OF THE MAGNETIC FORCE. 5 



and it will, therefore, in this place only be necessary to recapitulate in general the 

 results and to state the nature of the experiments there described. 



The first observations for the temperature co-efficient were made on July 16, 

 1840. Oscillations were observed alternately at the ordinary temperature and near 

 the freezing point, obtained by surrounding the box containing the magnet with 

 ice ; at the same time comparative oscillations of a bar in another building were 

 observed to furnish the necessary data to correct the bifilar results for any change 

 in the horizontal force during the progress of the experiments. The value deduced 

 was 2.8 scale divisions for a change of 1 Fahrenheit. No reliance was placed on 

 this result on account of the comparatively rude indications of the subsidiary 

 instrument, and also on account of an irregularity at a certain point in the curve 

 representing the connection of change of force with change of temperature. 



The method of deflections was tried, and abandoned on account of the small 

 amount of deflection at a distance sufficiently great to prevent the chance of per- 

 manent changes from the mutual action of the bars. 



On the 22d of February, 1841, comparisons by vibrations were again resorted to, 

 but with no better success, the correction for change of force during the interval 

 being unsatisfactory. The result deduced was 3.0 scale divisions for 1 Fahr. 



Applying the results to the readings of the bar when mounted on the bifilar 

 suspension wires in the observatory, they were so little satisfactory that it was 

 determined to get the change of intensity of the bar by heating and cooling the 

 observatory while the bar remained in situ. 



In January and February, 1842, a continuous series of observations was made by 

 allowing the observatory to attain the winter temperature on one day, and obtaining 

 thus a result by comparison with the preceding and succeeding days, when the 

 room was artificially warmed. The value found was 1.55 scale divisions for 1 

 Fahr. At this time the observatory was warmed by a soap-stone stove with copper 

 fixtures. 



About the close of the year 1842 an efficient set of subsidiary instruments was 

 mounted in one of the College buildings, the bifilar magnet being about nine inches 

 in length. After the relative value of the scales of the instruments had been 

 ascertained, comparative observations were made, six each day, in the morning and 

 afternoon. These observations and results are given in a table extending over 

 eleven months, in 1843, and over eleven months, in 1844. The results were fluc- 

 tuating, and the discrepancies proved conclusively, that other causes were at work 

 which would not be accounted for. The changes in the force were generally small. 

 In the course of these experiments I found, beyond a doubt, that instruments of 

 the same dimensions were required to give comparative results. During an aurora 

 the small instrument in the College gave by no means the same results as the large 

 instrument in the observatory; there were numerous comparisons determining this. 

 I had reason also to believe that the large bar had its induced magnetism easily 

 disturbed, and not regularly renewing itself, so that the correction for temperature 

 may be supposed compound, one part permanent and cne part temporary. The 

 following results were obtained : 



