take the average pressure of dry air" over large areas, so as to eliminate local irregularities, it is still certain that the resulting diurnal variation of this pressure of dry air is not much, if any, simpler than the diurnal variation of the barometer. All this is pretty certain, and yet, to my mind, it by no means follows, as some meteorologists have, perhaps, too hastily assumed, that the semidiurnal wave of pressure is not mainly due (indirectly though it may be) to the presence of aqueous vapour. We have to begin with the fact already mentioned that in the interior of great continental areas, where the air is necessarily dry, the night minimum becomes very small, whereas in maritime situations it is great. Again, the annual averages of vapour tension for each hour of the day at land stations give a curve with a strongly-pronounced double maximum and minimum. Also in such a place as Kimberley if we resolve both the diurnal variation of pressure, and of vapour tension, into their harmonic constituents, the epochs of the second, third, and fourth harmonic terms are almost the same in each. And it is also a curious fact that if from the total barometric pressure we subtract one-third of the observed vapour tension for each hour the resulting differences fall upon a curve which contains practically no third or fourth term. Now, it has been argued that the resemblance between the annual curves of the diurnal variation of the barometer and of the vapour tension is purely fortuitous since it is not found in the monthly curves. Such an argument does not seem to me to be quite sound. For although the diurnal curves of barometric pressure and vapour tension do differ from each other month by month, yet such variation as there is is in the same direction. For instance, if we draw winter, summer, and annual curves of the diurnal variation of barometric pressure and dew point as observed at Kimberley, we find that the winter deviations of both are positive before noon and negative after, as compared with the annual curves of each, and vice versâ in

summer.

Now, a moment's consideration will shew that this is exactly what it should be in all places where the relative humidity is not great, if the observed variations of barometric pressure indicate any movements at all of the air. For, supposing at any instant we have in a given space a barometric pressure of 26 140 inches, and that the vapour tension is 1 % of this, i.e., 2614 inch. Suppose further for any reason an indraught of air of the same composition, not necessarily lateral, raising the barometric pressure pretty quickly to 26.240 inches. Is it not clear that the vapour tension must have increased in the same time to 2624 inch? Therefore I say that all variations of barometric pressure that depend solely upon air movements must give rise to similar variations in the vapour tension. But it is truly said that over the ocean, or at great altitudes, the semi-diurnal curve of vapour tension does not exist, and that the observed curve is to all appearance a temperature curve. No doubt! But, then, it must be borne in mind that in all places where the relative humidity is high movements of air are as likely as not accompanied by changes of state in its contained water vapour.

When the air is unsaturated the barometric pressure may continually increase, and the vapour tension may increase in the same ratio; but when the saturation point is reached the proportion no longer holds, and the curve of vapour tension must necessarily be controlled by the temperature of the air. At the dew point a fall of temperature means in general a deposit of dew, or some condensation of moisture, and therefore a decreased tension; a rise of temperature in the presence of water no less means evaporation. Consequently if we are looking for relationships between barometric pressure and vapour tension, over the ocean will not be the place to find it.

Dalton's explanation of the diurnal oscillation of the barometer alluded to before was partly based on suppositions respecting the behaviour of the vapour of water which, under certain circumstances, are undoubtedly realities. Since it seems not to be generally known, and also refutes certain opinions which a later generation has credited him with, I take this opportunity to quote it :

"Another fact respecting the variation of the barometer indeed appears to be well established, by the attentive and careful observation of that instrument. I mean the diurnal variation first observed in the torrid zone, and since then traced through the temperate zone, though it is there blended with other and more powerful variations from other causes.

"Generally speaking, the fact is this: that early in the morning, about sunrise or soon after, the barometer is higher, all other circumstances the same, than it is afterwards; that it droops a little as the heat of the day advances, and is lowest nearly in the warmest part of the day; after which it rises as the air cools, and in the evening nearly recovers what it had lost since morning.

