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



the allowance for the curvature of the mercury, alone would easily account for the 

 underestimate. Besides, it is to be remembered that the reading of the external o™™ 

 for 152 atm. is only about 22 mm. ; so that a slight variation of surface-curvature of 

 the mercury would of itself explain a considerable part of the half per cent, deficit. 

 It is, however, a matter of no consequence whatever, as regards the conclusions of that 

 Eeport. 



Buchanan, in the paper already cited, gives for the compressibility of water 

 at 2°"5 C. the value 0-0000516 ; and at 12°"5 C, 0-0000483. The empirical formula, 

 which is one of the main results of this Eeport (Section VII. below), extended to 

 p = Q, gives 0-0000511 and 0D000480 respectively. The agreement is very remarkable. 



Amagat's 1 investigations, which were carried out by means of the electric indicator 

 already alluded to (which informs the experimenter of the instant at which a given 

 amount of compression is reached), have been extended to pressures of nearly 20 tons 

 weight on the square inch (3000 atm.). As a preliminary statement he gives the 

 average apparent compression (per atmosphere) of water at 17 0, 6 C. as follows : — 



From 1 to 262 atm., ..... 0-0000429, 



„ 262 to 805 „ 0-0000379, 



„ 805 to 1334 „ ..... 0-0000332. 



And he states that, at 3000 atmospheres, water (at this temperature) has lost about 

 1/10 of its original bulk. But Amagat has not yet published any determination of the 

 compressibility of his glass, so that the amount of compression shown by his experi- 

 ments cannot be compared with the results of this paper. The rate of diminution of 

 compressibility with increased pressure, however, can be (very roughly) approximated 

 to ; and Amagat appears to make it somewhat less than I do. He operated on distilled 

 water, thoroughly deprived of air. My experiments were made on cistern water, 

 boiled for as short a time as possible. The analogies given in the present paper 

 appear to show that this difference of substance operated on may perhaps suffice com- 

 pletely to explain the difference between our results. 



I am indebted to a footnote in the recent great work of Mohn 2 for a hint which 

 has led me to one of the most singular calculations as to the compressibility of water 

 which I have met with. As it is given in a volume 3 whose very raison d'etre is 

 supposed to be the minutest attainable accuracy in physical determinations, I con- 

 sulted it with eagerness. The reader may imagine the disappointment with which I 

 found that, as regards compressibility of water, its main feature is the amazing empirical 

 formula, — 



501-53 - 1-58995/ 1 - 0003141113f 2 ! 



1 Comptes Rendus, torn. ciii. p. 429, 1886, and torn. civ. p. 1159, 1887. 



2 Den Norske Nordkavs-Exped., Nordkavets Dybder, &c, Christiania, 1887. 



3 Travaux et Memoires du Bureau International des Poids et Me'sures, torn. ii. p. D30, Paris, 1883. 



