of the ocean, and likewise to investigate the decrease of heat in the higher regions of the atmosphere when applied to balloons.†

A register thermometer by Dr. John Rutherford, is described in the Edinburgh Transactions, which, at least for the minimum, is decidedly superior to any other. That for showing the lowest temperature is shown at Plate DXXV. Fig. 1. It is a spirit thermometer, in the fluid of which is placed a little thread of enamel enlarged to a globule at each end seen at a, and this constitutes the whole instrument. From the cohesive power of the spirit confined within the tube, it does not suffer this little index to pass it without a very considerable shock; it therefore carries it back to the lowest point of contraction and there leaves it, marking the greatest cold, and by expansion the index is not carried forward. It is adjusted merely by inverting the instrument till the index runs down to the end of the column of spirit, where it stops. It is then laid in a horizontal position. The maximum thermometer, Fig. 2. is exactly the reverse of this, the index A being pushed forward by a column of mercury, and left at the point of greatest expansion. The index was originally of ivory, but from the friction with the interior of the empty tube, it was found advisable to make it of steel and adjust it with a magnet like Six's. Even this, however, does not prevent the index from getting entangled with the mercury, and Mr. Adie of Edinburgh has more recently adopted the plan of placing naphtha above the mercury, in which the index floats. This has somewhat altered the character of the instrument for simplicity, though we suspect it to be almost a necessary evil. The minimum thermometer, as we have already said, is almost unexceptionable; we have had long experience of it, and find it extremely little liable to go out of order. It is not yet, however, constructed with all the delicacy of which it is undoubtedly susceptible.

A contrivance, intermediate between Six's and Rutherford's has been more lately proposed by Dr. Traill. It is the minimum thermometer of Rutherford, with a short piece of mercury A (Fig. 3.) introduced above the index a, and the spirit again covering the mercury contains another index b. It is manifest that the mercury, by pushing back and forward these indices, marks the two extremes; being of steel, gilded by means of galvanism, they are brought back by the magnet. To this elegant instrument, we fear from the trials we have made, that there are likely to be strong objections; we have not merely found the mercury extremely liable to separate, but there is reason to fear, from the certainty that a film of spirit lines the whole interior of the tube, that by the efforts of its corpuscular attraction, and the difficulty of moving a

sluggish column of mercury, however short, the position of the mercury, and consequently the true values of the degrees, will be changed.

Mr. Keith of Ravelstones has described a thermometer which, by a mechanical contrivance, registers for any given time, in a continuous form, the flux of atmospheric temperature. His instrument has the form represented in Fig. 4. the bulb A down to the point is filled with spirit, where mercury commences, which is continued up to the point h in the opposite leg. It bears a float, which by means of a marker b, indicates the temperature upon a scale affixed, and by a mechanical contrivance, leaves indices at the extreme points. By the adaptation of a pencil to the marker, and of a rotatory cylinder moved by clock work, in place of the scale, it is obvious that the atmospheric changes. will be transferred to it for every instant of time. This mechanical invention is very elegant, but has never met with, and is not calculated for, general adoption. Such mechanically drawn charts might please the eye or the fancy of the general observer, but we think we may lay it down as a maxim in meteorological science, that little is to be done in the present state of its advancement, by any mere mechanical contrivances not of the simplest possible nature, or which extend beyond the limits of a glass tube hermetically sealed.

Le Chevalier has proposed lately a plan of registration sufficiently simple in its idea, and which had long since occurred to us. It is the adaptation of an index to the metallic thermometers which have been described in another part of this article. The difficulty, however, is to find a method sufficiently delicate not to impede the motion of these most susceptible instruments. No plan has been given for this.

**

We have already mentioned that, besides the registration of the greatest heat and cold, thermometers have been proposed to record the temperature at any given epoch. Mr. Keith's, as we have seen, answers this purpose, but one more specifically for this end has been proposed by Mr. H. H. Blackadder. He has a thermometer A. Fig. 5. constructed exactly similarly to Cavendish's maximum, only that it contains mercury alone, and breaking the termination sharply over, fixes it into the case b without bringing it to a capillary termination, which we think a manifest defect. When held with the bulb A lowest, the column of mercury may obviously be made continuous; but if by a piece of clock-work the position be made horizontal, as shown in the figure, the column will be interrupted, and form a mark of the height of the mercury at that moment, when compared with the attached common thermometer B. But if the temperature were to rise after the given moment, the

Dr. Brewster's Journal of Science, vol. iii. Bulletin des Sciences, Mars. 1829.

