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



index is, and is not, in contact with it. The capillary convexity affects the maximum 

 and minimum indices in opposite ways. 



Further, I may observe (though it does not affect my work) that in these 

 thermometers the scale is at some distance from the mercury in the stem, and no 

 provision is made for avoiding parallax or personal equation. By merely altering the 

 position in which one holds the thermometer, it is possible to read the temperature 

 whether by the mercury column or the end of the index next it, to an amount different 

 in some of the thermometers by as much as a quarter of a degree, and in the great 

 majority of them by as much as a tenth. Thus if we get readings consistent within a 

 tenth of a degree we get all that the instruments are capable of furnishing. I have 

 therefore always read the thermometers in exactly the same position and (when so much 

 accuracy was attainable) only to the nearest tenth of a degree. And I have always made 

 my comparisons between successive positions of the index ; the only readings of the 

 mercury directly being taken roughly to find whether any permanent temperature-change 

 had been produced in the water of the press by pressure or otherwise, during the course 

 of an experiment. 



A great many different materials were tried for the framing of the thermometers : 

 and vulcanite was finally chosen, having been found to answer the purpose exceedingly 

 well. Wood warped, and metal was unsuitable for various reasons. It is rather curious 

 to find, as will be seen below, that this substance was one of the main causes of the very 

 large amount assigned to the pressure-correction. 



Captain Davis' Mode of Testing ; and his Correction for the Maximum Side. 



It is necessary to look somewhat closely into the mode in which Captain Davis 

 conducted his experiments, in so far at least as it differs from the one I afterwards 

 employed ; in order that we may be able to form an idea how, with nearly all the facts 

 before him, he yet failed to get their proper interpretation. Take, for instance, the way 

 in which he attempted to determine the correction which is due to the heating of 

 water by compression. This, of course, affects the thermometers while in the hydrostatic 

 press, but not when they are let down into the sea. When the water in the press is 

 compressed with the thermometers in it, it becomes hotter as the pressure increases (so 

 long at least as its temperature is above 4° C. or 39°'2 Fahr., that of its maximum 

 density). This is quite analogous to the heating of air in a cylinder when a piston is 

 suddenly forced dow r n ; when, as every one knows, tinder can be kindled by the heat 

 developed. So water is heated by compression, but not to anything like the same extent. 

 But it is necessary to remark that the amount of heating of water by a given compression 

 depends in a very curious manner upon the original temperature of the water. For 

 water taken at its maximum density is neither heated nor cooled by compression, but it 



