ates highly trained in the subjects with which they have to deal. And, finally, it is insisted that the physics of the air offers many opportunities to the creative scholar, and every university must realize that its paramount duty is the fostering of research and the training of investigators, for in no other way can it meet the growing and compelling demands of a progressive civilization.

It must be admitted, however, that it is not now easy to give a connected course on atmospheric physics, for there is no suitable text and the isolated articles upon which such a course must needs be based are scattered through the journals from Dan to Beersheeba and buried under a babel of tongues. But this is only a difficulty, and not, in the face of imperative needs, an excuse. A far greater and very real difficulty has, it is true, confronted most of us, for, until the last decade, or less, several important lectures in such a course would of necessity have been restricted to the same brevity as characterizes Horrebow's famous chapter on snakes in his "Natural History of Iceland"-there aren't any. Some of these lectures are still unwrittentantalizing challenges to the skill of the experimentalist and acumen of the analystwhile others have been at least partially supplied, a few of which it will be interesting to review in what follows.

TEMPERATURE OF THE FREE AIR

Although efforts to determine the temperature of the free air by means of thermometers carried by kites were made as early as 1749, the experiments being conducted at Glasgow by Alexander Wilson and his pupil Thomas Melville; and although, beginning with Jeffries in his ascent from London in 1784, balloonists have often carried thermometers on their flights, it was only after the development of self-recording instruments and the sounding balloon-both at the very end of the last century-that the vertical distribution of temperature up to even 7 or 8 kilometers became at all accurately known. As is now known, and as shown in Fig. 1, the average

temperature decreases slowly with elevation near the surface, then more and more rapidly to a maximum at some such considerable altitude as 7 to 9 kilometers, where it roughly approaches the adiabatic rate for dry air of approximately 1° C. per 103 meters.

These are the observed facts; but here too, as in the investigations of other physical phenomena, a knowledge of what happens is only so much raw material out of which some one happily may fashion the finished productwhy it happens. In this case the why is found in the presence of water vapor in the air, its condensation and the latent heat thus rendered sensible. As air is carried to higher levels by vertical convection it progressively expands against the continuously decreasing pressure, and thereby does work at the expense of its own heat. During the dry stage of this convection, that is, until saturation is attained, the cooling is roughly at the rate of 1° C. per 103 meters increase of elevation. Immediately condensation begins, however, latent heat is set free and the rate of cooling with elevation correspondingly decreased. But as the amount of vapor condensed per degree drop in temperature decreases with the temperature, it follows that the latent heat set free and the corresponding check in the rate of cooling with elevation also decreases. Hence the continuous temperature-elevation coordinates of a rising mass of saturated air form a curved line. Furthermore, the curve thus formed approximately coincides with the average temperature-altitude curve of the free áir throughout all cloud levels, or from 0.5 kilometer, say, to 9 kilometers, or thereabouts, above sea level. This agreement necessarily occurs more or less closely during every rain and in all deep clouds and, therefore, very frequently. Nor can there often be much departure from it between such occasions for during these intervals the whole of this portion of the atmosphere is, as a rule, simultaneously warmed or cooled, and thus the curve in question usually shifted essentially parallel to itself.

It appears, then, that the average tempera

ture gradient (rate of decrease of temperature with elevation) of the free air is approximately that of a rising mass of saturated air; and for the reasons (a) that frequently the air is rising and saturated, and (b) that departures from the thus established saturation curve develop but slowly, as explained, and are soon eliminated by its reestablishment.

THE ISOTHERMAL STATE OF THE UPPER AIR

In April, 1898, Teisserenc de Bort began at Trappes, near Paris, a long series of frequent atmospheric soundings with small balloons carrying automatic registering apparatus.

Among other things, he soon obtained temperature records that indicated the existence either of surprising errors in his apparatus, or of wholly unsuspected conditions in the upper atmosprere. The records generally were tolerably satisfactory up to some 10 to 12 kilometers-satisfactory, because through at least the upper half of this region they showed the temperature to decrease with elevation at, very roughly, the adiabatic rate for dry air. But somewhere in the neighborhood of 11 kilometers elevation everything seemed to go wrong, for from here on the records no

longer indicated a decrease of temperature with increase of elevation, but often even a slight increase! There were but two possible conclusions. Either the apparatus had developed, in actual use, faults that the cross questioning of the laboratory had failed to reveal, or else the upper atmosphere really was in a most unorthodox thermal state. However, numerous records obtained with sounding balloons at different places, by different people and with different apparatus all showed the same thing, namely, that the temperature of the upper atmosphere, though varying slightly from day to day, is, at any given time, substantially the same at all levels, as illustrated by Fig. 1.

