Tempei'ature and the Serenity of the Sky. 157 



with the remaining mean temperatures. In consequence, how- 

 ever, of the ever-varying position of the moon in her orbit, and 

 the greater or less speed with which she happens to be travel- 

 ling, there continually occur an unequal number of observations 

 between the quarters, equally distributed in a loug series of 

 years over the whole lunation. For example, in the case of the 

 first lunation in the present year (1859), new moon falling on the 

 4th day of January at 5'^ 2^^ a.m., the first quarter occurred 

 on the 12th day at 7^' 2201 a.m., and full moon at 11^^ 48'^ pm. 

 on the 18th day. It follows that there would be seven observa- 

 tions of mean temperature between new moon and first quarter, 

 and five only between first quarter and full moon. In the latter 

 case it is presumed that the deficiency would be not improperly 

 made good by repeating the observations of mean temperature 

 of the third day after first quarter, or the third day before full 

 moon, so as to complete the full number of observations in all 

 excepting the octant columns. The means of these columns 

 were not used in forming the curve of November temperature, 

 or the curves of yearly mean temperature, which were derived 

 from tables constructed in a similar manner to the one above 

 described. 



h I In the curve of M. T. for forty Novembers (fig. 5), which 

 was formed from this table, it will hardly be necessary for me to 

 draw attention to the remarkable alternation of high and low 

 temperature which prevails at regular intervals through the 

 lunation. Contrasted with the two curves of ten years' jM. T., 

 the first lialf follows the outline of the corresponding half of the 

 curve fig. 3. The other half, from the day before full moon, 

 very nearly resembles the corresponding period in the curve 

 fig. 3. The principal difference consists in the excess of effects 

 produced; and I cannot doubt that the more proaounced cha- 

 racter of the November curve, as compared with the line of the 

 10-year means, is due to the absence of the counteracting in- 

 fluences in other months, and of the action and reaction which 

 occurs in successive lunations, though the numerical difference 

 of the observations from which the means were deduced must 

 also be taken into account. 



7. That very sharp frosts occur on days when the sky is clear 

 requires no demonstrative i)roof; and so in summer nights 

 the disappearance of clouds, or prevalence of blue sky, at the 

 period of full moon, which appears to be established as a fact 

 on the evidence of Humboldt, Sir John Herschcl, and other 

 astronomers of eminence, might, it was seen, have afforded a 

 physical explanation of the depression in the curve at the 



ill eatli column represent the mean temperature of more than a single d;iy 

 of the lunation. 



