REPORT ON THE PRESSURE ERRORS OF THE THERMOMETERS. 11 



minimum index, they were subject, according to my interpretation of Captain Davis' 

 results, to a correction of very nearly half a degree Fahr. for every mile of depth. 



Now, even if the heating effect on the water in the press had been correctly deter- 

 mined, the result would have led to a deduction of at the utmost only about one-fourth 

 of the whole correction, thus still leaving a very formidable correction indeed. 



Theoretical Determination of the Direct Effect of Pressure. Experimental Verification. 



I therefore calculated the effect of pressure on a thermometer tube, assuming the 

 best data for the compressibility and the rigidity of glass. The investigation is given 

 in Appendix A to this paper. As the matter is of considerable importance, I have 

 developed the formula? sufficiently for application to any case of the kind which is likely 

 to occur. The result, so far as is required for the present argument, is that the internal 

 capacity of a glass tube (whose walls are thick in comparison with the diameter of the 

 bore) is reduced by about x^Vuth P^ f r eacn ton weight (per square inch) of pressure 

 applied from without ; the ends being closed. Hence, if such a tube be partly filled with 

 mercury, with an index above it ; the index should be displaced by lt) 1 00 th of the length of 

 the column of mercury for each ton weight of pressure applied to the outside of the tube. 



I tried the experiment with a thermometer tube, the length of the mercury column 

 being as nearly as possible a metre, and I found for every ton weight of pressure to which 

 the tube was exposed the index was displaced by one millimetre, the yuWtii part of the 

 length of the column precisely, being far more nearly than I had expected the result I 

 had already calculated from theory. Since, then, there is only a change of one-thousandth 

 in the length of the column, it is quite obvious that the amount of effect produced upon 

 the column of mercury in the Challenger thermometers (which is not above a sixth or a 

 seventh of a metre in length at the utmost), that is to say, the whole correction-difference 

 between the maximum and minimum indices is a matter of a sixth or seventh of a 

 millimetre ; or in general very nearly the same fraction of a degree of the scale. Thus 

 it is proved in two different ways that the correction supplied by the Admiralty, if it 

 is to be applied at all, ought to be applied almost in its entirety to the minimum index. 



Tltf Aneurisms. Their Object and Effects. 



There is another peculiarity of the Challenger thermometers which leads to a slight 

 but only a slight modification of this statement, viz., that at the lower end of each 

 of the two vertical columns there is an aneurism on the tube. These form a sort of 

 secondary bulb, making the tube faulty again after the primary bulb has been protected. 

 Their effect is slightly to increase the effective length of the column of mercury. 



I learned from Sir George Nares that the object of these aneurisms, and of another 

 which is situated close to the bulb, is to prevent the indices from being jammed at the 



