THE RETURN OF DAYLIGHT. 273 



but though this might be practicable for a week or two, 

 it cannot be entertained where months have to be taken 

 into consideration. 



Our mercurial thermometer is graduated to minus 

 49° ; but as mercury freezes, or is said to freeze, at 

 minus 39°, it is questionable whether its readings below 

 minus 39° are reliable. At all events, as its reading, 

 hour for hour, is lower than a spirit thermometer placed 

 alongside it, its reading is logged as a nearer approach 

 to tbe correct temperature so long as it is at or above 

 minus 49°. Below this point it suddenly contracts and 

 falls into the bulb, and there I presume freezes solid. 

 After that moment the spirit thermometers are perforce 

 read and logged. To-day, at the beginning, when the 

 mercury read minus 49°, the spirit thermometer read 

 minus 47°. At one A. m. the spirit thermometer read mi- 

 nus 48.5°, and soon after falling to minus 50° it finally 

 reached minus 53.5°. Before leaving New York, at 

 Collins' request, I directed Green to make thermome- 

 ters with bulbs of the prismatic colors, but, unfortu- 

 nately, in transportation to San Francisco, four of the 

 seven were broken, leaving us only red, violet, and 

 black. The object of these thermometers (filled with 

 uncolored spirit) was to determine the effect of the 

 sun's rays acting through prismatic colored bulbs, and 

 so obtain a scale of absorption. One of these (the vio- 

 let) was exposed to the air to-day, and when our ordi- 

 nary spirit thermometer read at midnight minus 53°, 

 this violet bulb read minus 47.5°. As this one has 

 agreed very well with our standard mercurial at read- 

 ings above minus 49°, it is possible that its present 

 reading is nearer the correct temperature than that 

 of the ordinary spirit. 



During the last few days I observed that on the port 



18 



