38 THE VOYAGE OF H.M.S. CHALLENGER. 



seated by the formulae (A) above, provided the first terms be made 499, 488, 477 

 respectively, i.e. provided 5, 2, and 1 be subtracted from the numbers for 1,2, and 3 

 tons respectively. Thus, while the amount of the compressibility is reduced, it is 

 made to depend on temperature precisely as before, but the way in which it depends on 

 pressure is altered. The rate of diminution of compressibility with increase of pressure 

 is now made constant at any one temperature, instead of becoming slowly less as the 

 pressure is increased. This is incompatible with the results of all of the first series of 

 experiments. The total amount of the compressibility is likewise diminished, by 1 per 

 cent, at 1 ton, by - 4 per cent, at 2 tons, and by 0"2 per cent, at 3 tons. 



Small as these differences are, their regularity struck me as very remarkable, and 

 as pointing definitely to some difference of conditions between the two sets of experi- 

 ments. Now there were undoubtedly many circumstances in which the series of 

 experiments differed : — 



First. The observers were not the same. All the readings in the first series were 

 made by myself ; but (in consequence of an accident which prevented me from working 

 in the cellar) I was unable to take part in the second series, and the readings for it 

 were all made by Mr. Dickson. Thus there may be a difference, of personal equation, 

 in the mode of applying the scale to the stem of the piezometer, or in the final . 

 adjustment of the manometer. Such an explanation is quite in accordance with the 

 results, as a constant difference of reading would tell most when the whole quantity 

 measured is least, i.e. at the lowest pressure. But a difference of a full millimetre in 

 the piezometer readings may be dismissed as extremely improbable. 



Second. It is possible that, during the second series of experiments, less care may 

 have been taken than in the first series to let off the pressure with extreme slowness. 

 Thus the indices may have been slightly washed down, and the record of compression 

 rendered too small. Even with the greatest care, this undoubtedly occurred in some, 

 at least, of the experiments of the first series ; and the screw-tap may have been altered 

 for the worse during the second series. 



Third. It is recorded in the laboratory book that, during the second series of 

 observations (which were made for the most part in the exceptionally cold weather of 

 last spring) the oil and treacle in the manometer had become very viscous, so that 

 it was difficult to make the pistons rotate. As artificial cooling, of the pressure 

 apparatus alone, was employed in the first series, this objection does not apply to it. 

 A constant zero error of 4 mm. only in the gauge would fully explain the discrepancy. 

 And there was another cause which may have tended to produce this result, viz. the 

 oxidation of the mercury in the manometric column, which had soiled the interior of 

 the lower part of the tube, and thus made it very difficult to read the zero. 



Fourth. The piezometers had been twice refilled, and of course slightly altered in 

 content, between the two series, and the hair indices had necessarily been changed. 



