Royal Society. 307 



of thermometers, I shall first describe generally the method pursued 

 in the graduation of standard instruments. 



Tlie plan of operations hitherto adopted has been that proposed 

 by M. Regnault, and consists essentially of the following steps : — 

 1st. Calibration of the tube : 2nd. Graduation of the scale : and 

 3rd. The determination of the scale coefficients. 



1. Calibration. — A tube having been selected as being tolerably free 

 from all visible defects, a short column of mercury, generally less 

 than one inch in length, is introduced. The tube is then attached 

 to the frame of Perreaux's dividing-engine, and by means of flexible 

 tubing is put in connection at both ends with india-rubber bags, the 

 pressure upon which can be regulated by means of screws. 'I'he 

 mercury is then brought to the part of the tube where the gradua- 

 tion is proposed to commence. The cutting-frame of the engine 

 carries also a small microsco])e with cross wires in its focus ; on 

 turning the dividing-screw, the microscope-wire is brought to coin- 

 cide with the first end of the mercury, and the screw is then turned 

 forward until the wire reaches the second end ; the length of the 

 column is thus given in revolutions of the screw. By means of the 

 india-rubber bags, tlie mercury is made to move along the tube until 

 the first end coincides again with the microscope- wire; the length 

 of the column is again measured, and the mercury again moved for- 

 ward ; the same process being repeated until the column has been 

 measured for each length of itself through the whole extent of the 

 proposed scale. Permanent marks are made on the glass at the 

 points of commencement and ending of the calibration. If the pro- 

 gress of the numbers shows any considerable irregularity in the 

 tube, and as a verification of the first set of measures, it is well to 

 repeat the calibration, commencing in this case at a point one-half 

 the length of the column in advance of the original starting-point. 

 A series of measures interpolated from the two sets may then be 

 adopted. Some exjjerience is necessary in order to bring with faci- 

 lity the end of the mercury exactly to the wire of the microscope ; 

 but w'hen care is taken to use very pure mercury and clean tubes, 

 the operation can generally, after a little trouble, be accomplished 

 with much accuracy. M. Regnault, I believe, recommends that the 

 motion of the mercury should be regulated by the breath, a drying 

 substance being interposed to prevent moisture entering the tube. 

 This method was employed for some of the first instruments made 

 at Kew, but was abandoned in favour of the elastic bags. 



2. Graduation. — The measured lengths of the column of mercury 

 in its successive steps along the tube correspond to equal volumes. 

 Assuming tliat the calibre of the tube does not vary throughout the 

 small length of the calibrating column, if we divide the sj)aces oc- 

 cupied successively by the mercury into an equal number of parts, 

 it is evident that the divisions will represent the same capacity, 

 although they may be of very diiTerent lengths. Before making the 

 tube into a thermometer, the divisions of the scale may be verified 

 by introducing a longer column of mercury, and examining whether 

 the column occupies an ccjual number of divisions in different parts 



X2 



