that derived by plotting the 5th series falls considerably below those obtained by plotting the 6th or 7th. A mere inspection of the table exhibits the same in particular cases. For example, a difference of temperature of 268° 64 between iron and platinum, corresponds in the third series to a difference of 13° 71 between bismuth and antimony; whereas in the 6th series, a difference of 268° 66 between the former corresponds to a difference of 17.77 between the latter; and in the 7th series, a difference of 268-56 is equivalent to one of 18° 60. It hence appears that the thermo-electric force of iron and platinum is relatively greater in the 6th and 7th series than in the 5th. We shall now endeavour to account for this hitherto inexplicable result. Turning to the table at page 85 of this Report, we observe that the current formed at the junction of hard and soft in an iron wire passes from hard to soft, which proves that the iron is rendered more negative when it is softened by heat. Let us now devote a moment's attention to the result with platinum wire at page 87. In the case of two homogeneous wires, the current passes from warm to cold, causing a deflection of 24° when both wires are hard. When a hard and soft wire are taken, and the former is heated, the current passes as before from warm to cold, causing, however, a deflection of only 13°. It thus appears that the soft wire is less negative, or what is the same, more positive than the hard wire. Consistently with this, if the heated wire be the soft one, the fact of its being hot and soft at the same time ought to make the current developed a maximum-this is the case. The deflection observed under these circumstances is 36°.

The general facts being thus established, that iron, when softened by heat, becomes more negative, and that platinum, when softened by heat, becomes more positive, let us apply them to the case before us. M. Regnault commenced his 5th series with a fresh couple of iron and platinum, increasing the difference of temperatures between the hot and cold junctions gradually until it reached 273°-46. The absolute temperature of the hot junction at this point was in all probability 300°. After the couple had been thus heated, it was allowed to cool, and the 6th series was commenced: here the anomaly before alluded to at once presented itself; a certain difference of temperature produced a stronger current than in the 5th series, a result which might be inferred à priori from the foregoing considerations. For the iron by being once heated to 300° has become more negative, as before proved, while the platinum has become more positive; the thermo-electric force of the couple has, in short, been increased, and a more powerful current is the necessary consequence. This is still more strikingly exhibited in the 7th series, where M. Regnault commences with a difference of 103°.80, and goes on in

creasing to 282° 18; then, without interrupting the series, allows the difference to sink again to 148°.97. The bismuth and antimony equivalent for this is 12°-30; whereas for a difference of 15229 between the iron and platinum, before the difference of temperature had reached the above amount (282°·18), the antimony and bismuth equivalent is only 11°-69. This fluctuation in the 7th series causes the curve derived from plotting to present somewhat of the appearance of a railway section over undulating ground, whereas in all the other cases it presents a gradual and almost uniform ascent. The sudden leaps' noticed by M. Regnault, whose cause he considered it impossible to ascertain, appear to be thus capable of satisfactory explanation.

XIV. Observations in the Alps on the Optical Phænomena of the Atmosphere. By Dr. HERMANN SCHLAGINTWEIT.

[Concluded from p. 16.]

COLOUR OF THE ATMOSPHERE. Different kinds of Cyanometers. Alteration of the intensity of the blue with the height. Determinations with the tricoloured Cyanometer. Cloud colours.

COLOUR OF THE ATMOSPHERE.

THE blue colour of the sky, as well as the transparency of the atmosphere, deepens as we ascend. De Luc* has already noticed this. Saussure and Humboldt have published a long series of experiments on the subject. The instrument used by both was the cyanometer of Saussure. It consists of a number of strips of paper, washed over with different shades of prussian blue. The differences of shade are so regulated, that two strips, which at a certain distance could not be distinguished from each other, constituted divisions upon the scale. As normal distance, Saussure assumed that at which the black circle of a diaphanometer 13" in diameter disappeared. Black was added by little and little until perfect black was obtained. At zero the scale was perfectly white, at the extreme end perfectly black. Within these two limits the scale was divided into 53 degrees. With this the colour of the sky was compared, and the nearest degree was set down as the expression thereoft.

* Modifications de l'Atmosphère, vol. iv. § 117, p. 930.

† As an example of Saussure's degrees, we may mention that the mean position of his cyanometer amounts—

For Germany, to

For the torrid zone

On Mont Blanc

15-17 20-24

Humboldt found (Tableau Physique, p. 103)–

In the tropics

On the peak of Teneriffe

On the Andes at 3000 toises.

which has also been observed by Gay-Lussac.

