i aaa 
Baar RF 
A ee 
Pa Se 
es 
SS 
= 
MAS by Sahn ee ey 
Wide fener 
f 
3 
396 Canadian Record of Science. 
Montreal clock having gained that much on the Toronto 
signals during their transmission. On the assumption that 
the time of transmission in the two directions is equal, 
which for land work, at least, we may safely make, the 
mean of the comparisons gives the precise difference in the 
clocks. As a matter of fact, clocks at distant stations which 
are connected by a well insulated land telegraph line can 
be compared and their difference ascertained with the same 
accuracy as when beside one another. When in the ex- 
change, clock signals only are sent over the telegraph line, 
the two systems of signals may be passing simultaneously, 
the clocks at A and B recording on both chr onographs. 
The comparison of clocks having been made a second set 
of star transits is observed. Thus the clock error being 
known just before and again just after the clock comparison 
may easily be computed for the instant of the comparison, 
The following taken from the Cambridge (U. S.,)—Mont- 
treal determination will serve as an example of what has 
been described :— 
Date Montreal Correc-| Montreal | Cambridge Correc-| Cambridge | Differ- 
: Clock | tion | Time | Clock | tion Time ence 
————$— 
a 
RECORD ON MONTREAL CHRONOGRAPH. 
3 Ite Ti, | s. | Pe fis She |flte. TSB) h. m. s. |m.s. 
1883 | 15 36 31.07 | +2.91 |15 36 33.08 | 1546 34.65 |—14:33]15 46 20.32 9 47.24 
RECORD ON CAMBRIDGE, CHRONOGRAPH. 
15 21 27.60 | +2.01 | 15 2L 29.61) 15 3131.26 | —14.33| 15 31 16.93] 9 47.82 
\—————| 
Mean difference of time of the stations As 9 47.28 
Now if the clock corrections here used were quite correct 
this result, 9m. 47-28s., would be our true and final value of 
the difference of longitude. Unfortunately, however, it is 
not an easy matter to obtain a clock error which is true to 
the nearest hundredth of a second or even to the nearest 
tenth. It would probably be nearer the truth to say that 
an uncorrected result of a set of.observations by a good ob- 
server is likely to be somewhere within a quarter of a 
second of the true error. If, however, he is a well trained 
observer he should always be in error (within a range of one 
tenth of a second) to the same amount and in the same direc- 
