rid of and bubbles of nearly pure vapour escape. The vapour rises into the bell just as gas does into a jar of water at the pneumatic trough, and would remain but for the heated bottom of the flask near to the source of heat, which is hotter than any other part of the arrangement. This superior heat causes the vapour in the little bell constantly to expand and discharge itself; but it does not do so continuously, it is an oscillation between vapour escaping and vapour charged with water entering an oscillatory motion which, as I have said, is only possible at the bottom of the vessel (at least in this form of the experiment), where the heat is most intense. If the little bell be raised some way above the heated bottom it ceases to act, no vapour escapes; and hence I say that this experiment does not show that air has the exclusive function assigned to it. The heated bottom of the vessel is the chief source of the bubbles of steam. They may be seen flashing from this hot surface, rising upwards and rapidly vanishing if the liquid has not yet attained its boilingpoint, or bursting on the surface when that point has been attained. There are also certain specks in the bottom of glass flasks, retorts, tubes, &c. which act as powerful nuclei; they discharge streams of bubbles with unceasing vigour, which does not decline although the action be continued during many hours. At the same time I do not deny, never have denied, that the air in solution has a useful part to perform in lessening the forces of cohesion and adhesion in the liquid, and in facilitating the formation of bubbles of vapour. All I contend for is that air is not the only nucleus, but that there are other modes of separating gas or vapour from their solutions, as already pointed

out.

I am quite prepared to admit that, seeing how vast a number of important vital functions depend on the solubility of air in water, a portion of air should cling obstinately to water even when raised to high temperatures, or, by virtue of this property, even when expelled by boiling, air may be, as Mr. Grove suggested, continually reabsorbed at or near the boiling temperature; and I should prefer to adopt this view rather than the idea that a bubble of air one millim. in diameter is an efficient cause in liberating half a million of bubbles of steam, each five millims. in diameter.

I arranged M. Marco's experiment on a larger scale by employing a short tube nearly 2 inches in length and about inch in diameter. A few coils of thin binding wire were passed round it near the bottom and also near the mouth of the tube; each coil terminated in a loop; and to each loop was attached, also by a loop, a longer and stouter wire extending beyond the neck of the flask. In this way, without removing the tube from the

flask, it could be easily arranged with the mouth downwards or upwards. A globular flask containing above ten ounces of distilled water was employed. When the mouth of the small tube was downwards the phenomena were much the same as before, only the tube did not rest on the bottom unless held there. On removing the lamp, the tube rapidly filled and sank with its mouth upon the heated bottom, which expanding the enclosed vapour caused it to disgorge the water and so to rise again. It did this two or three times; it then settled and filled; but there remained a small bead of air about the size of the head of a small pin. On again applying the lamp, the water boiled readily; but 75 seconds elapsed before the tube became active; and it was not until bubbles of vapour passed up into it and displaced the water that the tube was in a condition to pour out its intermittent bubbles; and it seems to me of importance to insist on this point, namely that on this as on many other occasions the flask boiled, so to speak, from its own independent resources, long before the little bell or the small tube came into action at all; and when either of them did so it behaved simply as a nucleus, increasing the amount of vapour that was given off by the boiling liquid. The flask boiled first from the action of the heated bottom and the presence of minute specks, probably of carbon, on the surface of the bottom, the importance of which I insisted on in my first paper. The air in the bell or in the small tube was displaced by expansion; and if at length the whole of the air was not got rid of, the reason seems to be the impossibility of doing so by expansion, just as it is impossible to obtain a perfect vacuum in the receiver of an ordinary air-pump. The air in the receiver goes on expanding until an exceedingly thin medium is left; and the air in the tube in boiling water is pumped out by expansion, and the minute portion that at length remains shows, not that this speck is necessary to carry on the ebullition, but simply that it is impossible to exhaust the tube by expansion only. Hence I see no reason to qualify the statement made in my first paper, namely that I cannot help thinking that too much importance has been attached to this small residual speck of air in the phenomena of boiling liquids. If a speck of air be thus left in the tube, a minute portion also probably exists in other parts of the liquid, although I have failed to detect it. Moreover by repeatedly boiling the flask at intervals of five or ten minutes, and allowing the small tube to fill after each boiling, the water becomes so far purged of air that the speck in the tube disappears, or is so minute that I have been unable to pronounce as to its existence. Of course it may have entered into solution even during the observation.

In some of the low-boiling liquids the speck rapidly dimi

nishes and disappears. For example, in distilling wood-spirit (boiling-point 148° F.), a small tube, held by a thin wire passing through the cork, stood with its mouth near the bottom of the retort. After the fifth boiling, when the lamp had been removed, the tube poured out about 150 bubbles of vapour before it began to fill. It occupied three minutes in filling, and then a minute speck was visible by the aid of a lens ; but within another minute the liquid closed over this speck and it disappeared, the temperature of the liquid being as high as 128° F. M. Gernez would say that the speck had entered into solution, after which the boiling would become difficult. On the contrary, the boiling was just as easy as before.

M. Gernez is so impressed with the necessity for the intervention of air in the phenomena we have been considering, that he cannot do without it in accounting for the line of bubbles which is produced by the friction of a chemically clean solid against the inner side of a chemically clean vessel containing a supersaturated gaseous solution or a liquid at or near the boilingpoint. I have already endeavoured to give a simple explanation of these phenomena*. "The glass rod or the steel knittingneedle, on being pressed against the side of the glass, displaces a certain small quantity of the liquid, and on moving the solid, with friction, against the side, successive quantities of liquid are thus displaced. A certain time, however short, must elapse before the water can fairly close in upon the moving points of the line thus traced; but however quick the water may be in filling up the void, the gas is quicker, and hence a friction line becomes a line of bubbles."

