of Dry and Moist Air. 29 



through the other gases, this superior propagation of heat can 

 only be a consequence of conduction. 



Another objection which Prof. Tyndall urges against my ap- 

 paratus is, that the pile is applied inside the experimental tube*. 

 He quotes an experiment in which he had cemented a thermo- 

 pile into the side of a tube, so that one side of it was inside the 

 tube and the other outside. When the air was now pumped out, 

 a very considerable current was produced. This result was what 

 might have been foreseen. But I may add that even when the 

 pile is entirely within the tube a current is produced on pump- 

 ing out the air, because the two sides of the pile and the neigh- 

 bouring parts of the apparatus are not cooled to the same extent. 

 The current, however, disappears entirely after a little time, 

 especially if the part of the tube containing the pile is surrounded 

 with water of constant temperature, as was always the case in the 

 apparatus that I used. 



When the pile is situated entirely outside the experimental 

 tube, the latter is cooled on being pumped out, just as though 

 the pile were in it ; this cooling, however, exerts a scarcely per- 

 ceptible effect on the pile applied externally, which proves that 

 the delicacy of the apparatus is now much less than it is when 

 the pile is placed entirely inside the tube. This inferior degree 

 of delicacy arises from the fact that the rock-salt plate, whereby 

 the tube is closed, absorbs, in proportion to its thickness and 

 transparency, a not inconsiderable amount of heat. When two 

 rock-salt plates are used, as in Prof. TyndalFs apparatus, the 

 delicacy of the apparatus becomes still less. 



By using two plates of rock-salt I have, even with my own 

 apparatus, found scarcely a perceptible difference in the amount 

 of heat transmitted across a vacuum and across a space filled 

 with dry air, a result which quite accords with Prof. TyndalPs 

 statements; when, however, the experiment was made without 

 rock-salt plates, the difference was quite decided. By the use 

 of the galvanometer mentioned above, I have convinced myself 

 that it is not quite so great as I had previously stated f; but it 

 amounts at any rate to several per cent. 



Por all gases, except atmospheric air, in the dry and moist 

 states, Prof. TyndalPs values agree with my own as nearly as 

 measurements of this kind could be expected to do. Hence the 

 method I have employed is at all events not so faulty as Prof. 

 Tyndall reproaches it with being ; for were his objections well 

 founded, the determinations of all the gases must have been 

 wrong. There must consequently be some other cause which 

 specially affects the determination of the absorptive power of 



* Phil. Mag. S. 4. vol. xxiii. p. 263. 



t Poggendorff s Annalen, vol. cxii. p. 524. [Phil. Mag. vol. xxii. p. 93.] 