Here, then, was a conflict between observational evidence and tradition. Actual measurements had declared the upper atmosphere to be essentially isothermal-declared it in the face of a tradition to the effect that the temperature of the atmosphere must steadily decrease to, or very nearly to, the absolute zero. The name of the joker who first perpetrated this scientific hoax may be lost to fame, but the worst of it is we physicists thoughtlessly perpetuated it. The qualification, thoughtlessly, is used advisedly, for it seems impossible than any process of reasoning could have led to such an erroneous conclusion. If the surface temperature of the earth is maintained, as we know it is, by the absorption of solar radiation, it is equally certain that in turn the temperatures of objects in the full flood of the necessarily equivalent terrestrial radiation can not drop to zero; nor, therefore, can the air, generally, cool by convection to a lower temperature than that which this radiation can maintain. These ideas, so simple that they seem hardly worth expressing, embody the fundamental explanation of why the upper atmosphere is essentially isothermal.

In addition to being exposed all the time to earth radiation the upper air is also exposed much of the time to solar radiation, but there is abundant evidence that the atmosphere at all levels is far more absorptive of

the relatively long wave-length terrestrial radiation than of the much shorter wavelength solar radiation. Hence in computing from á priori considerations the probable temperature of the isothermal region, or stratosphere, as it generally is called, it is sufficient, as a first approximation, to consider the effect of only the outgoing radiation, which, according to the work of Abbot and Fowle, of the Smithsonian Institution, is approximately equal in quantity and kind to that which would be emitted by a black surface coincident with the surface of the earth and at the temperature of 259° A. As a further simplification the surface in question may be regarded as horizontal and of infinite length and breadth in comparison to any elevation attained by sounding balloons, and, therefore, as giving radiation of equal intensity at all available altitudes.

Now consider two such surfaces, parallel and directly facing each other at a distance apart small in comparison to their width, and having the absolute temperature T,, and let an object of any kind whatever be placed at the center of the practically enclosed space. Obviously, according to the laws of radiation, the final temperature of the object in question will also be approximately T.. If, now, one of the parallel planes should be removed the uncovered object would be in substantially the same situation, so far as exposure to radiation is concerned, as is the atmosphere of the isothermal region in its exposure to radiation from the lower atmosphere. Of course each particle of the upper air receives some radiation from the adjacent atmosphere, but this is small in comparison to that from lower levels and may, therefore, provisionally be neglected. Hence the problem, as an approximation, is to find the temperature to which an object, assumed infinitesimally small, to fit the case of a gas, will come when exposed to the radiation of a single black plane at a given temperature, and of infinite extent.

But whether an object lies between two planes of equal temperature, as above assumed, or, like the upper air, faces but one,

it clearly is in temperature equilibrium when and only when it loses as much energy by radiation as it gains by absorption. Furthermore, so long as its chemical nature remains the same its coefficient of absorption is but little affected by even considerable changes of temperature. Therefore, whatever the nature of the object, since it is exposed to twice as much radiation when between the two planes as it is when facing but one, it must, in the former case, both absorb and emit twice as much energy as in the latter. That is,

STORM EFFECTS ON TEMPERATURE GRADIENTS

Another surprising and, for a time, disconcerting contribution of the sounding balloon to our knowledge of the air relates to the relation of the temperature of the atmosphere to storm conditions. It has long been known that, in general, areas of low pressure-cyclonic areas are accompanied by inwardly spiralling winds and precipitation; and, conversely, that areas of high pressure-anticyclonic areas are characterized by outwardly spiralling winds and clear skies. Certainly, then, the inwardly flowing winds of the cyclone must ascend, and the outwardly flowing winds of the anticyclone must be sustained by descending currents. And the next inference, namely, that the air of the cyclone is relatively warm and the air of the anticyclone comparatively cold, seemed equally certain; for, indeed, what else could cause ascent in the one case and descent in the other? But again the facts are not in accord with the

simplest and most obvious inference, but just the reverse, through all convective levels, that is, up to the base of the stratosphere, as shown by Figs. 2 and 3, except, in general, near the surface, during the winter. In short. quite contrary to familiar ideas about convection, the ascending air in this case is relatively cold and the descending air comparatively warm. And the stratosphere, as these figures also show, but further confounds this confusion, for here the temperature relations are again reversed, the warmer air being now over cyclones and the colder, above anticyclones.

The facts just stated were, indeed, for a time somewhat disconcerting, but they have helped to the realization that with reference to temperature there are two classes of extra

tropical cyclones, cold (migratory) and warm (stationary); and also two classes of anticyclones, warm (migratory) and cold (stationary).

That the atmosphere of a stationary anticyclone should average relatively cold, and that of the cyclone comparatively warm, is obvious from the fact that the former occurs only

The migratory storms, however, at least those of middle latitudes, are quite different. The relation of their temperatures to each other, level for level up to the stratosphere, is just the reverse of that which it would have to be if their circulations were of immediate

becomes, through a portion of its course, the prevailing winds from the west, that up to near the base of the stratosphere average stronger, and are more nearly constant in direction, with increase of altitude. Now, whatever the origin of the migratory anticy