8 DISCUSSION OF THE VERTICAL FORCE. 



standard temperatare to which all the vertical force readings, taken with the reflect- 

 ing magnetometer, have been referred. 



A close examination of the record of the Lloyd balance magnetometer, which 

 was used in June and July, 1840, in the College, and afterwards at the observatory 

 during five months, proved that in point of accuracy it would not compete with the 

 reflecting magnetometer mounted in January, 1841, and continued in use for four 

 years and a half. Owing to some imperfection in the first named instrument, its 

 indications were very unsteady, and at times fitfully changeable ; thus in September, 

 October, and December, there arc differences in the daily means (deduced from 

 twelve readings and referred to 32 Fahrenheit) of adjacent days of more than 200 

 scale divisions, and in one instance (October 19-20) amounting even to 256 

 divisions. In August there is a change of 389 scale divisions in three consecutive 

 days, and in October (17th to the 20th) one of 477 divisions in the means during 

 the same interval. There is besides a large progressive change, showing that the 

 instrument was in a very unstable equilibrium ; this change amounted in the first 

 month to over 300 scale divisions. An attempt was also made to deduce a tem- 

 perature coefficient by comparing mean daily readings of short and specially selected 

 periods of a few days each, with average high and low temperatures, but it failed 

 for want of sufficient uniformity in the indications of the instrument. In such a 

 series the disturbed indications could not be recognized and separated from the 

 regular readings. It was finally concluded to make no use of the observations 

 prior to January, 1841. 



Reduction of the Observations, bettceen February, 1841, and June, 1845, to a 

 uniform Temperature. A table has been constructed, with the observed tempera- 

 ture as the argument, giving the reduction for difference of temperature from 

 the normal temperature (66 Fahr.) ; by means of this table each observation has 

 been referred to its corresponding value as the standard temperature. Table No. 

 2 contains the monthly mean readings for each observing hour ; the time is local 

 time, and reckoned from midnight to midnight to 24 hours. The tenths in the 

 record have been omitted, as of no special value, since an error in the recorded 

 temperature of only 0.l affects the magnetometer reading by more than a scale 

 division. An increase of scale readings corresponds to a decrease of vertical force, 

 and one division equals 0.000033 parts of the force. Accidental irregularities in 

 the record are especially referred to in foot notes. 



The tabular values are directly taken from the manuscript tables containing the 

 single reduced readings and their monthly means. 



In the present state of our knowledge regarding the occTirrcnce of the dis- 

 turbances it is not safe to make any interpolations in the magnetometer record in 

 case of an accidental omission ; a rule which has been strictly adhered to. 



