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



The water in the compression apparatus, even when the large one was used, slowly 

 changed in temperature from one group of experiments to the next : — sometimes 

 perceptibly during the successive stages of one group. The effect of this source of error 

 was easily eliminated by means of the rough results of a plotting of the uncorrected 

 experimental data. From this the effect of a small change of temperature on the 

 compressibility at any assigned temperature was determined with accuracy far more 

 than sufficient to enable me to calculate the requisite correction. This correction was 

 therefore applied to all the experimental data of each group, for which the temperature 

 differed from that at the commencement of the group. The corrected numbers were 

 employed in the second and more complete graphical calculation. I endeavoured to 

 raise the pressure in each experiment as nearly as possible by 1, 2, or 3 tons weight per 

 square inch : — having convinced myself by many trials that this was the most convenient 

 plan. The cure for any (slight) excess or defect of pressure was at once supplied by the 

 graphical method employed in the reductions, in which the pressures were laid down as 

 abscissas, and the corresponding average compressibilities per atmosphere as ordinates. 



When this work has been fully carried out, we have still only the apparent com- 

 pressibility of the water or salt-solution. The correction for the compressibility of 

 glass, which is by no means a negligible quantity, — being in fact about 5 per cent, of 

 that of water at 0° C, —involves a more formidable measurement than the other ; but I 

 think I have executed it, for two different temperatures, within some 2 per cent, or so. 

 The resulting values of the true compressibility of water may therefore err, on this 

 account, by O'l per cent. This is considerably less than the probable error of the 

 determinations of apparent compressibility, so that it is far more than sufficient. With 

 a view to this part of the work the piezometers, whether for water or for mercury, were 

 all constructed from narrow and wide tubes of the same glass, obtained from one 

 melting in Messrs. Ford's Works, Edinburgh ; while solid rods of the same were also 

 obtained for the application of Buchanan's method. 1 



My results are not strictly comparable with any that, to my knowledge, have yet 

 been published, except, of course, those which I gave in 1883 and 1884. The reason 

 is that the lowest pressure which I applied (about 150 atmospheres, or nearly one ton 

 weight per square inch) is far greater than the highest employed by other experimenters, 

 at least for a consecutive series of pressures. I must except, however, the results of 

 Perkins and some remarkable recent determinations made by Amagat. 2 Perkins' results 

 are entirely valueless as to the actual compressions, because his pressure unit is 

 obviously very far from correct. They show, however, at one definite temperature, the 

 rate at which the compressibility diminishes as the pressure is raised. Amagat's 

 work, on the other hand, though of the highest order, is not yet completed by the 

 determination of the correction for the compression of the piezometer. 



1 Trans. Roy. Soc. Edin., vol. xxix. pp. 589-598, 1880. 2 Complex Rendus, torn, ciii., 1886, and torn, civ , 1887. 



