1823.] the Barometer to the Measurement of Heights. 163 



stances are they made? Is the temperature of the mercury takeni 

 into account? How are the thermometers and hygrometers con- 

 structed, and how placed? And what is the system adopted foic 

 the reduction of observations to a mean ? Such questions are inr 

 dispensable, yet it is to be feared many observers could not 

 answer them in a satisfactory manner. We have a great num- 

 ber of barometric means collected from numerous observations, 

 and yet no one can say precisely what these means are. 



It is thus that many long series of observations are in reahty 

 lost to science, and only furnish illusory documents to the phi- 

 losopher who wishes to draw inferences from the experiments of 

 his predecessors. Let it be our endeavour that these losses 

 shall be the last science has to sustain, and let us furnish our 

 successors with points of comparison less equivocal. 



Choice of Instruments. 



The siphon barometer is preferable for stationary observations, 

 as it possesses the peculiar property of annulling by compensa- 

 tion the effects of capillarity. 



With cistern barometers the correction for capillarity must be 

 applied as given in the table. But besides this, these barome- 

 ters require the apphcation of some means for reducing the 

 level of the mercury in the cistern to the zero of the scale, a point 

 from which it is continually deviating, as well by the ascent and 

 descent of the mercury, as by the less apparent, though not less 

 real, effect of changes of temperature. This is provided for by 

 giving the cistern such a diameter that the variations of level 

 become nearly insensible. This, however, is not sufficient for 

 very exact observations. 



It is almost superfluous to remark the necessity of rigid accu- 

 racy in the division of the scale. In barometers mounted in 

 wood, the scale is generally marked on a plate of metal attached 

 to the mounting. This will not satisfy those who are desirous 

 of great accuracy. Heat, cold, moisture, dryness, affect wood 

 in every direction, sometimes increasing and sometimes diminish- 

 ing the distance of the scale from the point where its divisions 

 are supposed to commence. The scale, therefore, should be 

 entire from zero up to the point of the greatest elevation of the 

 mercury, though it may be divided only in that part through 

 which the range of variation extends. Copper is the best mate- 

 rial as being the metal with whose pyrometrical dilatation we are 

 best acquainted. With a scale thus constructed, we know 

 exactly to what the variations in dimension which result from 

 variations in temperature are reducible. They are regular and 

 very small, and may generally be neglected ; but we thus know 

 their amount, and can supply the correction. '. 



The vernier should indicate the 1-lOOOth of an inch ; and this 

 is nearly the utmost which the barometer is capable of express- 

 ing without ambiguity : the visual ray must of course be perpen-^ 



M 2 



