APPENDIX. COMPARISONS OF THE ANEROID BAROMETER. 411 



Upon the last comparison before departure, these two aneroids possessed almost 

 exactly similar index-errors ( + 0-172 and +0-182), and upon return to Guaya- 

 quil their index-errors were not far apart (— 0-341 and —0-291). It would 

 have appeared legitimate to conclude that their working had closely corre- 

 sponded, but inspection of the last two columns of the table shows that such 

 a conclusion would have been extremely erroneous. The case of E, taken by 

 itself, is still stronger. This, in course of time, ' returned ' almost perfectly ; 

 and inasmuch as this instrument (like all the others) was tested before de- 

 parture, inch by inch, against the mercurial barometer under the air-pump, 

 and corresponded almost perfectly, it would have seemed right to conclude 

 that its readings in the interim must have been nearly free from error. Yet 

 this instrument, at the greatest height at which it was compared, was found 

 to possess a minus error of an inch and a fifth, the value of which, at the 

 elevation in question, exceeds two thousand feet (see § 9). 



§ 18. Some of the more important conclusions which must be arrived 

 at from consideration of the results of these comparisons of the aneroid 

 against the mercurial barometer are so obvious that I consider it unnecessary 

 even to point them out ; and, in the remarks which follow, I endeavour 

 more to indicate the ways in which the aneroid may be advantageously 

 used, than to emphasize the objections which might be urged against its 

 employment. 



A. It seems possible, without reference to a standard, by intercomparison 

 of a number of aneroids, to discriminate between them, and to select those 

 in which most confidence should be placed. 



B. That, with aneroids of the present construction, it is unlikely that 

 decent approximations to the truth will be obtained at low pressures, even 

 when employing a large number of instruments. The errors of the whole 

 series (A — G) were invariably minus ones, and in the worst cases amounted 

 to as much as two inches upon the mercurial barometer. 



C. That differences of level at great heights (low pressures) may be 

 determined with considerable accuracy with aneroids, even when they have 

 acquired very large index-errors. 



D. That in observations of this description a nearer approach to the 

 truth is generally obtained by employing the mean of ascending and descend- 

 ing readings than by taking ascending or descending readings separately. 



E. That the test which is commonly applied of comparing for brief 

 periods (minutes or hours) aneroids against mercurial barometers under the 

 air-pump is of little or no value in determining the errors which will appear 

 in aneroids used at low pressures for long periods (weeks or months). 



F. That, similarly, comparisons of aneroids against mercurial barometers 

 in balloon for a brief space of time afford little or no clue to the errors 

 which will be exhibited by the former when subjected to low pressures for 



