at D E, of such a length and at such a height above the cistern, as to include 

 all that space through which the level of the mercury in the tube usually va- 

 ries in the place in which the barometer is intended to be used. In these 

 countries the level of the mercury never falls below twenty-eight inches, nor 

 rises above thirty-one inches; consequently a space somewhat exceeding these 

 limits will be sufficient for the opening D E. The tube is permanently con- 

 nected with the cistern A B, and a scale is engraved upon the brass tube, near 

 the aperture D E, to indicate the fractions of the height of the mercury in 

 the tube. 



There is another method of avoiding the difficulty arising from the change 

 in the level of the surface of the mercury in the cistern, used in the barometer 

 hero represented. The bottom of the cistern moves within it in such a man- 

 ner as to prevent the mercury from escaping, and a screw is inserted at 

 B, by turning which the bottom of the cylinder is slowly elevated or 

 depressed. An ivory index is attached to the top of the cylinder, which is 

 presented downward and brought to a fine point, so as to mark a fixed level. 

 When an observation is made with the barometer, the screw V is turned until 

 the surface is brought accurately to the point of the index, by raising or low- 

 ering the bottom according as the surface is below or above that point. It fol- 

 lows, therefore, that whenever an observation is made with this instrument, the 

 surface of the mercury always stands at the same level, and therefore the di- 

 visions upon the scale C F, represent the actual change of height in the bar- 

 ometric column. 



Since the column of mercury sustained in the barometric tube is taken to 

 represent the pressure of the atmosphere, it is clear that no air or other elastic 

 fluid should occupy the part of the tube above the mercury. To avoid such a 

 cause of error is not so easy or obvious as may at first appear. Mercury, as it 

 exists in the ordinary state, frequently contains air or other elastic fluids com- 

 bined with it, and which art maintained in it by the atmospheric pressure, to 

 which it is usually subject. 



When it has subsided, however, in the barometric tube, it is relieved from 

 that pressure, and the elastic force of such air as may be lodged in the mercu- 

 ry, being relieved from the pressure which confined it there, it will make its 

 escape and rise to the surface, finally occupying the upper part, of the tube, and 

 exerting a pressure upon the surface of the column by means of its elasticity. 

 Such a pressure will, then, assist the weight of the column of mercury in bal- 

 ancing the atmospheric pressure, and consequently a column of less height 

 will balance the atmosphere than if the upper part of the tube were free from 

 air. To remove this cause of error it is necessary to adopt means of purify- 

 ing the mercury used in the barometer from all elastic fluids which may be 

 combined with it. 



The fact that the application of heat gives energy to the elastic force of gas 

 ' es, enables us easily *u accomplish this. For if the mercury be heated, the 

 | particles of air or other elastic fluids which are combined with it acquire sucU 

 1 a degree of elasticity that they dilate and rise to the surface, and there escape / 

 | in bubbles. The same process of heating serves to expel any liquid impurities I 

 i with which the mercury may be combined. These are converted into vapor > 

 [ and escape at the surface. 



The presence of an elastic fluid at the top of the tube is thus removed so far j 

 | as such fluid can proceed from the mercury. But it is also found that small par- ( 



> tides of air and moisture are liable to adhere to the interior surface of the > 

 I glass ; and when the mercury is introduced, and a vacuum produced at the top c 



> of the tube, these particles of air dilate, and rising, lodge at the top and vitiate / 

 | the vacuum which ought to be there ; the particles of moisture also evac rate ;' 



