course relatively much longer than the increased time of combustion would imply, and a large loss by dissociation and incomplete ammonification is easy to understand.

When a channel was left in the tube, so that the contact between distillation products and the soda-lime was less intimate, the loss of nitrogen was still greater. When, with the channel, the combustion occupied only three quarters of an hour, this loss varied from 0.4 per cent. to 15-4 per cent. of the total nitrogen. But when the combustion, with the channel, was prolonged to two and a half hours, the loss reached in one case 27 per cent. of the total nitrogen. This increased loss would seem most probably due to both incomplete ammonification and dissociation. At any rate, the conditions provided for less perfect contact between soda-lime and substance; for reduction of the supply of moisture which would provide hydrogen for ammonification and would also tend to prevent dissociation; and for a very long exposure of ammonia to heat.

When, with channel and long heating as before, the temperature of combustion was raised as high as the tubes would endure, the loss was very much greater, and amounted to 68.5 per cent. of the total nitrogen. The most natural explanation of this increased loss is the increased dissociation of ammonia. When the length of the anterior layer of soda-lime was increased, and with it, consequently, the amount of open space inside the tube, the loss of nitrogen was greater still, namely, from 75 to 79 per cent. of the whole.

It would, of course, be foolish to attempt to say how much of the loss in these latter cases was due to dissociation and how much to incomplete ammonification. Such questions can be answered only by more detailed experiments, in which the gases produced should themselves be analysed. But it does seem reasonably safe to say that both dissociation and incomplete ammonification must account for a large part, if not all, of the errors here observed, and that there is great danger of loss in these ways unless due precautions are observed.

To conclude: As regards loss by dissociation and oxidation, the facts above cited appear to warrant the inferences that, in combustions conducted by the method here described, in which (1) the tubes were so closely packed with soda-lime as to leave a minimum of free space inside, (2) full opportunity was provided for contact between distillation products and soda-lime (water-vapour at high temperature) by the anterior layer of soda-lime 12 cm. long or thereabouts, which was well heated before applying the heat to the mixture of soda-lime and substance, and maintained at a medium heat during the whole operation, and (3) the operation was completed in not over three quarters of an hour:

1. Provided the tube and contents are allowed to cool slightly before aspirating with air, there need be no loss by oxidation.

2. At a temperature sufficient to heat the tube only to dull redness, there need be no considerable loss by dissociation.

3. Ammonia may be dissociated and nitrogen lost by either very high heat, or by conducting the operation so slowly as to leave the ammonia exposed for a long time to heat. Very likely the danger of long exposure is heightened by lack of moisture from the anterior layer of soda-lime after the latter has been heated for some time.

4. A vacant space in the tube (channel as ordinarily recommended) may involve serious loss. This loss is probably due to both incomplete ammonification of distillation products and to dissociation of ammonia.

The danger of loss of ammonia from rapid flow of the gases through the acid solution in the bulb-tube is sometimes assumed to be considerable. Musso, for instance, lays great stress upon uniform and slow evolution of gas, and prolongs the combustion for several hours, the main purpose seeming to be to insure complete absorption of the

ammonia.* The experiments cited by Mr. Haynes in a previous article imply that the real danger from this source is very small. All his attempts to push the combustion fast enough to force ammonia through the solution were ineffectual. For instance, in a combustion of sulphate of ammonia, completed in 12 minutes, the ammonia, which was sufficient to neuteralise half or a little over half of the acid, was entirely absorbed. The volume of the acid solution was 10 c.c.; it was contained partly in the bottom bulb and partly in the outer bulb of an ordinary Knop and Arendt bulb apparatus, in accordance with our usual practice.

Bearing of the Results upon the Interpretation of those obtained by other Experimenters. The Difficulties with the Soda-lime Method observed by Seegen and Nowack, Musso, and others.