† Dr. Brewster's Journal of Science, No. xiv. p. 249. § Edinburgh Transactions, vol. iv. Edinburgh Transactions, x. 440.

mercury would obviously be expelled from the end of the tube, to avoid which two hair pencils are contrived, at the moment of immersion, to drop an evaporable fluid on the bulbs, until they are observed, when by adding the degrees vacant in tube Ab, to the height of the thermometer B, we have the temperature required for the given epoch. By terminating the tube of the thermometer A, in the manner shown in Fig. 6, it is obvious that the same effects will be obtained by a mere rotatory motion of the instrument, coinciding with the axis of the tube. We have not room to state our various objections to the instrument, but we conceive that the complexity of its construction, combining a double thermometer, a time-piece, and an evaporating apparatus, is such as to prevent its adoption beyond a very confined sphere.

So much for self-registering thermometers; we must now say a word or two on a modification of the common thermometer, in its most original form, which has been termed the differential thermometer. It is far from our intention to occupy our columns with any dispute regarding the true inventor of the instrument, which is exceedingly elementary in its principles, and is chiefly valuable from its applications. We will, however, say that we have examined the rare work of Professor Sturmius of Altdorff, entitled "Collegium Experimentale Curiosum," Nuremberg 1676, where the details and figures are so copious as to leave no doubt on the mind of the candid inquirer that Sturmius was the inventor of this modification of the thermometer. Professor Leslie and Count Rumford both revived it in 1804, under the titles of Differential Thermometer and Thermoscope, two forms of which are shown in Plate DXXV. Figs. 7 and 8. Its principle consists merely in showing the difference of temperature between two media, in which the balls A and B are placed, and these being filled with air, the excess of expansion of air in the one above the other is marked by the motion of the coloured fluid which fills the greater part of the bent stem, the height of which, h, is referred to a scale attached. As it is strictly a thermometer of differences, it refers merely to the relative elasticities of the air in the two balls, and being hermetically sealed, is free from the common defects of the air thermometer. The most important application of the differential thermometer to practical meteorology is the hygrometer invented by Mr. Leslie, on the Huttonian principle, of the difference between the temperature of a dry and moistened surface, a case exactly fitted for the interposition of the differential plan. We are happy to refer to a very elegant self-registering hygrometer on this principle, invented by the Reverend Mr. Gordon of Kinfauns, and described in the article METEOROLOGY, Vol. XIII. p. 167, in which the principle of Rutherford's minimum thermometer is elegantly applied to both the maximum and minimum indications of dryness. The differential thermometer has been made the

basis of several other instruments by Professor Leslie, adapted to meteorology, but which have not come into general use, and which it would be beyond the object of this article to detail.

A new application of the simple thermometer has recently been proposed by Baron Fourier, t which he names a Thermometer of Contact. It is shown in Fig. 9, and consists of a bulb and tube AB filled with mercury, as usual. By means of a case CDEF, it is surrounded with mercury retained by a leather bag, a b, and being placed with the bottom resting on any substance a, whose conducting power is to be examined, the time of the descent of the temperature from a certain point to which it has been artificially raised, to that of the substance under experiment, expresses, by an inverse ratio, its conducting power, which it is the object of the instrument to determine.

