78 BEXJ. PIKE'S, JR., DESCRIPTIVE CATALOGUE. 



that the telescope should, on every occasion, be raised or 

 lowered, by its proper screw, for making them perfectly so. 



The foregoing instructions for taking distances, apply 

 equally for taking altitudes by the sea, or artificial horizon, 

 they being no more than distances taken in a vertical plane. 

 Meridian altitudes cannot, however, be taken both back- 

 wards and forwards the same day, because there is not 

 time: all therefore that can be done, is, to observe the 

 altitude one way, and use the index-error; but even here, 

 you have a mean of that altitude, and this error, taken on 

 three different sextants. Both at sea and land, where the 

 observer is stationary, the meridian altitude should be 

 observed forwards one day, and backwards the next, and so 

 on alternately from day to day; the mean of latitudes, 

 deduced severally from such observations, will be the true 

 latitude ; but in these there should be no application of 

 index-error, for that being constant, the result would in 

 some measure be vitiated thereby. 



When both the reflected and direct images require to be 

 darkened, as is the case when the sun's diameter is measur- 

 ed and when his altitude is taken with an artificial horizon, 

 the attached dark glasses ought not to be used ; instead of 

 them, those which apply to the eye-end of the telescope 

 will answer much better ; the former having their errors 

 magnified by the power of the telescope, will, in proportion 

 to this power, and those errors, be less distinct than the 

 latter. 



In taking distances, when the position does not vary from 

 the vertical above thirty or forty degrees, the handles which 

 are attached to the circle are generally most conveniently 

 used ; but in those which incline more to the horizontal, 

 that handle which screws into a cock on one side, and into 

 the crooked handle on the other, will be found more appli- 

 cable. 



When the crooked handle happens to be in the way of 

 reading one of the branches of the index, it must be remov- 

 ed, for the time, by taking out the finger-screw, which 

 fastens it to the body of the circle. 



If it should happen that two of the readings agree with 

 each other very well, and the third differs from them, the 

 discordant one must not on any account be omitted, but a 

 fair mean must always be taken. 



It should be stated, that when the angle is about thirty 



