To investigate the point more closely, a further modification was introduced at 

 the commencement of series " C " by the use of three thermometers placed in different 

 positions. It was at once observed that the mean reading of the two thermometers 

 in air was by no means the same as the reading of the thermometer in the dummy 

 pendulum. With rising temperature, the latter gave a lower reading than the mean 

 of the air thermometers, not the reverse as would have been the case if the dummy 

 pendulum was in metallic connection with the base of the instrument. It was also 

 interesting to note that the considerable temperature gradient from floor to roof in 

 the hut was reflected in a similar temperature gradient within the pendulum case, 

 sometimes as much as |- C. in 20 cms. vertical height. 



As an example of the readings obtained, the uncorrected temperature observations 

 of Series " D " are given in Table LXV, the corrections being, however, applied in the 

 last column. 



TABLE LXV. 



80 