The thermometric properties of bodies have been applied to various mechanical contrivances of great ingenuity. One or two, which may rather be considered new forms of the instrument, than any thing else, deserve a place here. Dr. Cummings of Chester proposed a statical thermometer, which, by being constructed on a large scale, might open and shut windows of rooms or hothouses as the temperature became elevated or depressed above or below any standard temperature. A similar idea occurred, we believe, many years ago, to one of our most original philosophers, Sir James Hall, Bart. who, in the course of his long series of admirable experiments, which required a steady and intense temperature for a considerable period, contrived a valve for his furnaces which by its own pyrometric action, was opened or closed as the temperature increased or declined. The instrument of Dr. Cummings consisted of a large glass or iron matrass filled with air, but containing mercury in its stem, which was inverted into a cistern containing the same fluid. The matrass was supported from above by a cord over a pulley connected with the sash of a window, nicely balanced by it, so that it will easily be seen how, when the weight of the counterpoise was diminished by the expulsion of mercury from the stem as the excluded air increased in volume by temperature, the sash was let down from the top, and a communication with the external air effected; the reverse of course taking place as the temperature declined.

The balance thermometer of Kewley is represented at Fig. 10. It consists of an elongated bulb of alcohol A, which fluid, in the expansion of the tube at B, is succeeded by mercury continuing along the tube ff, and has free motion in a receptacle at c. The whole being suspended from a frame bc, by the point a, to which knife edges are adjusted, it is manifest that the expansion of the alcohol will, by expelling part of the mercury from the bulb B into C, destroy the equilibrium of the machine, and the quantity of the deviation may be measured by the divisions on the brass scale e,

moved by the pinion a, until the balance is restored. The machine, however, is most obviously applicable as a motive agent for mechanical purposes, such as the opening and shutting windows. Long before we were acquainted with Mr. Kewley's instrument, it occurred to us that such a contrivance would be applicable to the purposes of self-registration, but we believe it has never been put in practice.

Such we consider to be a pretty full detail of all contrivances of merit for the improvement and modification of the thermometer. We have stated the principal disadvantages of each, as they occurred, generally from our own experience. Register thermometers are the most imperfect, and open the widest field for ingenious adaptations, yet when we consider the very numerous experiments that have been made, and the few attempts at novelty, which of late years have been practically successful, we are tempted to consider much farther improvement in the theory as vain. There are, however, very great desiderata in the workmanship of these instruments, and much also to be looked for from observers, and especially that the instruments should be more widely diffused. They are generally made too large, or, if smaller, act imperfectly. In common day and night thermometers the objection of size is comparatively unimportant, since the maximum and minimum points of temperature are usually sensibly stationary, till almost any instrument has had time to acquire it; yet, even for such observations, much improvement is requisite, and how much more for those more refined inquiries which demand great sensibility in the indication of maxima and minima in short intervals; more especially in exploring the temperature of the atmosphere by balloons, or in hourly observations of the highest and lowest temperature. The clumsiness of most register thermometers is by no means an inherent defect, as many would have us believe; by proper care and skill in the manufacture, they may be made of great fineness, even where indices are required. On the Cavendish plan of capillary terminations with reservoirs, there is in some cases no limit to the degree of tenuity: we have very recently had a thermometer for maximum temperatures made with mercury alone, and having an attached mercurial thermometer precisely similar, to ascertain the values of the height of the mercury, made on the most delicate scale, of which the thermometer may, in some sense, be said to be capable, so that the bore containing mercury is little more than visible to the eye, the length of the degrees being considerable, and the whole calculated for the nicest experiments; it appears to work extremely well. For general adoption, however, we are here disposed to recommend Rutherford's thermometer, except that they are ill suited for carriage, a defect applying also to all thermometers containing mixed fluids.

It is very fortunate for the science of meteorology that the simple mercurial thermometer has become, under the present mode of manufacture, perhaps the most satisfactory, complete, and simple of any of our philosophical apparatus. We admit the carelessness with which it is often made, and VOL. XVIII. PART I.