Of the investigations which have been interpreted as casting doubt upon the reliability of the soda-lime method, those of Seegen and Nowack have, perhaps, been the most frequently quoted. These have carried great weight because the figures obtained were so wide of the truth despite the evident pains taken to secure accurate results, and because similarly bad results have been obtained by other chemists. The apparently unavoidable sources of serious error in the soda-lime method have been alleged as a ground for doubting the correctness of important series of researches in which the sodalime method has been used. The question of the validity of this doubt is a very important one.

In his account of his own experiments with flesh, Nowack states that care was taken to insure uniform development of gas, and the tubes were kept glowing hot for three hours ("die Röhre stets durch 3 Stunden glühen "). In the report of their joint experiments with animal and vegetable protein compounds, albumin, casein, syntonin, flesh, gluten, legumin, Seegen and Nowack state that they employed the same analytical methods and precautions (" die Analysen wurden in derselben Weise und midt denselben Cautelen ausgeführt, die . . . ausführlich erörtert sind ") as were employed by Nowack in his own investigations just referred to. The inference is that their combustions were likewise continued through a long time, if not at a high temperature. Musso, whose experiments were made with milk, whey, and cheese, says his combustions lasted hardly less than three hours, and in some cases they were prolonged for even six hours. the work of Nowack, of Nowack and Seegen, and of Musso, the figures obtained for nitrogen were very small. Taking those for nitrogen obtained by the absolute method as the standard, the loss of nitrogen in Nowack's experiments reached 20 per cent., and even more, of the whole nitrogen; in those of Seegen and Nowack the loss was from 1 to 20 per cent., while in those of Musso it was over 30 per cent. of the whole nitrogen. On the other hand, Gruber, who in reply to the criticism of Seegen and Nowack upon analyses by Voit and others in Munich, gives very satisfactory results obtained in Voit's laboratory with protein in flesh, never allowed more than half an hour for the combustions.

In

The experiments cited by Mr. Woods and myself in article No. IV. of this series seem to us to give satisfactory assurance of the reliability of soda-lime determinations of nitrogen in ordinary protein compounds when the combustions were made with proper precautions. Of the latter, the chief seem to us to be (1) sufficiently fine pulverisation and careful mixture of the substance with soda-lime; (2) close packing of the tube and sufficiently long anterior layer of soda-lime, so as to insure sufficient contact between soda-lime (heated water vapour) and distillation products, as well as to avoid too long sojourn of ammonia in the heated tube; and (3) maintaining the anterior layer at a moderate heat during the whole time of combustion, which should not be too protracted.

* Ztschr. anal. Chem. 16, 414.

Messrs. Seegen and Nowack do not state whether they left a channel in their tubes or not; but it is evident that the conditions of their work were such as to favour incomplete decomposition of the nitrogenous substance, imperfect ammonification of volatile decomposition products, and dissociation of ammonia. Prof. Seegen reports a nitrogenous residue in the tube after the combustions, and Dr. Nowack found the acid solution in the nitrogen bulbs so highly coloured as to interfere with titration with litmus solution. The combustion was prolonged for three hours, and during part of this time at a very high heat. The determinations of Muss were conducted very slowly, in some cases even more so than those of Seegen and Nowack, and the results obtained were likewise very low. Mr. Ball and I found that either leaving a channel in the tube, or high heating, or prolonging the combustion to two and a half hours, involved large loss of nitrogen, the loss with these conditions combined amounting, in some cases, to two thirds or three fourths of the whole nitrogen.

It would seem, therefore, that we have here cases of a kind not unfrequent in scientific research, in which the very effort to secure correct and reliable results involves grievous error.

I lay especial stress upon this matter because of the use that has been made of Seegen and Nowack's results to discredit the soda-lime method in general, and particularly to throw doubt upon the work of Voit and others in the Physiological Laboratory at Munich, in which this method has been employed in the study of the fundamental problems of nutrition. It was my fortune some time since, while working in that laboratory, to be permitted to observe very closely the ways in which determinations of nitrogen by soda-lime are there conducted. Although the effort to insure close packing of the tubes was perhaps less than in the experiments above detailed, and upon which the conclusions here given are based, yet on the whole the details of manipulation as I observed them were such as would, with the experience in this laboratory, lead one to expect correct results.

