Observations ivith Rater's Pendulums. 167 



discordantly than those of the other two pendulums. It was 

 feared that the ends of the knife edge sometimes touched one or 

 other of the guiding faces close to the grooves which receive the 

 knife edge when the pendulum is lifted off the planes. These 

 pieces were taken off by Mr. Ellery on 6 th SeptemV)er, and 

 another series of twelve swings, six in each position, were 

 obser\ed under the new condition. This makes in all sixty 

 swings of an average duration of five hours each. The general 

 practice was to commence a swing at about 9.30 a.m., conclude 

 it at about 3 p.m., commence another at 4 p.m., concluding at 

 9.30 p.m. Generally five coincidences were observed at the 

 beginning and end of each swing. The pendulum was always 

 stai'ted about half-an-hour before the first coincidence was 

 observed. 



Clock Rates. 



Until the 12th August the Shelton Clock was compared with 

 the Seth Thomas by eye and ear, the fraction of the second being 

 determined by coincidences of beat, with a Mean Time Chrono- 

 meter. The rates of the Seth Thomas were derived by chrono- 

 graphic comparison with the Standard Transit Clock, the error 

 of which was determined by transit observations. After the 

 12th August, Mr. Ellery mounted an electric contact spring on 

 the Shelton Clock, by which a signal was made at every sixtieth 

 second, on the Tape Chronograph, on which the beats of the 

 Transit Clock were simultaneously recorded, thus enabling a 

 comparison of the two clocks to be made with all the accuracy 

 obtainable. The uncertainty introduced by one of the weakest 

 points in pendulum observations was, by this method of 

 comparison, greatly reduced. A comparison was made at the 

 commencement and end of each swing; but when two sets of 

 swings were observed on the same day in succession, extending 

 from 9 a.m. to 10 p.m., the rate of the Shelton Clock was derived 

 from the two extreme comparisons in the morning and evening 

 only, neglecting the two intermediate ones. In this way it was 

 thought, although the rates for the two swings might be difierent, 

 the resulting mean of the two vibration numbers would be 

 improved. The Shelton Clock gave a good account of itself. 



