PHYSICAL PROPERTIES OF WATER, ETC. 27 



were employed. They were all previously tested, up to about 2^ tons weight per 

 square inch, in my large apparatus. As I was otherwise engaged, Professor Chrystal 

 and Mr. Murray kindly undertook the deep-sea observations; and I have recently 

 begun the work of reducing them. 



' The first rough reductions seemed to show that my pressure unit must be some- 

 where about 20 per cent, too small. As this was the all but unanimous verdict of 

 fifteen separate instruments, the survivors of two dozen sent out, I immediately 

 repeated the test of my unit by means of Amagat's observed values of the volume of 

 air at very high pressures. The result was to confirm, within 1 per cent, the accuracy 

 of the former estimate of the unit of my gauge. I then had the manometers resilvered, 

 and again tested in the compression apparatus. The results were now only about 5 

 per cent, different from those obtained in the 'Triton.' There could be no essential 

 difference between the two sets of home experiments, except that the first set was made 

 in July, the second in November, — while the temperatures at which the greatest com- 

 pressions were reached in the ' Triton ' were at least 3° C. lower than those in the latter 

 set. Hence it seems absolutely certain that water becomes considerably more com- 

 pressible as its temperature is lowered, at least as far as 3° C. (the ' Triton ' temperature). 

 This seems to be connected with the lowering by pressure of the maximum density 

 point of water, 1 and I intend to work it out. It is clear that in future trials of such 

 manometers some liquid less anomalous than water must be employed. 



" Another preliminary result, by no means so marked as the above, and possibly to 

 be explained away, is that by doubling (at any one temperature) a high pressure we 

 obtain somewhat less than double the compression. This, however, may be due to the 

 special construction of the manometer, which renders the exact determination of the 

 fiducial point almost impossible." 



In the winter of 1882 and the succeeding spring, I spent a great deal of time in 

 trying to get definite results from the records of the "Triton" trials, and in making 

 further experiments on those of the specially prepared piezometers which had not been 

 broken or left at the bottom of the sea. But this work led to no result on which I 

 could rely. I then directly attacked the problem of the compressibility of water at 

 different temperatures and pressures, having once more verified the unit of my pressure 

 gauge by comparison with Amagat's data for air. Results for one temperature were 

 published, as below, in the Proc. Roy. Soc. Earn, vol. xii. pp. 223, 224, 1883. [The 

 mercury content of the bulbs of the new piezometers was about 200 grm., and that of 

 100 mm. of stem about 2 "6 grm.] 



" The apparatus employed was of a very simple character, similar to that which 

 was used last autumn in the ' Triton.' 



1 [The reason for this remark will be seen in the second extract in Section XII. below. 20/0/88.] 



