]80 



but are entirely of an arbitrary value. If now we determine the 

 points of the scale at which the mercury stands in freezing and boil- 

 ing water, we can immediately convert the arbitrary scale-readings 

 into degrees of the ordinary scales of temperature. If a be the scale- 

 reading for the freezing-point, and h that for the boiling-point, the 

 temperature by Fahrenheit's scale corresponding to any reading 



'^'^—{^ — ^^180 + 32. The freezing-point is determined by placing 



the thermometer in finely-pounded ice, from which the water is 

 drained off as it melts. The boiling-point is ascertained by the form 

 of apparatus employed by M. Regnault ; the temperature observed 

 is that of steam, whose elasticity is the same as that of the atmo- 

 sphere. A small siphon water-gauge communicating with the in- 

 terior of the vessel gives notice to the observer when the ebullition 

 is being carried on too rapidly. The steam is generated from di- 

 stilled water. The height of the barometer is observed at the time 

 of the experiment, and the correction to a uniform height of 30 

 inches (reduced to 32°) is found from Regnault's table. In deter- 

 mining the fxxed points, the stems of the thermometers are kept 

 vertical ; if the subsequent comparisons with other instruments are 

 made in the same position, no error will arise from the expansion of 

 the bulb caused by the pressure of the column of mercury. If, how- 

 ever, the thermometers are intended to be used in any other than a 

 vertical position, it becomes necessary to determine the fixed points 

 also in a horizontal position. 



In accordance* with the plan here sketched, fifteen thermometers 

 have been completed with arbitrary scales. About thirty more tubes 

 have been calibrated, and the bulbs attached and filled, but the scales 

 not yet divided. The principal object in graduating the tube with 

 an arbitrary scale is the convenience it afl['ords of testing the di\d- 

 sions before it is converted into a thermometer. It is now proposed 

 to divide the scale at once into Fahrenheit degrees after the ther- 

 mometer has been made, and to test the accuracy of the di\dsions 

 afterwards by detaching a portion of the mercurial column and 

 making it move along the tube. If the scale should not then be 

 found correct, a table of its errors can be formed and furnished with 

 the instrument, or the thermometer rejected. The scales of these 

 thirty thermometers have not yet been proceeded with, as it is 

 desirable, before doing so, to allow the freezing-point to have 

 attained a permanent position. A few divisions have been cut on 

 the tubes near the freezing-point, and the reading with reference to 

 this short arbitrary scale taken from time to time in melting ice. 

 The period elapsed since the construction of the thermometers has 

 been too short to afford as yet much information as to the probable 

 constancy of the freezing-points. They have, however, already 

 shown generally a tendency to rise, in some cases to the extent of 

 nearly 0°*3 Fahr., but in most of them it does not yet exceed 0°'l 

 or 0°*2. Another peculiarity in connection with the freezing-point 

 has shown itself in almost all the thermometers yet tried. After a 

 thermometer has been exposed for some weeks to the ordinary tern- 



