THE SOURCES OF THE NITROGEN OF VEGETATION, ETC. 547 
Accuracy of the method for the determination of nitrogen by combustion 
with soda-lime, &e. 
In order to ascertain the accuracy of the method before relying upon it for the pur- 
poses of the investigation, a few preliminary experiments were made upon the determi- 
nation of small and known quantities of nitrogen, mixed with large quantities of soil, 
which had been previously freed from combined nitrogen as in the preparation of the 
soils for the plant-experiments. The nitrogenous substance taken for the purpose was 
the powdered crystals of purified quadroxalate of ammonia, te ‘ho, (C, 0;),-+-7 HO. 
The results were as follow— 
Experiment 1.—50 grammes of the prepared soil were mixed with quadroxalate con- 
taining by calculation 0:0024 gramme nitrogen; and on burning with soda-lime, and 
determining as above described, 00027 gramme nitrogen was found. 
Experiment 2.—100 grammes of the soil mixed with quadroxalate equal, by calcula- 
tion, to 00035 gramme nitrogen, gave on combustion 0:0037 gramme nitrogen. 
The error of analysis was, therefore, three-tenths of a milligramme of nitrogen with the 
50 grammes, and two-tenths with the 100 grammes of soil. These results were obtained 
at the commencement of the inquiry, with comparatively large quantities of titrated acid, 
and therefore before experience had suggested the precautions to be adopted to reduce 
the errors of determination to the minimum. They may hence be taken as examples of 
the maximum errors of analysis, but they are less than would affect the bearing of the 
results in the investigation on the question of assimilation. | 
Testing for Nitric acid. 
The indigo test, as recently refined by BoussincauLt*, and the protosulphate-of-iron 
test, were both employed. When nitric acid was sought for and not found, if practicable 
the negative result was always confirmed by the addition to some of the substance under 
examination of a quantity of nitric acid (in the form of nitrate) less than could affect 
any conclusions to be drawn from the fact of its presence or absence in the substance in 
question. In all the cases of such addition the re-examination showed the presence of 
nitric acid. 
The method of BovussineavLt was much more delicate than the protosulphate-of-iron 
test; but, on the other hand, the latter was much less liable to give deceptive indica- 
tions, dependent on other circumstances than the presence of nitric acid. In using the 
protosulphate test, the aqueous extract of the substance under examination was evapo- 
rated to a small volume with excess of fixed alkali, then transferred to a test-tube, and 
further evaporated till only a few drops remained. A considerable excess of concentrated 
sulphuric acid was then added, and on the surface of the liquid a concentrated solution 
of protosulphate of iron was carefully poured without agitation, by means of a small 
pipette with a mouth of almost capillary fineness. The characteristic brown tinge 
indicated the presence of nitric acid. 
* Ann. de Chim. et de Phys., vol. xviii. (1856) p. 158 et seg. 
