75 



several days by Prof. Kendall. It contained two tilt-hammer elec- 

 trotomes, one invented by Mr. J. J. Speed, jun., of Ithaca, N. Y., in 

 1847, attached to the minute wheel, and giving its signals every two 

 and a half minutes. The other (used for the occasion) was on the 

 plan invented by Dr. Locke, in 1848, and attached to the escape- 

 ment-wheel. The automatic clock register was graduated to two 

 seconds, usually occupying an inch of paper of the Morse's register- 

 ing fillet. 



Mr. Walker reports, that a comparison of sixty records, made by 

 the two registers at Washington, shows that the probable error of the 

 mechanical operation of printing and reading off, is only about fifteen- 

 thousandths of a second. 



This confirms the estimate of accuracy of the work made by him 

 in his report of Dec. 15lh last, viz., of a hundredth of a second for 

 the case of an automatic register of single seconds, with an inch of 

 paper to each. 



It further appears, from the Report, that when the star-signals 

 were given at Philadelphia, so that the clock and signal-waves had 

 the same local origin, all the registers at all the stations, marked 

 alike, within such limits as were indicated by the probable error just 

 mentioned. 



When, however, the star-signals were given at New York, small, 

 but appreciable, differences were noticed in the respective readings of 

 the apparent date, of the same event as recorded at the different sta- 

 tions. This discrepancy was still greater for the case of the Cam- 

 bridge star-signals, the graduating clock remaining always at Phila- 

 delphia. 



The following table contains the mean excess of the readings of 

 the date of each event in the time of the Philadelphia automatic clock 

 at each station, over that of each of the others, with the number of 

 single results, and the probable accidental error from the source al- 

 ready referred to. The stations compared are denoted by their ini- 

 tials. Those marked W, are for the mean of the two records made 

 at Washington. A further revision of these quantities may somewhat 

 change their amounts. 



Two kinds of readings were made, viz., break circuit signal read- 

 ings on a break circuit clock scale, and make circuit signal readings, 

 on a make circuit clock scale. 



The excesses indicated by the mean of the two series of readings for 

 the two scales, with the number of results and probable error of each, 

 are reported as follows. The times A, B, C and D, respectively de- 



