LEVELING, AKD USE OF BAEOMETER. 73 



fully compared before starting, one remaining at the lower station 

 while the other should receive sufficient time to reach the upper, 

 the observations to begin at a certain time and repeated at inter- 

 vals previously agreed u^Don, and continued until it is certain that 

 the distant party may have had time to make the necessary repeti- 

 tions. When not too distant a gunshot may answer for signal, if 

 beyond hearing distance the smoke from a fire may answer the pur- 

 pose, the signals applicable will generally depend upon the situa- 

 tion. 



Repetitions are necessary for several reasons. An imperfect con- 

 tact of the vernier scale or local and wave-like disturbances may 

 all tend to make the observation doubtful unless repeated, or the ap- 

 paratus may not yet; have partaken entirely of the local tempera- 

 ture. After carefully adjusting the index of the vernier to the top 

 of the column, the scale should be read, recorded, and the vernier 

 displaced, a second or third reading may verify the first, or one 

 another. The barometer requires two thermometers, one attached 

 to show the temperature of the mercury, (it being impossibe to in- 

 sert the attached thermometer into the mercury, it is only neces- 

 sary to place the same under as nearly as possible the same condi- 

 tions.) and one detached which should be moved about to give the 

 temperature of the atmosphere. Above all it appears necessary to 

 be well supplied with instruments when starting for a different al- 

 titude. Assistant Edwards, of the coast survey, started for the top 

 of a mountain in California with six barometers and arrived at his 

 station with one, one of the original syphons of which he carried 

 the mercurj^ in a flask in his pocket. 



It may be observed that valleys and abrupt inclinations should 

 be avoided, and isolated stations chosen where the atmosphere ap- 

 pears undisturbed. 



For reducing the observations the formula published in the 

 United States Coast Survey reports is most convenient in ordinary 

 cases, this is arranged for a mean temperature of fifty-five degrees 

 Fah., and is the product of the constant 55,000 multiplied by the 

 quotient of the difference of the Barometer readings divided by 



their sum, /55000 "^H^l and difi"ers but little from the truth. 



When we desire to make more accurate measurements, we find 

 the matter proportionatel}' complicated. If we assume that the 



