109 
the months of a single year; and if their difference depends 
upon the relative amount of the humidity of the atmosphere 
in each month, then it is clear that no regularity for a single 
year is to be expected. But if an average of six years be 
considered or the average of any number of years above six, 
and the following table presents an average of 10 years, then 
a definite order is manifest. According to this order, Dec- 
ember and January are the most humid months of the 
year, and May and June are the driest. And if the theory 
be true as to the cause of these differences, then during the 
past year the atmosphere was driest in the following 
months, September, March, August, May, April, and Feb- 
ruary, and was most humid in December, January, Novem- 
ber, July, June, and October. The fall of rain in the 
months when the atmosphere was driest was 19'072 inches, 
and in the months when it was most moist was 26T03 
inches, being a difference of 7’Q31 inches. In the same 
table I have given the ratios of this difference or excess of 
rainfall for 10 years, and placed them in juxtaposition with 
the mean humidity of the atmosphere calculated from obser- 
vations of the dry and wet bulb hygrometer, by means of 
Glaisher’s tables. A glance shows that the results are 
identical, and points almost certainly to the cause of the 
excess of rainfall between the two gauges, namely, the mois- 
ture of the atmosphere. 
But that another supposed cause may be fairly discussed 
I have prepared a table which gives the ratios of the differ- 
ences of rainfall in comparison with the calculated amount 
of moisture in the air, and the ratios of the relative amount 
of the movement of the air for every day of rainfall for the 
past year. From this table it will be seen that almost in 