Although the results of experiments by Gruber and others in the Munich laboratory, and accumulated experience elsewhere, make any vindication of its work superfluous, yet I trust these statements, in connection with an explanation of the error into which Seegen and Nowack have unwittingly fallen, may, in the interests of science, not seem out of place.

The experience of other chemists, as well as that in this laboratory, of which part has been described in these pages, leaves me with the decided belief that the Scylla and Charybdis of the soda-lime determination are incomplete ammonification of nitrogenous distillation products and dissociation of ammonia. The other difficulties are, in general, easy to overcome. They may all be prevented, in ordinary animal and vegetable protein compounds, by the precautions above enumerated. But in alkaloids and allied compounds, and in some amines, and amido and azo-compounds, loss by incomplete ammonification seems very difficult to avoid.

The observations I have had occasion to make, as well as printed accounts of the methods followed in different laboratories, have given me the decided impression that the most common difficulties are loose packing of the combustion tube so as to leave a channel, insufficient anterior layer of soda-lime, and too high or too long-continued heating.

The soda-lime method is beset with dangers, and requires great care to avoid them. Indeed, after the experience in this laboratory, covering a period of several years and including some thousands of nitrogen determinations, if I were going to undertake again a series of analyses like those for which these studies were made, and the details of the work were to be placed in the hands of an analyst, however expert, who had not been through some such experience as that above described, I should feel under the

necessity of asking him to do a considerable amount of preliminary work, including comparison of results by the soda-lime with other methods, before I could feel sure of his results. A case in point may be worth mentioning.

After most of the above detailed experiments had been made and we had the method in such control that we were wont to say jocosely that from a given protein compound we could obtain the whole or any desired fraction of the nitrogen as ammonia, at will, a gentleman of no little experience, and who afterwards proved himself a skilful and efficient analyst, came to our laboratory as assistant. We placed a number of substances, mostly animal tissues, in his hands for nitrogen determinations by the soda-lime method, instructing him in the details of the process, and going through a number of determinations with him. He was very confident that he could secure perfectly reliable results and at the same time make the determinations with considerable rapidity. After a list of substances had been analysed, I looked over his figures, and was somewhat surprised at the variations in the duplicate determinations, which, in some cases, as I now recall, reached nearly 0.2 per cent. This led me to question their accuracy. Mr. Woods, by whom most of the nitrogen determinations of this investigation have been made, repeated the analyses and found in almost every case a considerably higher percentage of nitrogen, so that the whole work had to be repeated. The most plausible explanation of the errors seemed to be that the determinations in question were made rather hurriedly, and either because the nitrogenous material was not well enough mixed with the sodalime, or from too loose packing of the tube, some of the nitrogenous distillation products were incompletely ammonified, and thus escaped detection in the titration subsequent to combustion.

At the same time I must repeat what I have already said, that our experience leads me to place the greatest confidence in the soda-lime method for the determination of nitrogen in ordinary protein compounds, provided the work be conducted with the precautions here insisted upon.

The perfection to which Kjeldahl's method has lately been brought, and its accuracy, convenience, and inexpensiveness, have led to its use in this as in many other laboratories. Our experience leads us to decidedly prefer it to the soda-lime method, though we find it advantageous to use both, making one check the other. But the danger of incomplete ammonification of some classes of compounds, e.g., alkaloids, makes us feel it necessary to control both by the absolute method for all classes of substances, except those for which they have been most thoroughly tested.

SUMMARY.

The experimental and other considerations presented in this and the previous articles on the determination of nitrogen by soda-lime, may be conveniently summarised, after first recalling the probable reaction by which the nitrogen is changed to ammonia, and the principal sources of error in the operation.

