258 On the Bain and Drainage -Waters at Rothamsted. 
Two of the samples of water analysed by Dr. Frankland are 
excluded from the mean given in the foregoing Table, and will 
be excluded from all subsequent discussion. They represent the 
collections from January 25 to 31, and from February 9 to 16, 
1870. Hard frost occurred on both occasions, and hoar-frost 
or snow remained exposed for many days on the surface of the 
gauge, and doubtless became contaminated to an unusual extent 
with atmospheric dust, as the resulting samples of water proved 
verv impure. 
The average amount of ammonia found by Dr. Frankland is 
equal to only 0-37 of nitrogen per million ; this is far smaller 
than that obtained in the earlier analyses of Rothamsted rain- 
water, which gave a mean of 0*95 of nitrogen as ammonia per 
million of water. We have already stated that Dr. Frankland's 
results cannot strictly be compared with those previously ob- 
tained, since his samples of rain-water did not include every 
fall of rain in a given period, but consisted of selected samples 
only. So large a difference cannot, however, be explained by 
the different range of the samples analysed, and we must there- 
fore look a little farther. The methods of analysis employed at 
Rothamsted and by Way had probably a small error on the 
side of excess, due to alkali, liberated by the action of steam on 
the glass condensing vessels, being reckoned as ammonia ; frona 
this error Dr. Frankland's determinations made by the Nessler 
process would be free. Another fact to be taken into account 
is the interval which has elapsed between the collection and 
analysis of the water. Dr. Frankland's experience, we believe, 
points to the necessity of a speedy analysis, as the ammonia 
in a water is apt to diminish on keeping. To this source 
of error, tending to deficiency, nearly all the analyses we 
have recorded are more or less liable. It becomes thus impos- 
sible to judge with certainty between the results of the different 
experimenters. The only mode for determining with exactness 
the proportion of ammonia in rain-water is to make the deter- 
existing in extremely minute quantity. Dr. Frankland has kindly informed us 
that the absence or presence of nitric and nitrous acid in his analyses of rain- 
water is to 1)6 explained by the method adopted for their determination. The 
amount of these acids in rain-water proved too small to be determined by the 
method he has usually employed in water-analysis, at least with the limited 
bulk of rain-water placed at his disposal ; he therefore in the later analyses 
converted tlie nitric and nitrous acids into ammonia, by the well-known aluminium 
method, and found tliat in this form the minute quantity of nitric and nitrous 
acid could be readily determined. When employing the latter method a small 
quantity of nitric or nitrous acid was always found in rain-water. 
In consequence of this communication from Dr. Frankland, we have disregarded 
the recorded absence of nitric and nitrous acid in certain of his analyses of rain 
and dew, and in calculating the mean quantity of nitrogen existing as nitric and 
nitrous acid have taken those analysea only in which nitric and nitrous acid were 
found. 
