1869.] 



of Thermometers in a Vacuum. 



325 



stant inrush of air, which vitiated the ultimate results. Only those expe- 

 riments are here given and discussed which were made in a smaller receiver 

 expressly constructed for my purposes by Mr. Casella. 



In this Table the corrected means of the individual observations are 

 given, while a larger Table, embodying also the latter, has been deposited 

 with the Royal Society for future reference. It is seen from this larger 

 Table that the average amount of error in these observations is not more 

 than about two-hundredths of a degree of Fahrenheit. In a very few cases 

 only, where the thermal effect was not quite completed when the readings 

 were taken, errors of about one-tenth of a degree occur ; care, however, 

 was taken in these solitary cases to ascertain the completion of the effect 

 by the more close agreement of a new series of observations. 



9. A glance at the preceding Table will at once show that the immediate 

 effect of exhaustion is a fall, that of readmission of air a rise of all ther- 

 mometers, and that there is at once a difference in the behaviour between 

 the thermometers A', B r , C, which are still surrounded by air, and A, B, C, 

 which are in a vacuum. But this difference is also observable to a certain 

 extent when the receiver is refilled, and when, as regards external pressure, 

 all thermometers are in the same condition ; hence this immediate difference 

 must have another cause than the supposed change in the capacity of the 

 instruments ; at any rate if a permanent difference is found afterwards in 

 consequence of such a change, it must be included in that difference which 

 shows itself immediately. The cause of the latter is obvious. The ther- 

 mometers in closed cases lag a little behind when they are affected by such 

 sudden fluctuations as those produced in these experiments, and they 

 assume, as the experiments have shown, the normal temperature a little later. 



The following Table gives the immediate fall and rise of all thermometers, 

 observed respectively on evacuating and refilling the receiver, and the im- 

 mediate mean difference between the differently placed thermometers. It 

 exhibits a very close agreement between the effect of exhaustion and that 

 of readmission of air ; but its more important practical purpose is to show 

 that an error of nearly two degrees of Fahrenheit is made in thermometer- 

 readings in a receiver immediately after exhaustion or readmission of air. 



Immediate effect of exhausting the Receiver. 



Thermometers falling. 



"~A. A'. B. B'. # C. C. 



. o o o o o o 



Experiment 1 1-84 0-91 177 1-05 v6i 1*34 



II r86 070 i*55 079 1-52 ri8 



III 179 C97 1*65 0-95 1-59 1-31 



rV r 75 °'9* VS7 0-97 1-52 1-23 



„ V 2-39 1*17 1-95 1-22 i-86 1-48 



VI 2-io 0-96 1*67 i-o2 i-68 1-29 



Means 1*95 0-94 1*69 i-oo 1-63 1*30 



Differences immediately 1 

 observable J 



