each for Colombo, Bombay, Aden and Malta, the 

 values were 



A„ = -11364, B„ = 1-816, C„ = 0*211. 



Using the former values, we note that Fremantle (Film 9) works out 

 as g = 978-904 instead of 979-485 given by pendulum observation, and 

 that, if the data obtained from Fremantle (Film 8) are used, the value 

 of g obtained is 978*736. There is thus no agreement between the values 

 obtained from consecutive films for the same port. Similar discrepancies 

 were obtained by using the second set of values, and other harbour 

 stations provided equally imsatisfactory results. 



The investigation was thereupon abandoned. 



Discussion of the Method. 



Apart from the possibility of leakage, which only was obvious in one 

 barometer on the return voyage, there are difficulties in the method. One 

 barometer could not be measured up because the image of the engraving 

 could not be seen on the film, though -visible during the adjustments. 

 The great difficulty, however, lies in the temperature measurement. The 

 value of the coefficient of U (C^, = -211) indicates that a variation in 

 ht of 0-1° C. means a change in the value of g of about -02 cm. /sec. ^ ; it is 

 essential chat this temperature should be measured on a thermometer 

 graduated to divisions less than 1-0° C. It is further necessary that this 

 thermometer should read the temperature of the mercury within the 

 barometer tube, and with the present arrangement this is not accomplished ; 

 it is important that the temperature lags of the barometer stem and of 

 the attached thermometer should be equal. 



