270 W. E. AND F. B. COOKE. 



some outstanding difficulties which will have to be met 

 and overcome. 



With those of a purely astronomical character we are 

 not here concerned. The object of this present paper is to 

 indicate a method whereby two clocks or chronometers 

 separated by some 5,000 miles or so may be compared with 

 .an error not exceeding one-hundredth of a second, and with 

 entire elimination of personal equation. 



In order that our remarks may be followed it will be 

 -advisable to briefly indicate the general features of the 

 whole operation, and the special difficulties to be attacked. 



Three, or perhaps four, stations are concerned. Let us 

 call the two astronomical stations (or those which contain 

 the clocks to be compared) A and B. A third station G 

 (the transmitting or signalling station) contains some piece 

 of apparatus which radiates a series of " dots" regularly 

 spaced at intervals of nearly, but not quite, one second 

 apart. This station should preferably be about midway 

 between A and B, but if this is impracticable, we shall 

 probably have two stations d and C 2 close to A and B 

 respectively, and each acting alternately as a transmitting 

 station. For definiteness we shall suppose a single station 

 G. For example, if A and B were the Sydney and Lick 

 Observatories, G might be at Honolulu : or with A and B 

 at Sydney and Honolulu might be at Nauru (near the 

 equator). 



All we expect from G is a series of regularly spaced 

 ""dots" at nearly a second interval. The exact times of 

 the dots are quite immaterial. It is the object of the 

 •observers at A and B to receive this series of arbitrary 

 signals and compare them with the beats of their 

 respective clocks, and this is accomplished by the well 

 known method of coincidences. Our problem now resolves 

 itself into this — to determine the exact seconds at which 