"The sun's power being greatest in the torrid zone, this effect of it (for it is evidently an effect of temperature) is there a maximum; and on this account it is more conspicuous there, as well as on account of the other variations being of less magnitude than in the temperate and frigid zones. The effect diminishes, in leaving the equator, in some proportion as the latitude, the seasons, and other circumstances.

"The sun is constantly heating the earth and air successively from east to west the air being heated expands in various direction to restore an equilibrium of pressure; if this expansion was only in a perpendicular direction, it would not disturb the barometer; but as the air will go in any direction where the pressure is least, it has a lateral motion as well as a perpendicular one; and hence the column pressing on the mercury is less in quantity during the high temperature; but when the excess of temperature is withdrawn, the air falls back into its former position.

"It will be perceived that the principle we have adopted in the Essay on the Variation of the Barometer is that an equality of elasticity in two vertical columns of air will in great part counteract an inequality in their weight. To illustrate this

position suppose a cylindric vessel of indefinite length were filled with hydrogen gas, and placed perpendicular to the ground plane, having no communication with the atmosphere; suppose then that a small hole were made in the side of the said vessel at a point where the elasticities of the two gases were equal. A communication being now open an intercourse would immediately commence; but this would not be occasioned by the specifically heavy gas rushing into the light gas, nor the light gas into the heavy gas exclusively. The two gases, having equal elastic forces, would by virtue of those forces be diffused through each other slowly and gradually according to the law. which I have pointed out in another Essay,—that of one gas being as a vacuum to another in regard to their mutual diffusion.

"The application of this principle in accounting for a temporary existence of a warm, vapoury volume of atmosphere supporting itself against heavier, but colder, volumes of atmosphere on the right and left of it is too obvious to be insisted upon.

"From these observations it is not difficult to conceive that a vertical column of warm, vapoury air may be projected into a heavier column of cold, dry air, such that their elasticities may be nearly equal for a time, but that the adjustment of their weights may require a slow and gradual operation, sufficient to account for the interval of time observed between the extreme and mean state of the barometer."

Buchan made the first material advance since the time of Dove in explaining the barometric oscillations, and he followed Dalton in distinguishing between weight and elasticity, without going to the lengths of claiming that the pressure on the cistern of a barometer at any instant is that of the whole weight in the vertical column of air resting upon it. His view I also quote in full from the ninth Edition of the Encyclopædia Britannica:—

"If the temperature of the whole of the earth's atmosphere were raised, atmospheric pressure would be diminished, for the simple reason that the mass of the atmosphere would thereby be removed to a greater distance from the earth's centre of gravity. Quite different results, however, would follow if the temperature of only a section of the earth's atmosphere were simultaneously raised, such as the section comprised between Long. 20° and 60° W. The immediate effect would be an increase of barometric pressure, owing to expansion from the higher temperature; and a subsequent effect would be the setting in of an ascending current more or less powerful, according to the differences between the temperature of the heated section and that of the air on each side. These are essentially the conditions under which the morning maximum and afternoon minimum of atmospheric pressure take place.

"The earth makes a complete rotation round its axis in 24 hours, and in the same brief interval the double-crested and double-troughed atmospheric diurnal tide makes a complete

circuit of the globe. The whole of the diurnal phenomenon of the atmospheric tides is therefore rapidly propagated over the surface of the earth from east to west, the movement being most rapid in equatorial regions, and there the amplitude of the oscillations is greater than in higher latitudes under similar atmospheric, astronomical, and geographical conditions. Owing to the rapidity of the diurnal heating of the atmosphere by the sun through its whole height, some time elapses before the higher expansive force called into play by the increase of temperature can counteract the vertical and lateral resistance it meets from the inertia and viscosity of the air. Till this resistance is overcome, the barometer continues to rise, not because the mass of atmosphere overhead is increased, but because a higher temperature has increased the tension or pressure. When the resistance has been overcome, an ascending current of the warm air sets in, the tension begins to be reduced, and the barometer falls and continues to fall till the afternoon minimum is reached. Thus the forenoon maximum and afternoon minimum are simply a temperature effect, the amplitude of the oscillation being determined by latitude, the quantity of aqueous vapour overhead and the sun's place in the sky. . .

