OBSERVATIONS OF THE EDINBURGH BOCK THERMOMETERS. 179 



deduced from them (Tables I. and II.). They have, however, been applied to the 

 weekly averages of twenty years used in the graphical drawing of the annual curves. 

 The corrections used for this purpose are the means of the corrections applied to the 

 old thermometers in the corresponding weeks over the ten years, 1837-46. The 

 similarity between the two sets of curves would seem to afford a sufficient assurance 

 that the corrections thus applied can be only inappreciably in error, allowance being 

 of course made, in comparing two curves, for the fact that the length of the longest 

 of the new tubes is only 250 inches, while that of the longest old one was 307 inches, 

 the other new tubes being also shorter than the corresponding ones of the old set. 

 This difference of length has, as has been said, but slight effect on the amount of the 

 correction to the readings, only about a maximum of 0"006 of a degree F. It has 

 however, a considerable influence on the form of the curve. As might have been 

 anticipated, a glance at Plates II. and III. will show, that in the new set the annual range, 

 is greater than in the old, but the retardation of epoch is less. While, then, it is to 

 be regretted that the corrections of the new thermometers cannot be determined with 

 the same order of accuracy and confidence with which those of the old set were worked 

 out by Prof. Forbes (Vol. XVI., Transactions, Royal Society, Edinburgh), it is reassuring 

 to be able to say, in the words of the late Prof. Piazzi Smyth, that " although these 

 new thermometers cannot compete with the old ones as instruments of the most 

 delicate Natural Philosophy chamber problems, I have been much pleased to find 

 that, step by step, they have shown their full sufficiency to keep up the differential 

 historical record of superannual cycles of temperatures." 



Having already no less than forty years' observations of the old set, with a complete 

 theory of their corrections, rigorously applied to each individual reading, and having 

 worked out the forms and the equations of the annual curves, and deduced therefrom 

 the conductivity of the rock, it would appear to be almost superfluous to attempt to 

 derive the same quantities from the twenty years' readings of the new set, when the 

 result, owing to the difficulty involved in the corrections, must necessarily be somewhat 

 less reliable. It was thought that this part of the work should be undertaken, however, 

 in the hope that a comparison of the results would form a more or less crucial test of 

 the propriety of the corrections which have been applied, and of the value of the new 

 set for carrying on the " historical record " alluded to by Prof. Piazzi Smyth. The value 

 of the conductivity of the soil can be derived either from the difference of the Napierian 

 logarithms of the amplitude at different depths, or from the difference of the epochs. Now 

 the corrections take the positive sign in the winter months and the negative sign in the 

 summer. Hence in the case of the deepest thermometers in each set, which have their 

 highest readings in winter and their lowest in summer, the range is increased ; whereas, 

 with the other three of each set the range is more or less diminished by the application 

 of the corrections. But the epochs are not similarly affected, because the corrections 

 follow a very smooth curve. Hence, if the two determinations of the conductivity by 

 means of the new thermometers agree fairly well with one another, and with the 



