Full Moon on Cloud. 219 



small effect, but rather that the nights near the time of full moon 

 are much more clear than those on which no moon is present, 

 as, for instance, the nights near to new moon. Now the num- 

 bers which I obtained gave no support to this idea; for, from 

 the observations made during 86 \ lunations at Greenwich, the 

 mean amount of cloud, as deduced from six observations taken at 

 intervals of two hours on five nights at each full moon, was 

 found to be 6*48, similar observations on five nights at each 

 new moon giving 6*07 (a clear sky being represented by 0, and 

 a sky wholly cloudy by 10). And I cannot help thinking that 

 the cause of the apparent deception is as I have stated it in my 

 paper printed in your July Number. But Mr. Harrison suggests 

 that sufficient account has not been] taken of the effect of the 

 omitted (principally Sunday) observations. I may here state 

 that from examination I do not find the omissions to occur 

 in the irregular manner which Mr. Harrison has supposed. 

 Now what would be the effect of supplying all the omitted 

 observations ? how would the numbers which I have quoted 

 be affected ? Practically one-seventh part of the observations 

 are wanting, the omissions being pretty equally distributed be- 

 tween full and new moon. Suppose now each omitted obser- 

 vation to be supplied, and let means be taken on two supposi- 

 tions, first, giving the omitted observations the extreme value 

 0, and, secondly, the extreme value 10. Such means are in each 

 case the least possible and greatest possible mean values. But 

 the one-seventh part of the observations is in each case a consi- 

 derable number. It is therefore in a high degree probable that 

 the means of the one-seventh part in each case will lie between 

 the before-mentioned least possible and greatest possible mean 

 values. We may therefore substitute for the omitted observa- 

 tions (instead of or 10) the least possible and greatest possible 

 mean values. Means being again taken, it is found that the 

 numbers 6*48 and 6'07 could not be altered by more than 0*13. 

 Admitting that the numbers may be altered by this amount, and 

 supposing even that the first is diminished and the second in- 

 creased, we should still have 6*35 for the mean state of the sky 

 at night at full moon, and 6'20 for its mean state at night at 

 new moon. These numbers not only negative any great effect, 

 but also the existence of any full-moon influence. 



Mr. Harrison further suggests that, by grouping several days 

 together, I have " merged and eliminated any distinctive character 

 they may individually have possessed." But I was not looking 

 for small effects. The sky was said to be much clearer at night 

 at the time of full moon. The seven years' Greenwich observa- 

 tions show that such peculiar effect does not exist. And this is 

 all that I wished to prove. 



Q2 