1. It seems decidedly probable that the change of nitrogen to ammonia is effected by union, at high temperature, with water vapour yielded by the soda-lime (or slaked lime in case the latter is used). It is essential that the contact between nitrogenous distillation products and water vapour be sufficient, and not at too high or too low temperature to insure conversion of all the nitrogen to ammonia, and that the ammonia be not dissociated or oxidised. The main objects, then, are to secure complete ammonification and to avoid dissociation and oxidation.

2. The chief difficulty in the way of complete ammonification of protein compounds appears to be the formation of gases which do not readily yield their nitrogen to be united with te hydrogen of the water vapour. With certain other classes of nitrogenous compounds, as leucine and its congeners, alkaloids, amines, and amido and azo-compounds,

this difficulty is greater, and sometimes apparently insuperable. The tendency of protein compounds to be decomposed by heat and other agencies into leucine, amines, etc., appears to explain the difficulty frequently found in getting all their nitrogen into the form of ammonia by heating with soda-lime. The evident means to secure complete ammonification must be sufficient contact with soda-lime at proper temperature.

3. The danger of dissociation of ammonia evidently increases with increase of temperature and time of exposure, and is probably diminished by presence of water vapour and other diluting gases. If this be so, the danger will be avoided by measurably rapid combustion at not too high heat, and by keeping the ammonia in contact with sufficient moisture from the soda-lime until it leaves the heated tube.

4. Leaving out of account substances such as nitrates, nitro-compounds, etc., whose nitrogen is imperfectly converted into ammonia by soda-lime, even in the presence of organic matter, and assuming palpable errors of manipulation, to be avoided, such as (a) loose packing of asbestos plug, which would allow particles of soda-lime to be swept into the acid bulb; (b) heat at anterior end of the tube, so low as to permit ammonia to be retained with moisture about the cork, or so high as to char the cork and give rise to acid or alkaline distillation products; (c) use of soda-lime containing nitrates or nitrites, which may, according to circumstances, either furnish nitrogen to be transformed into ammonia, or oxygen to burn the ammonia formed from the nitrogen of the substance; (d) use of distilled water containing ammonia for rinsing the acid bulb; (e) imperfectly-cleaned or incorrectly-calibrated burettes; the principal sources of error above discussed involve loss of nitrogen, and may be recapitulated thus:

I. Loss from imperfect ammonification of the nitrogenous substance, due to:a. Incomplete decomposition of the substance, part of the nitrogen being, from coarseness of the particles of the substance, imperfect mixing with the soda-lime, insufficient heat, or other cause, left behind in the charred residue.

b. Change of nitrogen into compounds other than ammonia, either such as may remain in the tube, e.g., cyanogen; or volatile distillation products which escape ammonification and pass through the acid solution unabsorbed, or, if absorbed, are not accurately determined by the titration or other means used to find the amount of nitrogen in the solution.

c. Escape of nitrogen in the free state.

II. Loss of ammonia through :

a. Dissociation at high heat in the combustion tube.

b. Oxidation by air present in the tube before, or introduced in aspirating to wash out ammonia after, the combustion.

c. Neutralisation by acid products, e.g., of sugar, where the latter is used in the combustion.

d. Incomplete absorption by the acid solution.

5. Complete decomposition of the substance has, in our experience, been readily secured by pulverisation fine enough for it to pass through a of seive of 1 mm. aperture; thorough mixing with soda-lime; avoiding the shaking by which the particles gather at the top of the soda-lime; and heating to low redness.

6. With sufficient soda-lime, not too dry, we have found no reason to fear the formation of cyanides, nor have we been able to obtain any indication of the escape of free nitrogen when the operation is properly conducted, although it might, perhaps, occur by oxidation of ammonia if there were nitrites or nitrates present, or if aspiration with air were done while the tube and contents are hot. Turning off the flame before aspirating has, in our experience, sufficed to avoid oxidation by air. At least, if ammonia has been oxidised, the quantity has been too small to be detected.