numbers corresponding to equal intervals expressed in wave lengths, and is based upon Angström's measurements alone. For the sake of completeness, I give this table in full, together with the wave lengths for the same lines, as found by myself by Cauchy's method of interpolation, and the differences. For the sake of comparison I have changed the position of the decimal point in the wave lengths determined by Mr. Stoney, so as to correspond with my own results.

AM. JOUR. SCI.-SECOND SERIES, VOL. XLVII, No. 140.—MARCH, 1869.

Mr. Stoney does not state what method he employed in interpolating; the correspondence between his values and my own is in general very close, though I employed the data obtained by Ditscheiner, in place of those of Angström.*

Cambridge, Jan. 1st, 1869.

ART. XVI.-On the condition of our knowledge of the Processes in Luminous Flames; by EUGENE W. HILGARD.

FOR more than half a century after Sir Humphrey Davy's important investigations of the subject of flames, the experiments and explanations of that eminent philosopher have passed, unchallenged and almost unchanged, into text-books and lectures. As regards luminous flames especially, he established the necessity of the presence of a solid incandescent body to produce useful luminosity; and in reference to the flames of hydrocarbons in particular, he suggested that the liberation of carbon in them was owing to the combustion of the hydrogen of the compound in advance of its carbon, the latter being heated to incandescence by an oxhydrogen flame, as it were, and failing to be consumed until all the hydrogen was first oxydized.

It is remarkable that an explanation so directly at variance with the daily experience of blacksmiths, and with a lecture experiment performed even in the most elementary course of chemistry, could so long have passed current; for the decomposition of steam by ignited charcoal into hydrogen and carbonic oxyd gases is an old observation. Nay, most of us who have performed this experiment on the lecture table, have been in the habit of passing the gas through lime water or potash lye, to free it from carbonic acid before showing its properties. And yet we were taught, and have ourselves taught, that in the flame, hydrogen burnt first and carbon afterward, and any doubt on the subject was quieted by a reflection about the influence of "chemical mass on affinity.

When, in 1852, I was a student in the laboratory at Heidelberg, to which Bunsen had then but just been called, the latter suggested to me as an interesting subject of investigation, the composition of the gases in the various portions of the flame; stating at the same time that the prevalent opinion concerning

* It is proper to state here, that since this paper was read before the National Academy of Sciences, all the calculations have been revised and many corrections and alterations have been introduced. The original title was, "On the determination of wave lengths by the method of comparison," and a portion of the memoir has already appeared with this title. This Journal for May, 1868.

the respective affinities of hydrogen and carbon for oxygen were certainly erroneous-that, when mixtures of carbonic oxyd and hydrogen were exploded with oxygen in the eudiometer, the distribution of the latter took place according to some law dependent upon the relative proportions between the combustible gases, and also the amount of oxygen present, but discriminating widely in favor of carbon. The experiments upon which this opinion was based, are probably those given in his "Gasometric Methods" at the conclusion of the chapter on the combustion of gases.

In order to study the processes taking place in the flame, it became necessary to investigate first of all, the composition of the gases in the interior cone, from which the flame derives its substance. I constructed a lamp adapted to the introduction of a suction tube into the flame from below, and made a series of fifteen analyses of the gases so collected at various points of the interior cone of the flames of beef tallow, and of wax. The low temperature known to prevail in this portion of the flame rendered it least adapted to elucidating the mooted question of affinities; but being the generator from which the other portions of the flame are supplied, a knowledge of its component gases was indispensable. My analysis (published in Liebig's Annalen der Chemie und Pharmacie, vol. xcii, p. 129, 1854) showed, however, the existence even in the highest portion of the cone, of free hydrogen with a large excess of carbonic oxyd and carbonic acid; the amount of hydrogen varying but little from base to point, while the carbonic acid increased in about the same ratio as nitrogen, i. e., in proportion to the oxygen entering the flame. Bunsen as well as myself failed, however, to draw the legitimate conclusion from these facts, at the time; the more as, with the materials I used, it was impossible to follow the formation of water by progressive oxydation.

The latter difficulty was avoided by Landolt, who, two years later took up the same subject, the failure of my health having rendered doubtful the prospect of my ever being able to resume it so as to carry out the proposed investigation of the other parts of the flame.

Landolt used illuminating gas of known composition, and was therefore enabled to determine the deficient factor in my analysis, viz., water. So far as comparable his results in general confirmed mine. He felt compelled by the increase of free hydrogen in the higher parts of the flame examined by him, to assume the occurrence of a reaction between free carbon and (preformed) water; but he also failed to draw the inevitable conclusion as to what must happen in the luminous cone.

