CHAP. XVI.] SEA OBSERVATIONS 365 



where we laid-to the night before, and in many instances unless 

 we know in what direction to go, it is useless to move at all. 



In such cases observations should be taken throughout the star 

 night ; for, though they will give but approximate results, latitudes, 

 latitudes by pairs of stars north arid south of zenith by the 

 same observer should give under favourable circumstances a 

 position within five miles of the correct latitude, which will 

 tell us if we are drifting much in the line of the meridian, and 

 also affords us an approximate latitude to work longitude by, 

 in the morning. 



The worst of it is. that circumstances apparently favourable 

 are often not really so, as the great source of error in night ob- 

 servations is the impossibility of being certain of the horizon. A 

 false horizon will frequently look so well defined as to mislead the 

 best observer, and will, of course, throw out the resulting lati- 

 tude greatly. Thus we can put no great faith in latitude by stars, 

 and none whatever in a single observation, or even in a single 

 pair, as the horizon in one direction may be true and in another 

 false. It is only in a series of pairs of stars through several 

 hours that we can have any confidence, as if the results of these 

 agree fairly, or steadily show movement in one direction (the 

 effect of a current), we may then feel pretty sure of our position 

 as far as latitude goes. 



Night observations at sea for longitude are not of much use ; star 

 but, under unusually good circumstances of horizon, the mean Observa- 

 of two star chronometers, one east, the other west of meridian, Longi- 

 may be better than nothing. ^^^^ 



If, however, when the day has sufficiently broken to enable g^^^.^ ^^ 

 the horizon to be clearly seen, we can get observations of bright Daybreak, 

 stars or planets on different bearings, we can obtain an excellent 

 position by Sumner's method from whichtostart our day's work. 



Or, if we can only get one daybreak star, as soon as we can 

 get an observation of the sun, we can combine it with the day- 

 break observation. 



The most satisfactory position is that obtained with a twilight 

 horizon from the quadrilateral formed by the intersection of the 

 lines of position derived from observations of a pair of stars 

 on nearly opposite bearings, combined with observations of 

 another pair of stars on bearings at right angles to the first 

 pair. The patent log being read when each star is observed, 