39

23

41

43

To attain a more varied change of tint, Parrot* made use of a rotating disc on which were laid sectors of prussian blue; he thus obtained a mixed colour capable of far greater modification. His instrument was also divided into degrees. It seems, however, very difficult to obtain instruments of both descriptions which are quite capable of being compared with each other. After some experiments, we found it advantageous to apply the colours in a different manner; and instead of expressing the tint in degrees, to express it according to the proportions of the mixture. We constructed two cyanometers, the first was of the same form as that used by Parrot. A disc 20 centimetres in diameter was covered with a layer of white lead, a substance which, when properly manufactured, possesses everywhere the same degree of whiteness, whereas different descriptions of white bleached paper vary greatly from each other in this respect. The rim was divided into 100 degrees (1° being =3.6 of the usual divisions), and by means of these the whole surface was divided into distinct sectors. This disc was fixed upon another of pasteboard by means of little supports, which sustained the centre and the rim merely. The rest of the space between both discs was hollow. From three points situated 33-3 of the rim divisions apart (120° in the common sense), a knife was drawn along the corresponding radii. Through the slits thus formed, blue and black segments could be pushed in until the required portion of them was visible upon the surface; the remaining portion slid into the hollow space between the discs. The blue segments were coloured by a layer of cobalt (oil colour) carefully laid on; on the others was placed a layer of raven-black (oil colour). These colours can be found everywhere, and exhibit such slight deviations of shade that they may be regarded as constant. Pl. I. fig. 3 exhibits the mechanical arrangement of this apparatus; the section of it is given at B. a is the plate of paper on which the layer of white lead is laid; b is a disc of pasteboard parallel to the latter; at c are the sections of the supports which connect both discs at the centre and rim; d is the projecting periphery which carries the graduation; e is a small cylinder of wood, 2 centimetres long, which is fast glued behind. Around this passes a strap, which being pulled downwards, imparts a rotary motion to the disc sufficiently quick, and of sufficiently long continuance, to permit of comparing the disc with the portion of the firmament to be investigated. The screw f holds the instrument fast to the upright which supports it during the rotation; at g are plates used to strengthen the apparatus.

In fig. A the surface of the cyanometer, as fitted for experiment, is represented. The blue sectors partially cover the white * Physik der Erde, § 278, p. 102.

surface.

As the radii of the sectors are the same as those of the disc, the exposed surfaces of both are proportional to the number of degrees embraced by the circular contours. We have in the present case

If the disc be now set in rotation, we shall obtain a mixed colour the same as if we had blended

33 per cent. of blue,

and 67 per cent. of white

most intimately together. It will be afterwards seen, that in this way a colour may be obtained, which, although it approaches very near to that of the portion of the firmament under examination, still does not necessarily possess that tone which we denominate the colour of the air. It would thus be possible to attain the brightness corresponding to the position of the instrument shown in the figure in another manner, that is, by omitting blue, and setting in its place a sector of black (of course much lighter). For the simplicity of the process and the comparability of the results, we have found it more advantageous never to use black as long as pure cobalt, which itself is a very dark colour, was not lighter than the firmament*.

As the setting up of the apparatus and the rotation of the disc demanded considerable time, we found it convenient to have an instrument similar to that of Saussure, that is to say, coloured strips of paper, with which, however, neither the prussian blue on the white paper, nor a division into degrees, was made use of, but which was so arranged that the per-centage of cobalt could be immediately ascertained. In the construction we proceeded in the following manner:

A uniform cylindrical glass syringe was divided into 300 equal volumes, and then filled alternately with Kremser white and carefully prepared pure cobalt (both oil colours and of the same consistency); a series of equal volumes of white and cobalt were now placed beside each other on a palette. We had thus constant colours, capable of being casily imitated by subsequent experimenters. Oil colours, further, permit of being very inti

*Compare Arago's ingenious cyanometer, in which a plate of quartz, cut perpendicular to the axis of the crystal, is used for the production of the blue with which the colour of the sky is to be compared.-Annales de Chimie, vol. iv. p. 98.

mately mixed, and of being uniformly laid on the surface, which in our case was that of weakly-sized Bristol-board. In this way we obtained fifteen divisions of a scale; the first of which was white, and the last pure cobalt. The difference from one division to the next was a matter of indifference in the application of the instrument, as the per-centage content of cobalt and not the number on the scale was noted. The increase of cobalt from one division to that next to it was not uniform. We endeavoured to have the differences of shade from leaf to leaf tolerably alike; and here we remarked, that a uniform addition of cobalt becomes less appreciable when a considerable quantity of the colour is already present. In the last leaves, therefore, we used a greater proportion of cobalt than in the first; the immediate object of the latter was to render the instrument more uniform.

Cyanometrical experiments are, in general, determinations of the brightness rather than of the colour; it is, however, of some interest to investigate the shades of the latter a little more closely. A mixture of white and cobalt cannot fully accomplish this. The most direct way of proving this, is by looking at a landscape painted in oil, where only white and blue are used in the treatment of the sky. An addition of red or yellow is always necessary. As the shades of colour exhibit considerable changes, it seemed to us not unimportant to determine their relations, at least approximately, for different elevations. The colour which is generally added to complete the sky tone is light ochre (hydrate of iron); this unfortunately is a colour which, strictly speaking, cannot be regarded as constant in all manufactories. But the smallness of the quantity used, which never exceeded 11 per cent., served to render the disturbance arising from this less appreciable.

In the construction of this second scale, and of a third for the colour of the clouds, we have been assisted by the advice of that distinguished landscape-painter, M. A. Zwengauer of München, to whose kind and friendly support we take this opportunity of expressing our deep obligation.

The basis of the tricoloured cyanometer, consisting of a union of cobalt, white and ochre, was formed by three different mixtures of the last two colours. The first consisted of 20 parts of white and 1 part of ochre; the second of 20 parts of white and 2 of ochre; and the third of 20 parts of white and 3 of ochre. To each of these separately were added 4, 8, 12, 20, and 50 parts of cobalt, so that for every tone we had five divisions of the scale; we had, therefore, fifteen divisions in all. In the formation of a scale for judging of the colours of the clouds, such a simple process could not be followed. The most suitable procedure appeared to us to be that of imitating the most marked