"When the daily maximum temperature is past, and the temperature has begun to fall, the air becomes more condensed in the lower strata, and pressure consequently at great heights is lowered. Owing to this lower pressure in the upper regions of the air, the ascending current which rises from the longitudes where at the time the afternoon pressure is low flows back to eastward, thus increasing the pressure over those longitudes where the temperature is now falling. This atmospheric quasi-tidal movement occasions the PM. increase of pressure, which reaches the maximum from 9 p.m. to midnight according to latitude and geographical position. This maximum is therefore caused by accessions to the mass of the atmosphere overhead contributed by the ascending currents from the longitudes of the afternoon low pressure immediately to the westward.

"As midnight and the early hours of morning advance, these contributions become less and less, and at length cease altogether, and pressure continues steadily to fall. But between the time when the increase of pressure from the overflow through the upper regions of the atmosphere ceases and the time when pressure increases from the heat rays, direct or indirect, of the returning sun, or during the hours of the night when the effects of nocturnal radiation are at the maximum, pressure is still further reduced from another cause. Radiation towards the cold regions of space takes place, not only from the surface of the globe, but also directly from the molecules of the air and its aqueous vapour. The effect of this simultaneous cooling of the atmosphere through its whole height is necessarily a diminution of its tension. Since this takes place at a more rapid rate than can be compensated for by any mechanical or tidal movement of the atmosphere from the

regions adjoining, owing to the inertia and viscosity of the air, pressure continues to fall to the morning minimum. This minimum is thus due, not to the removal of any of the mass of air overhead, as happens in the case of the afternoon minimum, but to a reduction of the tension or pressure of the air consequent upon a reduction in the temperature through radiation from the aerial molecules towards the cold regions of space."

This differs somewhat (and for the better, I think) from Buchan's explanation given in the Challenger Report, in which the morning maximum is said to depend in part upon the "vitally important principle. that a portion of the aqueous vapour of the atmosphere passes from the gaseous to the liquid state, thus reducing the tension," and an inverse process as a contributing cause of the morning maximum.

The great virtue of Buchan's theory, to my mind, lies in his definite acceptance of the principle that a great portion of the barometric oscillation may be due to tension as distinguished from lateral movement, and rejection of the principle that the diurnal oscillation differs only in degree from the annual circulation. Indeed, the weak spot in his theory is in the hypothetical nature of these same lateral movements, and for which he adduces next to no evidence. The high velocities that would be requisite if an air particle is to move all the way from my to M-about a thousand miles an hour at the equator-seem to have escaped attention; and also the complicated nature of the circulation necessary to restore equilibrium. For the movement in the first instance will not be a simple flow from where the atmospheric strata are most elevated to where they are most depressed; but will (or should) operate most vigorously where the gradients are steepest. That is, any outward tendency of the wind due to an elevation of the atmospheric strata should display its most marked effect near the times of sunrise and sunset, when temperatures are changing most rapidly, and should be directed to some point not the shortest path down the slope. As we have seen, this is pretty much what happens at Kimberley, and, as I have shewn elsewhere, our normal wind directions at any hour are tangents to an imaginary spiral curling outwards from the light hemisphere and into the dark. Another point which seems to want demonstration, or, at any rate, elucidation, is the local character he endeavours to give to his theory, namely, that the barometric oscillations are generated by absorption and radiation in the regions where they occur. It is difficult to believe that physical processes which can only generate a total range of temperature over the ocean of one or two degrees can give rise to a range of pressure often greater than it is in continental places, where the range of temperature is twenty times as great. Moreover, it has yet to be shewn that the absolute humidity (upon which absorption depends), and the relative humidity (upon which radiation mainly depends), are materially less at great altitudes over the land than they are over the ocean. Över Kimberley, at any rate, the upper air must be nearly or quite as rich in aqueous vapour as the upper air is over the eastern Atlantic. For our upper currents seem to set almost constantly from