210 ■ Love and ^nieal : 



completely carried out ; it was replaced by another one, whose 

 bulb was similar but shorter, no. 4 in the tables. The mate- 

 rials were cut from samples of fine Japanese silk, fine linen 

 cambric, and book muslin of open texture ; the pieces were 

 thoroughly cleansed by boiling three times in succession, in 

 caustic soda solution and in distilled water alternately. They 

 were then sewn firmly round the bulbs, one thickness only, 

 the rest hanging below. 



The dry thermometer was also of Jena glass, similar to the 

 others ; it formed part of the condensation hygrometer used as 

 a control. This was a Regnault double hygrometer, consisting 

 of two precisely similar thimbles, one of which contains ether, 

 and through both of which a steady current of air is drawn by 

 an aspirator. Exception has sometimes been taken to the Reg- 

 nault hygrometer on the ground that the silvered surfaces 

 cannot be kept polished, being attacked by the gases of the 

 air. This difficulty is avoided by using nickeled surfaces, 

 which are foimd to keep a splendid polish until the nickel 

 wears off. The curved surface is a slight disadvantage, but 

 experience shows that it is of no consecpience. The dew-point 

 thermometer was of English glass, graduated in fifths of a 

 degree Centigrade. The following table gives the dimensions 

 and range of the several thermometers : — - 



Bulb. 10 decree Rari^e. 



Tliey were calibrated by direct comparison with a standard 

 Negretti and Zambra, No. 65043, tested at Kew in March, 

 1888. The scale errors of this standard were all zero, except 

 — 0.1° at 100^ C. The zero error, determined immediately 

 before the calibration, was 0.2^. The scale errors of the 

 other thei-mometers were in all cases small, those of tlie 



