352 



MADE AT THE KEW OBSERVATORY. 



2. That in tliis respect small aneroids are somewhat less trustworthy than large 

 ones. 



3. That if the instrument read he previously tested and its coiTections ascer- 

 tained, we may consider it trustworthy (making use of these corrections) for up 

 readings throughout a greater range than if it had not been so tested. 



I come now to consider whether a rapid change of pressure affects an aneroid 

 after the experiment has been completed. 



The following table will exhil;)it the results obtained in this direction : 



Correction before experiment 



Immediately after experiment 



Eigliteen hours after experiment .. 

 Forty-ciglit hours after experiment 



Three days after experiment 



Three weeks after experiment 



Aneroid No. 



—.12 

 + .03 

 —.0.3 

 —.04 

 -.05 

 —.10 



—.12 



+.06 



'— .OS 

 —.11 



—.04 

 +.06 

 —.01 



—.01 

 —.07 



16. 



—.05 

 + .04 

 + .01 



—.01 

 —.06 



It thus appears that if an instrument reads correctly before it is put into the 

 receiver it will read too low immediately afterwards, and that it may be some 

 considerable time before it recovers its previous reading. The instrument cannot, 

 therefore, be safely trusted for absolute determinations if it has been recently 

 exposed to rapid changes of pressure. 



The experiments hitherto recorded, in which an inch of pressure has been taken 

 away or added every 10 minutes, are perhaps analogous to ascents in a balloon, 

 or descents from a mountain; they are not, however, precisely analogous to moun- 

 tain ascents, since a longer time than 10 minutes is usually taken to produce a 

 change of pressure equal to one inch. 



At the suggesticm of Mr. Charles Brooke, a couple of aneroids were tested in 

 April, 1868, with the view of rendering the experiment more analogous to a 

 mountain ascent. 



The pressure was reduced by half an inch at a time, and at intervals of 30 

 minutes, the aneroids being well tapped. 



The following corrections were obtained for down readings, (instruments sup- 

 posed right at 30 inches :) 



These results, when compared with the previous determinations for these same 

 instruments, would seem to show that a somewhat better result is obtained when 

 the exhaustion is carried on more slowly, and hence that the corrections depend, 

 to a considerable extent, on the nature of the treatment received. 