M. Gernez says:-"As it results from these divers experiments that in each bubble of vapour disengaged there is always a small quantity of air, it is natural to suppose that the friction of solid bodies in the middle of a liquid should determine the separation of a small quantity of gas which was in a state of supersaturated solution and which serves as an atmosphere into which the liquid emits vapour."

Considering how minute a portion of air or gas liquids at or near the boiling-point, and especially after repeated boiling, can hold in solution-considering also that such liquids when removed from the source of heat, can be made to boil up again and even to boil over, by the mere friction of a hard solid, I cannot conceive that these multitudes of bubbles are called into existence by first liberating air; but I can conceive that the vapour, being already in solution in the hot liquid, is in a condition to rush into the vacuum formed for it by a solid moving against the side.

* Phil. Mag. for November 1874.

There are some other points arising out of the long memoir of M. Gernez, especially on the subject of superheating, that call for comment, but must be deferred to another occasion on account of the great length of this paper.

Highgate, N. May 10, 1875.

LII. Remarks on M. Goldstein's Observations on Spectra of Gases. By Professor WÜLLNER, of Aachen *.

N the Monatsbericht of the Königl. Akademie der Wissenschaften for August last are described by M. Goldstein observations on gas-spectra †, which are said to prove that my. explanation of the band- and line-spectra observed in one and the same gas (Pogg. Ann. vol. cxlvii. p. 321) is untenable. My explanation was to this effect-that from an illuminating gas a band-spectrum is always obtained when a thick layer of gas is made use of as the source of light, but the line-spectrum when only a few molecules of the gas are ignited. I deduce this from experiments which showed that in tubes filled with rarefied gases the line-spectrum always and only appeared when the inductioncurrent traversed the tube in the form of a proper spark, but the band-spectrum when the discharge was without a spark. In my experiments at that time the sparkless discharge was that which I afterwards (Pogg. Ann. Jubelband) named the luminous tuft it gives light which more or less fills the entire tube; the proper spark ignites a minimum quantity of molecules, only those which lie in the line of the spark.

The experiments of M. Goldstein which are said to contradict my explanation consist, so far as I understand his description, essentially in this :- In the circuit in which is the spectral tube he produces an interruption at one plaee, so that there the induction-current passes in sparks; sometimes he also inserts a Leyden jar; and then he observes the spectra of the tube. He expects under these circumstances line-spectra must appear in the tube, but finds, as a rule, in tubes filled with air, the bandspectrum, and only when the air presents greater resistance a mixed spectrum of bands and lines. No detailed description is given of his observations on hydrogen; he merely remarks (p. 602), "The observations on hydrogen correspond." He specifies, however, that, of two tubes simultaneously placed in the circuit, one containing air, and the other hydrogen, the former gives the band-spectrum, and the latter the line-spectrum. As the same

* Translated from the Monatsbericht der Königlich Preussischen Akademie der Wissenschaften zu Berlin, December 1874.

† See Phil. Mag. for May 1875, pp. 333-345.

rhythm of the discharge was present in every part of the circuit, M. Goldstein believes that when at one place there is a spark it also appears at all places in the circuit at which interruption takes place, that consequently, according to my explanation of spectra, the line-spectrum would everywhere show itself.

I see, on the contrary, in the experiments of M. Goldstein in general a corroboration of my view, which requires a bandspectrum whenever extensive masses of gas are rendered luminous. The error in M. Goldstein's assumptions is this :-When at one place in the circuit the discharge passes in a spark, he infers that it must also do so in all spectral tubes inserted, because the rhythm of the discharge is everywhere the same. This is not the case; the form in which the discharge takes place in spectral tubes depends on the pressure of the gas and on the dimensions of the tube.

That like rhythm does not establish like form, I have already shown in my treatise on the origin of spectra of different orders (Pogg. Ann. vol. cxlvii. p. 337), where, in one and the same hydrogen-tube, the spark extended only from the positive electrode about halfway down the tube, but below this it was dissolved. If the slit of the spectrometer was on a level with the spark, it gave the line-spectrum; if it was in front of that part of the tube in which the spark was lost, the band-spectrum

was seen.

M. Goldstein himself has also observed that in spaces containing air sufficiently rarefied, in spite of the insertion of a spark-distance, no spark comes, when he (in p. 603 of his communication) speaks of several centimetres thickness. He even alternated sparks with a discharge as rapid as the spark. That (as M. Goldstein truly remarks) such a discharge does not give a line-spectrum, is the best proof of the correctness of my interpretation of the phenomena; for in that case the whole of the light filling the tube shines, and not, as in proper sparks, merely a few molecules lying in the line of the spark.

Last year I made a great number of experiments on the passage of the induction-current through tubes filled with rarefied gases; and in those experiments, exactly as M. Goldstein did, I interposed spark-distances, and sometimes also Leyden jars. I have not been able to finish those experiments, because working at the new edition of my Experimentalphysik has claimed the whole of my attention. I have on this account communicated only a small portion of them, treating of the forms of the positive luminous tuft, in tubes filled with air, in its dependence on pressure and on the dimensions of the tube. Let me be permitted here to communicate a series of experiments from March 1873, carried out in conjunction with Dr. WinkelPhil. Mag. S. 4. Vol. 49. No. 327. June 1875. 2 I