316 Transactions of the South African Philosophical Society. 



These results are decidedly curious. At 2 p.m. there is a double 

 maximum and minimum during the course of the lunar month : 

 the first minimum falling just after new moon, and the second just 

 before full ; while the first maximum occurs at the first quadrature, 

 and the second maximum early in the last quarter. There are 

 also indications (which are, however, plainer in the daily averages- 

 of the large Table) of another maximum at full moon. The curious 

 thing about these minima and maxima is that they belong 

 to a daylight hour, and are therefore quite independent of moon- 

 light. 



At first sight the 8 p.m. results are pretty much what would be 

 expected on the assumption that the moon often makes thin clouds- 

 visible which would, without the help of the moon's light, often 

 escape observation. That is, we get considerably more cloudiness 

 between the third and eighteenth day than we do between the 

 eighteenth and third. For the moon sets before 8 p.m. (on an 

 average) in the first octant, and does not rise before 8 p.m. in 

 the sixth octant. But it is not so easy to explain the mini- 

 mum of the thirteenth day, nor the very pronounced full-moon 

 maximum. 



The mean results agree on the whole better with the 8 p.m. 

 averages than they do with the 2 p.m. averages. This is worth 

 attention, because they are derived from four daylight observations 

 and only two night observations. The minimum in the fourth 

 octant is particularly well marked as well as the full-moon maxi- 

 mum, both, as it happens, corresponding with the same turning- 

 points on Meissner's curve. In fact, the correspondences here 

 between the Potsdam and Kenilworth results are much more exact 

 than there seems to be any physical reason for. 



Assuming the cloudiness of the sky to be apparently less at night 

 when the moon is below the horizon than it ought to be, we ought, 

 to expect the greatest differences between the cloud percentages of 

 5 p.m. and 11 p.m. from the first to the sixth day, because the moon 

 sets before 11 p.m. ; and from the twenty-second to the thirtieth day 

 because the moon does not rise until after 11 p.m. To a certain 

 extent we get this result. But we get more. There is no sudden 

 decline in the value of the differences as soon as the moon is still 

 above the horizon at 11 p.m. ; on the contrary, the decline is gradual 

 until the tenth day — after which it is rapid. This is interesting, 

 because the light of the moon is considerable throughout the third 

 octant. But it is still more interesting that the value of the decline 

 runs up again to a maximum at full moon. This fact would doubt- 

 less have pleased Sir John Herschel, who fancied that he had 



