Dec. 28, 1882] 
AMERICAN RESEARCHES ON 
WATER-ANALYSIS 
JE the instance of the U.S. National Board of Health, a 
comprehensive investigation relating to analysis of water 
has been recently conducted by Prof. J. W. Mallet, F.R.S., of 
Virginia, assisted by several chemists and others, It was pro- 
posed to examine carefully the chief processes in use for chemi- 
cally determining the organic matter or its constituents, in drink- 
ing water, to test the absolute and relative accuracy of the 
results these processes are capable of yielding, and, as far as 
possible, to ascertain the nature and scope of the practical con- 
clusions available for sanitary purposes, which might thence be 
secured. 
A preliminary report of this inquiry has appeared in a recent 
supplement (No. 19) of the National Board of Health Bulletin, 
put we propose here to give our readers some idea of the nature 
of it. 
Three chemists took charge severally of the so-called ‘‘ com- 
bustion process” of Frankland and Armstrong, the ‘‘ albuminoid 
ammonia process” of Wanklyn, Chapman and Smith, and the 
permanganate process as advocated by Tidy; they were, Mr. W. 
A. Noyes, who worked at Baltimore, Dr. Ch. Smart, U.S.A., 
at Washington, and Dr. J. A. Tanner, U.S.N., at Virginia. 
The water-samples were collected and distributed to these three 
places by Prof. Mallet, and all three chemists were required to 
examine their several samples on the same day. Prof. H. Newell 
Martin undertook a simultaneous microscopic examination, and 
pathological observations on the effect of injecting the waters, 
concentrated by evaporation at a very low temperature, under 
the skin of rabbits. 
After a large amount of preliminary and special work, nine 
classes of waters were obtained or prepared, for the main series 
of test analyses, as follows :— 
Class I.—Natural waters, believed to he wholesome (including 
the water-sup)»ly of some of the principal cities). 
Class IJ.—Natural waters which were believed to have actually 
caused disease in those who drank them. 
Class III.—Natural waters of doubtful, but more or less 
suspected character. 
Class 1V.—Artificially prepared waters, made by adding to 
wholesome water, certain amounts of various infusions of 
vegetable organic matter, such as drinking water is lable to be 
contaminated with. 
Class V.—Waters prepared with various forms of vegetable 
refuse, from manufacturing or industrial operations. 
Class VI.—Waters prepared with azima/ (or partly animal) 
organic matter of natural origin. 
Class VII.—Water prepared with azimal refuse from manu- 
facturing or industrial operations. 
Class VIII.—Water prepared by adding morbid products of 
human disease. 
Class I1X.—Solutions, in distilled water, of carefully deter- 
mined amounts of pure substances of definite chemical com- 
positions, 
The report first deals with the degree of accuracy of the three 
processes examined, a matter which may be looked at in two 
ways : first, as to the concordance of the results of each process 
in duplicate or triplicate experiments on the same water; and 
secondly, as to the agreement of the results with the actual, 
quantitatively known, contents of a particular water. 
In the case of multiplied experiments on the same water, 
then, the author first shows in three successive tables (for the 
three processes), the divergence of individual results from the 
mean, as fercentaze on the mean. It appears that, on the whole, 
the most closely concordant results were furnished by the per- 
manganate process, and the least so by the combustion pro- 
cess, the albuminoid-ammonia process holding the intermediate 
position, 
Next, a table is given showing the extent of agreement of 
results obtained by the different processes with quantities of 
organic constituents known to be actually present. ‘The figures 
strikingly indicate certain important defects of the several pro- 
cesses, though (it is pointed out) they must be looked at in a 
broad, general way, remembering the small number of organic 
substances treated and their special characters. 
