176 O. T. Sherman— A Pendulum Study. 



For better distinction the threads were divided into tallies- 

 five black and five white. The interval between the threads 

 was about seven-tenths of a millimeter. The observer then 

 sitting behind the telescope made at each apparent contact of 

 the knife edge with a thread, a signal on the chronograph at 

 once representing the tally and thread which the k 

 seemed to touch. The result was so unforseen as for some time 

 to cause a doubt in the mind of the observer as to its correct- 

 ness. He assured himself of its reality by obtaining fairly 

 identical results again and again. Over some threads the knife 

 edge passed with rapidity. On others it rested for an interval 

 2 with the decreasing rapidity of the movement. 

 Many it touched, passed, retouched, repassed. A representa- 

 tion is annexed. The abscissae being the interval in minutes 

 since the commencement of observation. The ordinates, the 

 number of threads from the center. 



It is interesting to compare with the above paragraph a 

 remark of the Coast Survey Observer, who notices that the dis- 

 crepancies of the separate observations of the intervals at 

 which his pendulum reached a given arc are several times lar- 

 ger than can well be attributed to errors of observation. Is it 

 not quite probable that there was in his case an indeterminate- 

 ness similar to that shown in the diagram? 



It seemed likely that this was due to a walking of the pen- 

 dulum, such as might effect a change in the absolute position 

 of the bar during the time of swing. I therefore fixed on the- 

 bar a small bit of looking-glass in which was reflected a scale. 

 At the beginning of the swing a thread was set so that when 



