72 



with the present plan, the thermometer part may be made 

 most sensitive. It should not he forgotten that in these 

 severe climates the errors through bad position of the 

 instruments is often considerable, whereas with an electric 

 system they can be placed far from the houses. Therefore, if 

 with a metal thermometer, such as was used before, the error 

 should even rise as high as a half a degree Centigrade, and 

 a quicksilver thermometer placed near a house should give, 

 in consequence of the position, an error of two degrees C, 

 then we obtain the most reliable figures from the least 

 sensitive instrument. 



The following table shows the temperature as measured 

 in Davos Dorfli in my screen, fifty metres from the hotel. 

 The nine a.m. figures were always taken with a mercury 

 thermometer; at one p.m. usually with the electrical resis- 

 tance, but on a number of days I used the mercury ther- 

 mometer. Except on a few days in February, when a 

 friend took the observations, the seven a.m. and nine p.m. 

 were taken regularly with the electrical resistance. The 

 coil used was the one described as a spider's web, in which 

 the resistance was 450 ohms at — 7' Fahr., and 550 ohms 

 at 83 Fahr. 



As the whole thing was really in an experimental stage, 

 to see what could be done with the apparatus I possessed, the 

 results must not be looked upon as reliable within half a 

 degree Fahrenheit, but I think that they may be taken as 

 so far correct, 



7 A.M. 9 A.M. 1 P.M. 9 P.M. 



1884. Nov.*- 8-lC.(17-4F.)- 3-5 0.(257 F.) + 2.6C.(36-8F.)- 7-3 0.(18-9 P.) 

 ,, Dec. - 7-9 0.(17-8 P.)- 6-5 0.(20-2 F.)-l-l C.(30-0F.)- 7-80.(17-9P.) 



1885. Jan. -13-2 0.( 8-2P.)-ll-3 C.(ir7 P.)-l-9C.(28-6F.)-ir7C.(irOF.) 

 „ Feb. - 5-6 0.(21-9 F.)- 2-8 0.(27-0 F.) + 3-8 0.(38-9 F.)- 37 0.(25-4 F.) 



The results obtained are just about two degi'ees Cent, 

 colder than those published from Davos Platz, and ppssibly 

 Platz is a trifle colder, but the main difference must be 



* November, 1884, at 3 p.m. the mean was— 0-9 C. (30-4 F.) 