The Report proceeds to deal with the effects on the results by 
the different processes of varying the extent of dilution of the 
same organic substances in water. Here, as regards the com- 
bustion process, distinct confirmation is had of the existence of | 
NATURE 
On 
two forms of constant error affecting evaporation. The weaker 
the solution the less is the amount of organic carbon obtained, 
and the larger the figure for organic nitrogen. If the usual 
interpretation of Frankland’s C : N ratio be applied, the curious 
and important result of these sources of error follows, that the 
more dilute an organically-polluted water is, the more animal- 
like in origin will its polluting material seem to be ; while the 
stronger it is, the greater will be the tendency to refer the con- 
tamination to a vegetable source—a distortion of conclusions 
manifestly in the opposite direction to that commonly assumed 
to be safe. 
As to the albuminoid ammonia process, the weaker the solu- 
tions, the higher are the results obtained, both for free and 
albuminoid ammonia, the influence common to both being 
probably that of imperfect condensation in the distillation 
having a less effect as the quantity of ammonia is stronger. The 
lower results from the stronger solutions for albuminoid am- 
monia, may be partly due to the fixed charge of alkaline per- 
manganate to the quantity of organic matter to be acted upon. 
‘The results of the permanganate process are shown to be 
much less influenced by varying dilution within the limits of 
those experiments than those of the other proces-es. 
The following, briefly stated, are the author’s 
Special Conclusions as to the Combustion Process 
1. The combustion itself, carried out according to Frank- 
land’s directions, is a process of great delicacy, and quite satis- 
factory in its details, with proper precautions, and in trained 
hands, Gaseous volumetric analysis with the aid of the 
Sprengel vacuum, is sufficient. 
2. The Frankland process is quite within reach of the 
manipulative skill of any fairly-trained chemist, but it requires 
practice and probably pretty constant practice. It is better 
adapted for large public laboratories, where many samples of 
water are examined, than for occasional use by a private 
individual, 
3 The defective point is the failure of the evaporation to 
leave a residue representing the original organic matter ; this 
(suspected hitherto) is regarded as now established. There are 
two constantly present errors (greatest with little organic matter), 
viz., loss of carbon, and gain of nitrogen from the atmosphere, 
the latter probably partly balanced by loss of that originally 
pre-ent in the water. 
4. These errors, affecting unequally the carbon and nitrogen, 
are liable to alter the statement of the C : N ratio, and so distort 
sanitary conclusions. 
5. The result for organic nitrogen, by the combustion process, 
is also affected indirectly by the errors connected with determina- 
tion of the ammonia (‘free ammonia”’), as the nitrogen be- 
longing to this has to be subtracted from the gross result. 
6. A further, indirectly operative, cause of error as to the 
nitrogen arises from tbe varying loss of ammonia by dissociation 
of its salts during evaporation, The time occupied in evapora- 
tion, as well as the amount of ammoniacal salts present, will 
influence the amount of loss. Such error will mostly be very 
small, but may affect considerably the result for organic 
nitrogen. 
7. The presence of nitrates presents a special difficulty in the 
combustion method. Mr. W. Williams’s application of the 
copper-zine couple deserves more precise quantitative examina- 
tion than it has yet had. The simultaneous presence of urea 
with nitrites and nitrates, presents a case of pecul:ar difficulty, 
with several sources of error. 
8. The formation of sulphuric acid during evaporation with 
sulphurous acid seems te occur oftener than has been recognised. 
It is much to be deprecated, though its effect, at any rate on the 
carbon determinations, has not seemed so great as might have 
been expected. 
9. The combustion process, in its present form, cannot be 
considered as determining the carbon and nitrogen in a sense to 
justify the claim of ‘‘absolute” value for its results, which has 
been denied to those of other methods. It is but a method of 
approximation, involving sundry errors, and in part, a balance 
of errors. 
10. There is, however, good ground for believing, that in 
many, perhaps most cases, its results for organic carbon may, 
with proper precautions, be made more valuable than the indica- 
tions of the permanganate process, and its results for organic 
nitrogen more valuable than the indications of the albuminoid 
ammonia process. 