the want of agreement of instruments of the same construction which sometimes occurs; but we assert that this want of agreement has been overstated, and that the very backward state of this, the most elementary thermometric branch of meteorology, and the slow and doubtful steps which it even yet makes, is owing but very trivially to this cause, and almost entirely rests with the observers themselves. Most persons who observe the thermometer with even more than ordinary attention, are quite ignorant of the real difficulties in the practical use of the instrument, and how their labours may contribute best to the advancement of the science they wish to promote. In the most meagre register, the choice of hours, the accuracy to which they are kept, the position of the instrument, are all subjects of considerable difficulty, and requiring perseverance and practical knowledge; but should inquiries with the instrument be pushed beyond so meagre a diary, the consideration of time, and place, and affecting circumstances, whether remote or direct, rise to paramount importance, and if not duly attended to, destroy every value of such observations. Under the manifold difficulties of the subject, and the want of extensive and systematic co-operation under which so valuable an instrument as the simple thermometer, and the light which it throws on the state of the atmosphere, have so long laboured, we are disposed to think the most important foundation-stone ever laid for the science of meteorology, to be the series of hourly observations of the thermometer made at Leith Fort, under the direction of the Royal Society of Edinburgh, and at the instance of their late secretary Dr. Brewster, which was continued with unremitting vigilance for four years, 1824-27, and the results of the two first years of which have been published by Dr. Brewster, in a most important paper in the Edinburgh Transactions, Vol. X. The light there thrown upon the daily thermometric curve, renders it, we think, very well fitted for a brief notice in this place, being, as we have no hesitation in saying, far the most scientific tribute which has been paid to the value of the instrument. The horary thermometric curve for 1824 and 1825, is represented in Plate DXXV, Fig. 11, and it will easily be seen how nearly it approaches the form of a parabola. That it really does so, is shown by the following table,

where the difference never exceeds a quarter of a degree of Fahrenheit. In this paper, too, a complete solution is given of one of the greatest problems in the application of the thermometer to meteorology, the proper hour of observation; the mean temperature occurred in 1824 at 9 h. 13 m. A. M. and 8 h. 26 m. P. M.; in 1825 at 9 h. 13 m. A. M. and 8 h. 28 m. P. M., a most extraordinary coincidence, which sets the value of these tables in a true light, demonstrates the real accuracy attainable in the subject, and proves the value of two years of consistent and systematic observation, above the undigested mass of heterogeneous materials collected by the labours of single observers in a century. Dr. Brewster has given tables for reducing the observations for any hour of the day, or pair of hours, to the true mean, for which we must refer the reader to the paper itself, glad, however, of having had an opportunity of introducing to his attention so important a subject.

We propose now to give some hints for the application of the thermometer to practical meteorology, a subject, as we have just observed, most palpably neglected, and the importance of which has been so overlooked, that, as far as we know, nothing like a systematic set of experiments have been published on the influence of disturbing causes in the determination of atmospheric temperatures, nor has any attempt been made to classify these sources of error. In some of our most practical works, all directions on the subject are concentrated in such meagre terms, as "to avoid all kinds of radiation;" but unless observers have attended to the nature of these, the causes by which they are produced, and the extent of the error to which they may give rise, such general cautions are wholly vain. We have but to look at the absurd statements given in the meteorological contributions of ephemeral journalists, of preternaturally high temperatures alleged to have occurred in remarkable seasons; even the temperature of 100° and upwards, have often been stated as occurring in this country in the shade. We may cite the very extraordinary summer of 1826, as a striking example of this, and we may refer to a paper, on that occasion, in the

Edinburgh Journal of Science for October of that year, as a specimen of the nice precautions requisite in ascertaining with accuracy such extraordinary temperatures. When the temperature of the air remains for any time above 80°, no attention can be too great to avoid errors of radiation and reflection. The great importance of the subject, the neglect it has received, and the absolute necessity of attending to it, to do the thermometer any justice as a meteorological instrument, induces us, before concluding this article, to give a bare outline of the subject, as one deserving of much attention in filling up.

The author of the present article has devoted some time and care to this inquiry, and particularly in the summer of 1828, made at his leisure some experiments connected with it, but without any immediate view of publication, and consequently without such systematic arrangements, and such a provision of apparatus as he would otherwise have adopted. During the present summer (1829) he intended to have continued these upon an enlarged and improved plan, but the extraordinary changeableness of the season, and an almost unremitting course of cloudy weather, deprived him of his best opportunities. He is therefore compelled to draw for illustration of his ideas upon his older sources of private information, of which, though upon an imperfect scale, he vouches for the accuracy, and they will aid him in pointing out at least what he conceives to be the track in which such an

inquiry should be pursued.

