49S TRANS. ST. LOUIS ACAD. SCIEN'CE. 



means of which are printed in full-face type, to distinguish them. 

 It will be seen, however, that the means of these periods do not 

 progress, rise and fall, much more evenly than the single daily 

 means ; compare, e.g., the mean of Jan. 31st to Feb. 4th. which 

 is so much lower than the mean of the foregoing or the following 

 five days that one might suspect a regular and normal decline of 

 temperature in these days, and not a mere accident. 



A few facts must strike every one who examines the ta- 

 bles. The first is, that a time e\en as long as 47 years fails to 

 give us anything approaching absolute and reliable means ; and 

 we come to the painful conclusion, that observations even contin- 

 ued for double that time, or for a century, may not yet obtain that 

 desirable object. It seems that the excessive extremes of one or of 

 a few days such as we often observe in our climate of extremes, 

 especially in the winter season, will influence — or, I ma}' say, 

 vitiate — the means of a long series of observations ; and the ques- 

 tion with me arises, whether such extremes ought not to be elim- 

 inated from the series, and thus truer means be obtained. At the 

 same time we may justly be astonished that from such heteroge- 

 neous data so much order and system result — which gives us 

 hope that we cannot be quite on the wrong track. 



Another fact, which strikes us in looking over the tables, is 

 that the mean temperatures do not increase and decrease evenly, 

 but sometimes quite rapidly, and at other times they may become 

 almost stationary for a time. These points come out most strik- 

 ingly on a diagram which embodies the essential parts of the 

 results and shows the daily progress of the temperature. Thus 

 we find very little change from the middle of December to the 

 first part of Februaiy, though the temperature proves to be low- 

 est from January 4th to 13th ; then we notice a rapid rise from 

 Feb. 6th to 3oth, a slower rise to the middle of March, then a 

 rapid one to the end of the month ; in the forepart of April a slow 

 and after the middle of that month a very marked one ; then fol- 

 lows a tolerably even, at last quite slow% rise to July 9th, when 

 between this date and the i8th the greatest elevation of the curve 

 is obtained. After that the mean temperature falls slowly to the 

 middle of August, followed by a more rapid decline to ihe end 

 of September ; after a slight pause in the first week in October, 

 a more rapid fall takes place for the following two weeks and a 



