424 



Profs. John Perry and W. E. Ayrton. [May 27, 



jars has been tested, and we find the same kind of confirmation. For 

 example, the following is one of the many experiments we tried : — 



September, 1877. 



A circular glass rod 128 centims. long, 0*832 centim. in 

 diameter, rested on two knife edges 117 centims. apart, and was 

 loaded with a weight of 400 grms. hung at its centre. The de- 

 flections of the rod were observed by a ray of light coming from a 

 rigidly fixed lamp, falling on a mirror rotated by the bending of the 

 rod, and reflected on to a rigidly fixed vertical scale. To prevent 

 errors arising from slight alterations of the position of the flame, it 

 was always focussed so that the image of two cross-wires fixed in the 

 lamp slit and formed by a fixed lens always occupied the same position, 

 when not reflected by the mirror attached to the deflected rod, and it 

 was the position of this image, after being reflected by this mirror, 

 which was observed on the vertical scale. After the beam had been 

 loaded for 145*55 hours, observations of the increasing deflection being 

 taken all the time, the load was removed. For several hours after the 

 load had been removed, the actual observations themselves needed 

 very little correction by drawing a curve, afterwards the slight wavi- 

 ness of the true curve due to changes of temperature had to be elimi- 

 nated. In the several tables of observations of loaded and relieved 

 beams (each experiment lasting two or three weeks), we have given 

 the temperature of the room at the time of making the observation. 



By the method of reduction adopted, we find that during the recovery 

 of the rod from t equal 20, to t equal 240 (1 = 1 "2 5 hours) the curve is 

 almost purely logarithmic, and we find by calculation that 

 ^=10-64- 224- 79€-o-oi296^ 



and the column of values of (iv-iv^), in the following table shows that 

 this expression satisfies the observations, very accurately for a very 

 long period, namely, 220 on our scale of time, or 275 hours. Taking 

 these numbers, we have calculated w 2 , the next term in the solution of 

 the differential equation, and find that 



^ 3 = 138-7e-0-3125^ 



so that the last column of the table shows the inaccuracy of the as- 

 sumption— 



^=10-6 + 224-79e-o-oi296^ + 138- 7 6 -o-3i25^ 

 an inaccuracy which is of very little importance from 



£=2-4 to £=240, 



and even this inaccuracy we have found is to some extent due to the 

 fact that the temperature of the room was steadily falling at that time 

 from 26°C. to 21° C. 