Pogg. Ann., vol. xcix, p. 389.

A later research of Lunge on the composition of the gas contained in the interior cone of the flame of a Bunsen's burner, must have led to the truth of the matter, by showing how little oxygen sufficed to render a flame non-luminous when previously mingled with the gas. I have not seen Lunge's memoir; but he likewise seems to have failed to draw the important conclusion of which his analysis must have contained the elements.

Next, in June, 1860, comes a memoir of Erdmann,† who in discussing the principles upon which his gas-tester (a modification of Bunsen's burner) is based, first distinctly enunciates that according to his experiments, the carbon in a flame is oxydized before the hydrogen, and that the separation of carbon upon which luminosity depends, is due to heat alone, as would be the case were the gas passed through a red hot tube.

Finally, eighteen months later, we have an elaborate and elegant research by Kersten, wherein he proves by eudiometric experiments that (at least within the limits of the proportions employed by him) whenever a hydrocarbon is exploded with oxygen insufficient to burn more than the carbon to carbonic oxyd no hydrogen at all is oxydized; but that as between carbonic oxyd and hydrogen, the formation of carbonic acid on one hand and of water on the other depend upon "chemical mass," as Bunsen had already shown.

This question has therefore been peremptorily settled by decisive experiments, as much as eight years since. Yet the latest editions of text-books published in this country, and even those which, like the excellent work of Messrs. Eliot and Storer, claim to represent the latest state of the science, still retain the old error regarding the succession of oxydation.

There is another point which, though I took special pains to demonstrate the facts fourteen years ago, is still incorrectly stated in almost all text-books as well as books of reference. I allude to the definition of the several essentially distinct parts of the flame. Three only are usually mentioned, viz., the inner cone, the luminous portion, and the outer, faintly luminous envelop or veil. Yet Berzelius already distinguished the very important fourth part, viz: the blue cup-shaped zone surrounding the base of the flame, which is as sharply defined from the inner cone, as from the outer veil with which it is usually confounded.

That this blue cup is identical with the blue cone of the blowpipe oxydation-flame, is stated by Plattner in the first edition of his work on the blowpipe. Strangely inconsistent with his * Annalen der Chemie und Pharmacie, vol. cxii, p. 205.

J. pr. Chem., June, 1860, p. 241.

J. pr. Chem., Dec. 1861, p. 290.

own definition, he nevertheless teaches that the blue cup is formed by the combustion chiefly of carbonic oxyd, produced by the "first and weakest effect of the heat on the fuel"—an assumption as little justified by experiments regarding such action, as was that of the combustion of hydrogen previous to carbon. It is palpably inadmissible in reference to the blue oxydation cone, which is of course identical with the flame of a Bunsen's gas burner-the supply of oxygen being, in both cases, sufficiently great, and so mingled with the entire gas as to suppress the separation of carbon.

The part performed by the blue cup, viz: that of a selfheating retort with walls impervious to oxygen, in which dry distillation is accomplished; its theoretical import, as the counterpart of the luminous portion, where the same gases are burnt with evolution of light, render the neglect with which it has been treated doubly surprising. That it is totally distinct from the outer veil is readily perceived when the eye is protected from being dazzled by means of a screen of the shape and size of the luminous hollow cone. The veil is then seen surrounding the blue cup as well as the higher portions of the flame, and is thus proved to be nothing more than a zone of glowing gas; which of course, however, cannot be strictly defined from the luminous envelop, the oxydation being a gradually progressive one, from the highly luminous central portion to that brownish, semi-transparent zone of transition, where the carbonic oxyd, burning simultaneously with hydrogen, fails to produce its characteristic blue tint because of the excessive temperature existing there. The same is the case when it is burnt by itself from a jet kept at a white heat.*

The inner cone, too, is still incorrectly defined as "the space containing the combustible vapors and gases generated from the wick." This would lead any one to suppose, that the external atmosphere had only the effect of burning off the outside of this gaseous mass, and some text-books have gone far in the graphic delineation of this process. My analyses first proved fourteen years since, that even in a tallow flame, characterized by an excess of fuel over the available oxygen, the percentage of nitrogen gas does not fall below 59 per cent in the lowest part of the flame, and increases to 76 per cent at the point of the inner cone, with from 10 to 15 per cent of carbonic acid. The products and educts of combustion therefore greatly exceed the combustible gases and essentially modify the processes thereafter occurring, for it is clear that in the luminous portion, the water and carbonic acid must in part at least again yield up their oxygen, before final combustion.

*See Gmelin's Handbook, art. Carbonic Oxide.