158 



MR THOMAS HEATH ON 



plan was not adopted with the new set, as it was thought possible that the hard material 

 had something to do with the unaccountable fracture in 1861 of the second longest of 

 the old set. After the new thermometers had been some time in position, it was noticed 

 that the sand was gradually settling down, and that the shorter thermometers were 

 slowly sinking. In 1880, as measured by Prof. Piazzi Smyth, their bulbs were then 

 below the surface of the rock, 



^ 250 inches, t 2 127*5 inches, t 3 57*0 inches, and t A 34"5 inches. 

 In May 1889 they were again measured by Prof. Copeland, and found to be below 



the surface, 



t 1 25 If inches, t 2 130-| inches, t s 58^ inches, t i 35^f inches. 



In August 1900 their positions were once more measured, and they appeared to 

 have sunk still further, their bulbs being then below the surface, 



t x 251| inches, t 2 133| inches, t s 60J inches, t 4 37J inches. 



Altogether from the date of their establishment in the rock up to last year, the 

 thermometers have sunk, 



t-y If inch, t. 2 8 \ inches, t 3 10^ inches, t± 12£ inches. 



Whether anything can be done, without risk to the safety of the instruments, to 

 replace them in their original position or to prevent them sinking further into the 

 bore-hole, is a question which will have to be taken into consideration at an early date. 



Plate I. gives a graphic representation of the rise and fall of temperature of the 

 four thermometers, as shown by the annual means taken quarterly. These are compared 

 with the shaded air temperature curve and the rainfall in Scotland, and with the sun- 

 spot curve. The last-named curve has been plotted from the sunspot numbers accord- 

 ing to the late P. Wolf, and A. Wolfer. It is remarkable that the series of maximum 

 temperatures shown for 1882 correspond to a period in the sunspot cycle which is further 

 removed from the sunspot minimum than has ever occurred with this series of observa- 

 tions, including those of both the old and new thermometers. Another remarkable 

 feature is a dip in the curve at 1898. which goes through all the thermometers, affecting 

 even t 1 to a slight extent. It is no doubt connected with the rise in the sunspot curve 

 in the same year. The air temperature and rainfall curves have been laid down from 

 the table printed on page 186, the numbers used being the quadruple annual means for 

 quarterly periods. The sunspot curve has been plotted from the following table : — 



SUN-SPOT NUMBERS ACCORDING TO R. WOLF AND A. WOLFER. 



1877, . 



. Ill 



1883, . 



. 65-3 



1889, . 



. 6-3 



1895, . 



64-0 



1878, . 



. 3-8 



1884, . 



. 633 



1890, . 



. 71 



1896, . 



41-8 



1879, . 



. 77 



1885, . 



. 49-9 



1891, . 



. 35-6 



1897, . 



262 



1880, . 



. 31-5 



1886, . 



. 26-1 



1892, . 



. 730 



1898, . 



26-7 



1881, . 



. 54-4 



1887, . 



. 13-5 



1893, . 



. 84-9 



1899, . 



12-1 



1882, . 



. 58-1 



1 888, . 



. 67 



1894, . 



. 78-0 







