ABSOLUTE THERMAL CONDUCTIVITY OF NICKEL. 



379 



above 220° C. The corrections at 0° C. were zero for most of the thermometers. One 

 read '55° too low, but that was due to a small particle of the mercury having been shaken 

 up into the top of the stem — probably during transit — from which it could not be again 

 dislodged. It was on the strength of these results that '0001 ISt 2 was used to give the 

 stem correction in the Forbes bar experiments. 



The following table gives the corrected mean readings obtained from the last three 

 experiments, together with the values of the conductivity calculated from them. 



Date of 

 Experiment. 



31/12/97 

 4/1/98 

 6/1/98 



Temp, 

 of Air. 



Corrected Mean Temperatures of Holes in Bar 



No. 1. No. 2. No. 3. No. 4. No. 5. 



14-8 

 13-2 

 133 



39-3 

 47-0 

 40-5 



74-1 

 79-0 

 62-3 



115-9 



118-9 



89-4 



168-1 

 170-0 

 123-9 



242-7 

 2432 

 170-3 



Temp, at 



End of 



Bar. 



28-5 

 37-7 

 34-3 



Gradient 



at End of 



Bar. 



366 

 3-23 



2-08 



Mean 

 Temp, of 



Water. 



16-8 

 221 



21-8 



Flow of 



Heat in 



Calories 



per Second. 



8-12 



7-225 



4-86 



Conduc- 

 tivity at 

 End of 

 Bar. 



■130 

 •131 

 •136 



§11. Theory of the Method. — Let K be the conductivity, 6 the temperature, X 

 he distance from some fixed point on the axis of the bar, of a section of the bar of area 

 i, across which H units of heat pass in unit of time, then 



ax 



dd 



Corresponding values of 0, H, and -j- are given in the above table for the end section of 



he bar whose cross-section is 17'1 square centimetres (diameter is 4'663 cms.). The 



alues of H given are subject to two corrections : (1) a correction for heat lost by radia- 



ton from the brass cap ; (2) correction for the changes in the thermal capacity of unit 



lass of water with temperature. An estimate of the former error shows that it never 



xceeded 1 per cent., so that it is probable that these corrections combined do not 



xceed 2 per cent. They are rather smaller than the corresponding corrections in a 



)ecific heat determination. 



dB 

 The values of -r- are liable to error from two sources : (1) thermometric errors in the 



imperature of the nearest thermometer hole ; (2) arithmetical or geometrical errors in 

 fferentiating the temperature curve. Errors from both of these causes would have 

 Jen reduced by having more thermometer holes, and what discordance there is between 

 ie values of the conductivity found from the three sets of readings given above is 

 'obably mostly due to errors in estimating d9/dx. Differences amounting to 2 or 3 per 

 mt. are only to be expected. All these sources of error effect Forbes's method, — 

 jid, of course, also Angstrom's — but to these are added in Forbes's method all those 

 ising from the cooling experiment. 



The measurements referred to only determine the conductivity at temperatures some- 

 what above that of the air, but the conductivity could be found in a similar manner at 

 <jher temperatures (such as slightly over 100° C, by allowing the water in the cap to be 



