88 



SCIENCE. 



[Vol. XIX. No. 471 



causing danger of Are, bodily injury, or even loss of life. 

 Pecuniary liabilities in such cases may be very great. The 

 financial officers of the university regard such risks as more 

 than offsetting the receipts for the time-signals. 



One of the greatest advantages of the time-service to the 

 observatory has been that it kept before the public the prac- 

 tical value of astronomical work. Many thousands of per- 

 sons who take no interest in work of a purely scientific 

 character recognize the great financial value to the public of 

 an accurate system of time. The observatory desires to con- 

 fer this benefit on the public, and it would be ready to do 

 so even at a financial loss. But recently the time-signals of 

 the United States Naval Observatory have been offered to 

 the public at very low rates, through the Western Union 

 Telegraph Company. This can the more readily be done 

 since the expense of furnishing the time is borne by the 

 people through a government appropriation, while the com- 

 pany has the largest facilities for the maintenance of tele- 

 graphic connections. The Harvard College Observatory is 

 therefore relieved of this duty. If the public is to be the gainer, 

 signals of equal accuracy and continuity must be furnished. 

 Unfortunately, signals sent to a great distance are liable to 

 frequent interruptions from trouble with the telegraph lines, 

 and therefore secondary clocks must be used in each large 

 city if continuous signals are to be distributed. These 

 clocks must be constantly compared and corrected if great 

 accuracy is to be attained, and it is still a question whether 

 satisfactory results can be secured outside of an astronomical 

 observatory. If the results prove unsatisfactory, however, 

 the responsibility for trying the experiment will not rest upon 

 this observatory. 



In view of the facts stated above, it has been decided to 

 discontinue the time-signals furnished by this Observatory 

 after March 31, 1892. An earlier date would have been se- 

 lected, but for the desire to give our subscribers sufficient 

 time to make other arrangements for securing signals. 



The most important events in the history of the time-ser- 

 vice are given below. The first transmission of time from 

 the observatory to Boston was over a line hired for the pur- 

 pose and used occasionally for the compirison of clocks in 

 Boston with the standard clock at Cambridge. From 1856 

 to 1862 the observatory owned a line for the same purpose. 

 Up to the close of 1871, no charge was made for the time 

 thus furnished, which was used for many years for striking 

 the fire-alarm bells of Boston at noon, and for other purposes. 

 The regular transmission of signals and the receipt of com- 

 pensation for them began in 1872, the service being under 

 the direct care of Professor Winlock, who had devised the 

 system. After his death in 1875, Professor W. A. Rogers 

 took charge of the service and introduced the custom of 

 telegraphing information as to the error of the signals at a 

 given hour daily. In 1877 Dr. Leonard Waldo took charge, 

 and during the next year, with the liberal cooperation of 

 the Equitable -Life Assurance Company, the Boston Time- 

 Ball was erected on top of the building of that company. In 

 1878, also, a correspondence was opened with the railways 

 of New England relative to a uniform system of lime and 

 the practicability of introducing it by legislation. A plan 

 for establishing a bureau for the testing of fine watches and 

 thermometers was considered, and abandoned on the ground 

 that such work would be commercial rather than scientific, and 

 therefore not within the scope of the observatory. ' In 1879, 

 Professor Frank Waldo, who had previously assisted his 

 brother, took charge of the time-service. The error of the 

 standard sidereal clock was determined every day at 10 a.m. 



from the latest comparisons with the stars, assuming the rate 

 to continue uniform. The mean-time clock was compared 

 with this, and for several years the difference had been com- 

 municated every day by telegraph. This practice was aban- 

 doned, since it was easy to reduce this difference to zero, 

 and it did not indicate the true error of the clock. Especially 

 during continued cloudy weather, large changes might take 

 place in the rate of the sidereal clock, which could not be 

 determined until observations could be made of the stars. 

 At this time the signals were sent to New York, and were 

 used in the time-service of that city in combination with 

 similar signals sent from the Naval Observatory and Alle- 

 gheny Observatory. It developed the interesting fact that 

 the differences, sometimes amounting to several seconds, 

 were much greater than were expected, or than would be 

 derived from combining the supposed errors of the different 

 time-services. This was regarded as a preliminary trial of a 

 plan which was developed later, and appears to be the only 

 way of effecting a great increase in the accuracy of time- 

 signals. It is easy to keep the errors of a clock small if the 

 weather is clear, and frequent comparisons can be made with 

 the stars. During long periods of cloudy weather, however, 

 when no observations of the stars can be made, it is very 

 difficult. The slight changes of rate to which even the best 

 clocks are liable may cause serious errors at the end of sev- 

 eral days. The remedy is co-operation between observatories 

 so distant that it would seldom happen that clouds would 

 prevent observations at all of them. The time would be 

 determined at each observatory every evening, when it was 

 possible, and the result transmitted telegraphically to a cen- 

 tral station ; also when called for, as soon as it cleared, 

 whatever the hour. The central station would report daily 

 to each observatory either the results of each observation re- 

 ceived or a corrected error derived from them all. Each 

 observatory might send its own time or receive signals from 

 a normal clock at the central station. Mr. J. Rayner Ed- 

 mands, who has had charge of the lime-service from June, 

 1881, to the present time, rendered important aid in forming 

 this plan. He postponed the record of the errors occurring 

 during cloudy weather until observations could be made for 

 determining them. The apparent errors were thus increased, 

 but the actual errors were repre.sented with much greater 

 accuracy. The practice of making the error at 10 a.m. es- 

 pecially small was abandoned, and attention was given to 

 keeping the signals as accurate, and the daily rate as small 

 as possible at all hours. The genera.1 introduction of stan- 

 dard time was effected at noon on Nov. 18, 1883. After the 

 change was decided upon, a large part of Mr. Edmands's 

 time for several weeks was devoted to securing the assent of 

 the public throughout New England to the proposed change. 

 In 1885, a new time-ball was erected on the Boston post oflBce 

 building, with the aid of an appropriation from the city of 

 Boston. Experiments were made in various matters asso- 

 ciated with the distribution of accurate time. Among others, 

 a delaying apparatus was devised, by which the signals of a 

 clock could be retarded by any desired fraction of a second, 

 so that, without disturbing a clock, its apparent error could 

 be vai'ied at will. In 1889 some interesting experiments 

 were made by Mr. W. P. Gerrish on distributing time accu- 

 rately by flashes of magnesium powder. Signals were thus 

 sent from a station on Blue Hill, twelve miles distant. They 

 were readily visible, and the exact time to within a fraction 

 of a second could be taken from them. These flashes were 

 also seen from Princeton and Mount Wachusett, forty-four 

 miles distant, and from numerous nearer points. From an 



