PHYSICAL PROPERTIES OF WATER, ETC. 41 



This seems to throw light on the question raised in last section, and to show that 

 the main cause of the discrepancy between the first and second series of observations is 

 not due to a difference in the substance operated on. The constant difference of the 

 differences is due to such a cause, being at once traceable to the fact that the sea-water 

 put into some of the piezometers for the second series of experiments was taken from 

 the same Winchester quart bottle as was that with which they had been filled two 

 years before. During these two years the sea-water had probably, by evaporation, 

 become slightly stronger, and, therefore, less compressible. The change of com- 

 pressibility is less than 0'5 per cent, of the whole, and is therefore practically (as it is 

 in the third significant figure) the same for all three pressures. If we now look back to 

 the suggested explanations in last section, we see that the above remarks entirely 

 dispose of the fifth and sixth so far as fresh water is concerned, though the sixth, in a 

 modified form, has to do in part with the discrepancy between the two series of 

 observations on sea-water. 



To decide between the two series I made a new set of observations, employing the 

 two piezometers of large capacity spoken of at the end of Section III. These are 

 called M L and M 2 . On the first day of experimenting M x held sea-water from a 

 Winchester quart filled at the same time with the first, but which had remained 

 unopened. M 2 had fresh water. On the second day M 2 held sea-water, and M x fresh 

 water. The object of this was to discover, if such existed, errors in the calibration of 

 the piezometers, and then to eliminate them by a process akin to that of weighing with 

 a false balance. 



One of the ordinary piezometers (v), filled with fresh water, was associated with 

 the others as a check. I quote the results of one experiment only, made on the second 



day : — 



5/6/88 



5 9°-4 



422 

 5 



Thus we have the following comparison of estimates of true average compressibility 



for the first additional ton : — 



Fresh Water. Sea-Water. 



( 1st Series 474 434 



9°-4j2n,l „ 469 427 



(New „ 473 434 



A few of the experiments were not thoroughly decisive ; none were in favour of the 

 second series. This seems (so far as the first ton is concerned) to settle the question in 

 favour of the first series. 



The formulae (A) and (B) may therefore, for 1 ton at least, be regarded as 



(PHYS. CHE5I. CHALL EXI>. — rART IV. — 1888.) " 



