300 HYDROGRAPHICAL SURVEYING [chap. xiii. 



watch is beating. For instance, supposing that at the 40- 

 seconds' mark on the dial the hand falls a little short of the mark 

 one beat and a little in advance at the next, unless he knows 

 which of those beats is meant for the 40 seconds, he may be 

 gi\angthe time four-tenths of a second wrong. This, of course, 

 refers both to comparing and taking time for the observations. 

 Method of In comparing, which is perhaps best done by two persons, 

 ing. the " Stop " is given on an exact second of the box-chrono- 



meter (wliich beats half-seconds), and the time by the pocket- 

 watch noted. A check is then taken by comparing the 

 reverse way, calling the " Stop " at an exact second by the 

 pocket-watch, and noting the seconds and parts of seconds of 

 the box-chronometer. As parts of seconds have to be esti- 

 mated on both watches, these two comparisons will frequently 

 differ two-tenths of a second. This is as near as we shall 

 probably be able to arrive at the truth ; but if the difference 

 exceeds this, more checks should be taken, until we are satisfied 

 which was wrong. 



No operation requires more care or more practice than com- 

 paring, and while the simple method given above will give 

 good results when observers are experienced, varying the beat 

 on which the word " Stop " is given has the effect of varjdng 

 the fraction of a second to be estimated by the other observer, 

 and tends to avoid bias on his part. 



Thus, an even second, an odd second, and a half-second 

 beat should be used successively in giving the word " Stop " 

 in the case of a chronometer, and any or all of the five beats 

 in the case of a pocket-chronometer. 



At least six comparisons on different beats should be inter- 

 changed to secure a good result. 



In the same manner checks should be taken when com- 

 paring one box-chronometer with another. 



If two pocket-watches are available, it will not be amiss 



to use both, even when there is only one observer, as it helps 



to eliminate errors of comparison. In tliis case half the sets 



will be taken with one watch, and the other half with the other. 



Com- There will be a coincidence of beats at intervals of 3 minutes 



I a nson of 



Solar and 4 seconds when using chronometers beating half -seconds, 



Sidereal ^^^^^ ^^ intervals of 6 minutes 8 seconds in the case of clocks 



Chrono 



meters, beating seconds. The comparison is made when the chrono- 



