On the Rain and Drainage - Waters at Rothamsted. 203 
only compare rains of similar quantity, and falling in a similar 
season of the year. This can hardly be done to any practical 
extent with the small series of analyses now before us. The 
chief points insisted on by Dr. Frankland, namely, that the 
south-east wind produces rain richest in ammonia, and the 
north-east wind rain richest in chlorine, are probably correct. 
London lies to the south-east of the rain-gauge ; the south- 
east wind, which is naturally rich in ammonia, may probably 
derive some addition to its contents from this source. To 
the north-east of the gauge lies the nearest wide expanse of 
ocean — the North Sea ; it is easy therefore to understand that 
winds from this direction should supply the largest proportion 
of chlorides. 
The last investigation on the Rothamsted rain-waters which 
we have to mention is one now in progress respecting the quan- 
tity of chlorides present in the annual rainfall. This investiga- 
tion commenced in June 1877, and has been continued down to 
the present time. A proportion of each day's rain, at the rate 
of one gallon for every inch of rainfall, has been set aside in a 
glass carboy provided for the purpose. At the end of each 
inonth the contents of the carboy are well mixed, and a sample 
is taken for analysis. 
The chlorine has been determined by the volumetric method 
employed by Dr. Frankland. The amount of chlorine present 
being often extremely small, and the results obtained when 
working on the unconcentrated water depending greatly on the 
conditions of the experiment, the greater part of the deter- 
minations has been made on water concentrated in a glass basin. 
One litre of rain-water, with 10 cubic centimetres of lime-water 
(since May 1880, 5 cubic centimetres have been employed), have 
been evaporated over a gas burner to rather less than one-quarter 
litre, then filtered, and the clear liquid diluted with distilled 
water till exactly one-quarter of a litre in volume ; in this solu- 
tion chlorine has then been determined by the method already 
mentioned. This mode of working gave much sharper results. 
We have, however, quite recently found that the chlorine deter- 
mined on this plan is somewhat below the true amount. We 
therefore now concentrate two or three litres of the rain with a 
little lime-water to a small bulk, filter, precipitate with nitrate 
of silver, and collect and weigh the precipitate. Eight monthly 
determinations of chlorine in rain-water made by this gravi- 
metric method have given a mean of 2*89 parts of chlorine per 
million, while the results of the volumetric method for the same 
months show a mean of 2*73 of chlorine per million. 
In the following Table will be found the monthly determina- 
tions of chlorine made by the volumetric method. The results, 
