162 Mr. R. H. M. Bosanquet on Sudden Changes in 



typical parallelogram, as the corresponding change of tempe- 

 rature usually took place in the small hours of the morning. 

 The diagram May 4th-5th shows a case in which this part of 

 the curve was obtained. This exhibits a subsidiary effect, 

 which consists of a tendency to increase the elongation * in 

 the neighbourhood of the lower limit when the temperature 

 is constant. Another very clear case of this effect will be 

 found on the diagram for April 28th. My impression is that 

 conditions of wire are common in which this effect is consider- 

 ably developed. In such cases changes of temperature give 

 rise to excursions, after which there is a more or less rapid 

 return to a mean position. 



The observations and diagrams hitherto referred to deal 

 with changes in the angular position of a body hanging from 

 the wire at rest. Observations were also made of the periods 

 of vibration, for the purpose of ascertaining what changes 

 took place in the torsional rigidity of the wire. The results 

 of these observations are as follows : — 



Periods of vibration. 



Mean period. 



Mean elongation. 



Number of 

 observed periods. 



m. s. 







4 45 



89-2 



4 



4 53 



88-5 



4 



4 41 



85-4 



13 



5 



84-6 



4 



5 39 



83-7 



4 



6 3 



77-0 



4 



5 52 



72-1 



12 



6 33 



57-9 



4 



4 23 



54-6 



2 



5 33 



46-4 



4 



5 16 



37 



2 



3 56 



35-0 



8 



3 4 



323 



2 



2 29 



28-6 



2 



2 40 



27-5 



4 



It appears from these numbers that the period of vibration 

 is least when the elongation is in the neighbourhood of the 

 lower limit, increases rapidly as the limit is departed from 

 up to a maximum value, and diminishes again slightly as the 

 upper limit is approached and attained; that is to say, the 

 torsion force has 



a maximum at the lower limit of elongation; 

 a minimum between the limits; 

 and an increase up to the upper limit. 

 The observations for the period are necessarily complicated 

 * I. e. to return from the lower limit towards the mean position. 



