1869.] 



of Thermometers in a Vacuum. 



327 



1 1 . These readings tell all the same thing, and taken separately agree 

 closely with the mean result. If the result, found by another method, for 

 the thermometers tested for the vacuum- correction during the pendulum- 

 experiments for the Indian Survey, which gave o, 43, is added, it may be 

 stated, as first result of these experiments, that a thermometer of common 

 form and size will, if used in the vacuum of a receiver, require an addi- 

 tive correction of four-tenths of a degree of Fahrenheit'' 's scale, provided 

 that no readings are taken until the immediate effect of exhaustion, which 

 amounts to nearh/ tivo degrees, is equalized. 



12. The two large thermometers gave the following differences: — As 

 some of them are in an opposite direction, I denote the expected differences 

 by the sign + , and those on the wrong side by — . 



C— C. 



c— c. 



Experiment IV — o*o6 



O'OO 



V -o-o8 



+ 0'20 



VI -o-i7 



— O'OI 



Experiment I — o*o6 



O'OO 



„ II -0-05 



-f-o'o6 



III -o-o6 



+0-03 



These results only strengthen the validity of the others ; for obviously 

 we have, in regard to these large thermometers, in fact no other difference 

 but that arising from experimental errors, local currents, &c. 



The first explanation of this behaviour that suggested itself was, that 

 the thermometer which was supposed to be surrounded by air, had some 

 flaw in the glass envelope, which allowed the air to escape during the 

 pumping, so that there was really no difference of condition between it and 

 its companion thermometer. A most careful examination of the case did 

 not lead to the discovery of such a cause of leakage ; and as the thermo- 

 meter in the closed case lagged behind the other in the same manner as 

 the other thermometers in a similar condition did, I can only come to the 

 conclusion that thermometers with large bulbs and stems really behave 

 differently, or that the permanent effect of exhaustion is imperceptible. 



[With a view of determining whether the exceptional behaviour of the 

 large thermometers could be accounted for by greater strength of their 

 glass bulbs, Professor Stokes kindly suggested to me a comparison of the 

 relative thickness of the glass of the bulbs by placing on it a very minute 

 opaque dot, and measuring the apparent distance of the dot from its re- 

 flected image by a lens. I found the following results : — 



1st. The thickness of the glass varies not inconsiderably in different parts 

 of the bulb of one and the same thermometer. 



2nd. The thickness of the glass in the bulbs of the large thermometers 

 was, on the average, twice that of the small spherical bulbs. 



3rd. The thermometers with cylindrical bulbs had nearly three times 

 the thickness of those with small spherical bidbs ; this thickness, 

 however, was considerably less at the base of the cylinder. 



