DR BREWSTER on the Register of the Thermometer. 363 



In directing these observations, it became necessary to select 

 two hours of the day most convenient for marking the state of 

 the Thermometer, and the Mean Temperature of which ap- 

 proached nearest to the Mean Temperature of the day. The 

 hours adopted were 10 o'clock A. M. and 10 p. M., which had been 

 previously recommended by the Reverend Mr GORDON. The 

 observations were accordingly made at these hours, during 

 three years ; but it appeared to me, upon a more attentive consi- 

 deration of the subject, that the Thermometer should be obser- 

 ved at the two times of the day at which the Mean Tempera- 

 ture occurred ; for if one of the observations was omitted, the 

 other still possessed considerable value, as an approximation 

 to the Mean Temperature. Unfortunately, however, there were 

 almost no observations in existence from which the times of the 

 daily Mean Temperature could be deduced. Professor DEWEY 

 of New York had observed the Thermometer once every hour 

 during five days at a time, in the months of March, April, 

 July and October of the year 1816, and during eight days 

 of January, and two of February in the year 1817*; and 

 Mr COLDSTREAM of Leith registered the Temperature of 24 

 successive hours once every month, from July 1822 to July 

 1823. From this last series of observations the Mean Tem- 

 perature appeared to occur at \ past 7 o'clock in the morn- 

 ing, and \ past 8 in the evening ; and these hours were accord- 

 ingly used in most of the registers for 1824 and 1825. It was 

 very obvious, however, that these observations, though made 

 with great care, were too limited to afford any thing like an ac- 

 curate result ; and hence it became a desirable object to extend 

 the hourly observations of the Thermometer over a greater por- 

 tion of the seasons, or, if possible, to record its indications for 

 every hour of a complete year. 



* Memoirs of the American Academy of Arts and Sciences, Vol. IV. Part II. 

 p. 392. 



