Meteorology of New Haven. 



237 



month of one year, so unlike that of the same month in some differ- 

 ent year. 



If we make the comparison by seasons, tlie numbers are 



The final result is that the mean temperature of New Haven by 

 the last 45 years, is one-fifth of a degree lower than by the first 41 

 years ; but this quantity does not exceed the probable zero error of 

 most of the thermometers employed in the observations ; and we must 

 conclude that if the mean temperature of New Haven has changed 

 at all since 1778, the change amounts to only a small fraction of a 

 degree, and cannot be certainly decided from the observations. It iis 

 however, noticeable that the difference for the summer months is 

 quite large, considering the length of the period from which the re- 

 sult is deduced, and seems to indicate a slight moderation in the heat 

 of our summers. A careful examination of the preceding tables will 

 show a remarkable excess of high temperatures for summer at the 

 mid-day observation during the years from 1778 to 1803, from which 

 it is reasonable to conclude that the thermometers of Pres. Stiles and 

 Messrs. Beers and Howe were not as well protected from reflected 

 heat as were the thermometers in most of the subsequent observations. 

 Some influence may be ascribed to the increased number and size of 

 our shade trees, within the last half century, by which means ther- 

 mometei's are generally better sheltered from the rays of the sun re- 

 flected from buildings or a sandy soil. 



One of the most important features of a climate is its liability to 

 late frosts in spring, and early frosts in autumn. We have endeav- 

 ored to determine from the journals the average date of these frosts, 

 but it is obvious from an inspection of the records that the frosts in 

 many instances were not recorded. It is thought that we may obtain 

 more reliable data for comparison, if we fix upon a certain tempera- 

 ture as indicated by the observations of the thermometer. We have 

 therefore selected from the journals for each year the last instance 

 before midsummer, in which the thermometer sunk as low as 40°, and 

 whenever mention is made of a frost at that date, the record is cop- 

 ied in the following table. We have also selected from the journals 

 for each year, the first date after midsummer in which the thermom- 



