260 On the Haiti and Drainage - Waters at Rotliamsted. 
These small deposits, condensed from the lower stratum of 
the atmosphere, contain on an average three or four times the 
amount of organic carbon, organic nitrogen, ammonia, and nitric 
acid, found in the analyses of rain-water. The total quantity of 
solid matter, and the amount of chlorides, is also larger, but the 
difference is much smaller than in the case of the other ingre- 
dients. The mean proportion of organic nitrogen to carbon 
is 1 : 3-5. 
We must now consider briefly the variations in the com- 
position of the Rothamsted rain-waters displayed in Dr. Frank- 
land's analyses. The extent of variation in 69 samples has 
been already given in Table XI. A glance at this Table will at 
once show that the amount of variation is enormous ; and the 
range of difference becomes still larger if we include in the 
same view the small deposits of dew and hoar-frost. 
As the composition of rain-water greatly depends on the 
quantity of the fall, we shall in the first place classify the 
analyses according to the quantity of rain which they represent, 
and next according to the season of the year in which they fell. 
Only fifty-four of the samples analysed by Dr. Frankland fairly 
represented a known quantity of rainfall, but in the case of four 
other samples the bulk of the sample bore such a high pro- 
portion to the bulk of the rainfall that it may be accepted as a 
tolerably fair representation of the whole fall. From the whole 
number two analyses are omitted for reasons already given ; 
there remain therefore fifty-six available for discussion. 
In the upper division of Table XIII. the whole of the analyses 
of rain-water are arranged according to the quantity' of the fall 
which they represent. It is evident on considering this division 
that the proportion of each constituent tends to diminish as the 
amount of rainfall increases, the decrease being most rapid in 
the case of the chlorides, and least marked in the case of the 
organic elements. The quantity of nitrogen as nitrates and 
nitrites is frequently the mean of so few analyses that the figures 
are necessarily irregular. 
Since the quantity of the fall exercises such a preponderating 
influence on the composition of the rain-water, it is clear that 
in tracing out other conditions affecting the composition of the 
water we should only compare together analyses which represent 
rainfalls of similar amount, a truth which, though self-evident, 
has been very much overlooked in discussions of this nature. 
In the second and third divisions of the Table the analyses 
are still arranged according to the quantity of the fall, but they 
are now divided so as to show the differences between the rain 
of summer and winter. The series of analyses we are now dis- 
cussing is by no means favourable for the exhibition of differ- 