The deviations of the thermometric results from the real temperature, are either from internal or external causes. The former arise from the inaptness of any index of expansion we can use to assume the identical temperature of the air at any moment. Upon these we mean not at present to dwell; they have already been partly discussed in a former part of this article, when alluding to the variable capacities of bodies for heat. Alcohol thermometers we have shown to be more sluggish in their impressions than mercurial ones, and air is more rapidly affected than either. Hence the most essential error of the Sympiesometer, an instrument already alluded to: its whole accuracy depending on the perfect coincidence of the indications of the gaseous and mercurial bulbs, a certainty of which in many cases it is very difficult to attain.* In observations where the change of temperature is rapid or considerable, the sluggishness of alcohol thermometers may sometimes be sensible; but in most cases it is not so, as will be shown by the accompanying table, in which case, however, circumstances were rather favourable, the wind being in an advantageous quarter for promoting an equilibrium, the exposure of the window, close to the sill of which both instruments were suspended, being N.N.W.

The various sizes of the bulbs of thermometers is another great cause of inequality, but which is easily reducible to a known law; for since all the particles of the substance cæteris paribus require equal times to acquire a new temperature, the sensibility will be inversely as the masses; but as such a change depends upon an equable effect produced by the extent of surface exposed to the acting medium, it will be directly as the surfaces. But as in spherical bulbs the surfaces vary as the squares of the diameters, but the mass as the cubes, the sensibility will be as the power of the diameter, that is, inversely as the diameter simply.

2

3

But external causes, or those independent of the expanding substance itself are more complex and diversified. They may perhaps be classified in the following way.

I. Radiation, a solar, b Terrestrial.

II. Reflection of heat by buildings and other objects.

III. Absorption. a, By the wall or object to which the thermometer is attached. b, By the scale and mounting. c, By the colours of neighbouring objects.

IV. Humidity. a, At the time of dew falling. b, By the proximity of water.

I. Of the existence and great force of radiation, both of heat from the sun impressing the thermometer, which interrupts the free passage of its rays, and elevates the temperature above that of the atmosphere, and of heat from the earth in cold nights, which by transmitting rays to distant points of a clear sky, when the equilibrium is disturbed by the absence of the sun-it is not our present object to speak-but both are sources of error in observations conducted for other purposes than to measure them. The direct action of the sun's rays upon the naked bulb of the thermometer is much smaller than many persons imagine, for so vast a portion is reflected by the brilliant polish of the mercurial bulb, that it requires to be blackened to indicate

the true effect, by absorbing the luminous and calorific rays. We have often seen a thermometer very little influenced by the direct rays of the sun, if his altitude was not great, and there was a breeze of wind. From direct radiation it is unnecessary to warn the attention of the observer, it is by reflec tion and absorption that its influence is so difficult to be avoided. It is not merely in ascertaining great colds, as at night,* that the influence of terrestrial radiation is sensible. Even where more than half the sky is sheltered from the instrument, and in ordinary summer weather, its influence must be avoided. This will be illustrated by Table II. in which a large thermometer hung at a window, in the usual way, is compared with one having the scale detached and laid among cotton on the window sill exposed to the northern sky. The differences are very striking, and show how much errors are to be avoided in experiments, even at ordinary windows. We shall quote some other observations connected with terrestrial radiation, when we touch on the subject of Dew.

II. REFLECTION.-In making observations in towns this is one of the greatest difficulties to be struggled with, since opposite houses, and the of extraneous heat, even in the best exposures. It street below, never fail to reflect great quantities tion of a thermometer in country houses, all the heat is very frequently to be observed, that from the posiof the afternoon's sun is reflected from some projecting wall, yet the indications of the instrument are unsuspectingly observed to the last moment that the sun's rays do not touch the bulb. This is a fertile source of error to the amount of very many degrees, among ordinary observers-but in its more refined forms, reflection is an almost invincible obstacle, even with the greatest care-the sun's rays are reflected even from grass at great distances.† We have often seen thermometers hung at a north window, on the third floor from the ground, materially affected, in the middle of the day, by the sun's rays upon the grass, thirty feet below, and at a very considerable distance from the foot of the wall. The preventions of reflection are sufficiently simple in description, though very difficult to be