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



The mean thus obtained coincided very closely with the mean of all the experiments. 

 Hence the average linear compressibility per atmosphere for the first ton is, at 9° - l C, 



°' 0102 „ = 0-000000884 



152-3 x 75-75 



whence the compressibility of glass is 



0-00000265 



The two series of experiments agreed fairly with one another, and appeared to 

 show an increase of compressibility with rise of temperature, and a diminution with 

 rise of pressure, but these are not made certain. Considerably greater ranges, both of 

 pressure and of temperature, are necessary to settle such questions. 



As I cannot trust to a unit or two in the last place, (i.e. the seventh place of 

 decimals), my results for the apparent compressibility of water, and as an error of 

 reading of the external gauge may easily amount to 1 per cent, of the whole ton 

 applied, I have taken from the above experiments the number "0000026 as expressing 

 with sufficient accuracy the compressibility of the glass of the piezometers throughout 

 the range of temperature 0° to 15° C, and of pressure from 150 to 450 atm. This 

 number is simply to be added to all the values of apparent compressibility. Had I 

 pushed the pressures farther than 450 atm., this correction would have required 

 reduction, as shown in Appendix I). 



vi. resume* of my own experiments on compression of water 



and of Sea-Water. 



The following details are, where not otherwise stated, taken from my laboratory 

 books. I was led to make these experiments by the non-success of an attempt to 

 determine the exact unit of the external gauge (described in my former Report). Not 

 being aware of the great discovery of Canton (in fact, having always been accustomed 

 to speak of the compressibility of water as 1/20,000 per atm.), I imagined that I 

 could verify my gauge by comparing, on a water piezometer, the effects of a pressure 

 measured by the gauge with those produced by a measured depth of sea- water, without 

 any reference to the temperatures at which measurements were made ; provided, of course, 

 that these were not very different. The result is described in the following extract : ' — 



" To test by an independent process the accuracy of the unit of my pressure gauge, 

 on which the estimated corrections for the Challenger deep-sea thermometers depend, it 

 was arranged that H.M.S. 'Triton' should visit during the autumn a region in which 

 soundings of at least a mile and a half could be had. A set of manometers, filled' with 

 pure water, and recording by the washing away of part of a very thin film of silver, 



1 Proc. Roy. Soc. Edin., vol. xii. pp. 45, 46, 1882. 



