Temperature and the Serenity of the Sky. 159 



adapted to the present inquiry. In addition to the blanks caused 

 by the cessation of observations on the Sunday, which could not, 

 in an investigation into lunar influence, be supplied by means 

 subjected to ordinary modes of correction, there were errors to 

 be allowed for which appeared inseparable from the measurement 

 of rapidly-passing and ever-changing bodies of vapour. No di- 

 stinction, also, could be made in the daily means between the more 

 dense and rarefied clouds. Notwithstanding this, the evidence 

 derived from the above observations is most impoi'tant. Fig. 1 

 is a curve formed from tables of the mean daily amount of cloud 

 for seven years at Greenwich, arranged in the same manner as 

 in the tables of mean temperature, which have been already de- 

 scribed, excepting that the sums of each vertical column were 

 divided by the number of observations actually recorded : they 

 varied from 70 to 76. This cloud-curve will be found to agree 

 in a remarkable manner with the curve of M. T. for ten years at 

 Greenwich (fig. 2), with which it accords most nearly in point of 

 time. 



10. Struck by the apparent regularity and boldness of the 

 lines of November temperature, I then formed a curve of the 

 mean amount of November cloud for the years 1841-48 at 

 Greenwich. It was formed in the same manner as the 7-year 

 curve, with this exception — that the arithmetic means of the 

 days immediately before and after the blank days in the tables 

 were from necessity introduced to make up an equal number of 

 seven observations in each column throughout the lunation. It 

 was found that the mean of the means of these columns was 

 the same as the mean amount of cloud for November, as given 

 in the results of the Greenwich observations, viz. 7*3. The curve 

 (fig. 6) which was then formed proved a perfect reflexion of the 

 mean temperature of November for forty years, as obtained from 

 the lunar tables. 



11. The results of the Greenwich observations for 1840-47 

 supplied a further test, or rather index of the serenity of the 

 sky at different periods in the lunation, and so of the eS'ects on 

 temperature jjroduced by an absence of cloud. On extracting 

 the cloudless days, or what might be considered cloudless days, 

 there enumerated, and arranging them in their proper position 

 on the lunar curves of mean temperature, in the three first years 

 of the above-named j)criod, out of twenty-three clear days it was 

 found that six occurred between new moon and first quarter, 

 eight before and after full moon, and five from the day of last 

 quarter to the fourth octant. Of the four other clear days, 

 three occurred in a single year (1842) on the second day after 

 first quarter, in the months of Aj)ril, July and August. The 

 amount of cloud in two of the exceptional instances was 0'3. 



