Report of the Gravity Survey Covuuittce. 215 



sti'ucted for use in the French geodetic surveys an instrument 

 which he terms a "reversible invertible pendulum:" this 

 instrument has two iixed knife-edges, afid two interchangeable 

 weights, which are transferred from end to end of the pendulum 

 in each observation. Commandant Defibrges is of opinion that 

 such a mode of construction will do away with the errors 

 inseparable from the use of so-called invariable pendulums. 1 

 cannot agree with him here : for it seems to me very doubtful 

 whether we should ever get the weights after displacement back 

 to exactly their original positions : and the errors thus com- 

 missible, though small, are quite likely to be of a higher order of 

 magnitude than those which the method is designed to eliminate. 

 For instance, suppose a fine dust particle should get lodged 

 between the weight and its seat : this might well be the ten- 

 thousandth of an inch thick, and yet escape observation : yet it 

 would alter the vibration number in a half-seconds pendulum by 

 one and a half vibrations per day. Moreover, the large amount 

 of extra handling, screwing up, etc., involved seems to me to 

 militate seriously against the trustworthiness of the arrangement. 

 I have entered into this matter at considerable length because 

 Commandant Deftbrges is looked upon, and rightly so, as one of 

 the foremost of living geodesists, and his work demands very 

 careful consideration. 



Positing then the necessity for invariability, as far as it can 

 be attained, let us see how we are to seek it. 



I am strongly of opinioii that the pendulum should be rigid ; 

 this, indeed, was the iirst point forced on my attention when 

 handling the Indian pendulums. However much care be taken, 

 a five-foot bar of thin brass, with comparatively heavy weights at 

 its ends, is continually undergoing fiexures of a rather alarming 

 magnitude during the processes of handling which necessarily go 

 on in the course of the transfer of the pendulum from its case to the 

 cylinder in which it is vibrated; I have serious doubts whether any 

 amount of care can prevent such flexui'es as will introduce a per- 

 manent set and alter the vibration number. Nor are these doubts 

 without solid foundation ; we know that these pendulums have been 

 bent in the past, and have had to be re-straightened —a process 

 which necessarily destroys the continuity of the observations 

 made with them -and one of them is cei'tainly not quite straight 



