of upper winds shown by the higher clouds, namely, south-westerly in the northern hemis heré and north-westerly in the southern.

Halley in 1680, antfl Hadley in 1735. ave erroneous or imperfect explanations of the mechanical rinci is that bring about these winds. As some errors in regard) to this subject are still current, it is necessary to say that it is erroneous to teach that atmospheric air weighs less on being heated, or by reason .of the infusion of more moisture, and that therefore the barometer falls. The addition of more moisture must increase its weight as a whole; heat, being im onderable, cannot directly affect its weight either way. We are liagle to disseminateerrorby the careless use of the world “ lighter, " since it means both a diminution in absolute weight and a diminution in relative weight or specific gravity. Heat and moisture may diminish the specific gravity of a given mass of air by increasing its volume, or of a given volume by diminishing its mass, but neither of them can of themselves affect the pressure shown by the barometer so far as that is due to the weight of the atmosphere. It is not proper to say that by warming the air, thereby diminishing its specific gravity and causing it torise, so that colder air fiows in to take its place, we thereby diminish the barometric pressure. It is easily seen that in the expression ${\displaystyle \scriptstyle p=\mathrm {RT} /v}$, which, as we have before said, is the law of elasticity, ${\displaystyle \scriptstyle \mathrm {T} }$ and ${\displaystyle \scriptstyle v}$ may so vary as to counterbalance each other, and allow the pressure ${\displaystyle \scriptstyle p}$ to remain the same. Within any given room or other enclosure hot air ma rise on one side, How over to the opposite, cool and return, and the circulation be ke t up indefinitely without any necessary change in pressure. Tfie problem of the relation between wind and pressure in the free atmos here is more complex than this, and involves the consideration of the inertia of the masses of air that are in motion with the earth around its axis. The air is so extremely mobile that it moves quickly in response to slight differences in pressure that cannot be detected by ordinary barometric measurement. The gradients or differences of pressure that are shown on meteorological charts are not directly, but only very indirectly, due to buoyancy, as caused by heat and moisture. The pressure gradients, so-called, are not merely the prime causes of the winds, but are equally and essentially the results of the winds. They are primarily due to the fact that the atmosphere is rapidly revolving with the surface of the earth around the earth's axis, while at the same time it ma be circulating about a storm centre. Inappreciable differences ofy pressure start the winds in motion, and the air moves towards the region of low pressure, just as in the pneumatic despatch tubes the flow of air towards the low pressure carries the packages along. But in the free air, where there are less important resistances to be overcome, the freedom of motion is greater than in these pneumatic tubes. No sooner is the atmosphere thus set in motion by pressure from all sides towards the central low pressure than it rapidly acquires a spiral circulation, and thereb there is superimposed (in the northern hemisphere) a decided diminution of pressure on the' left hand side of the wind, and an equally rapid increase on the right hand side. The gradient of pressure in the direction of the wind overcomes resistances, but the gradient of pressure, perpendicular to the direction of the wind, is far greater than that in the direction of the wind, and is that which produces the areas of decided low ressures that appear as storm centres on the daily weather map. Therefore, in general, the wind cuts across the charted isobars in obli ue directions and at angles which are nearly'9o° for the feeble winds far removed from the centres, but which are almost zero for the most violent winds near the low centre. The winds acquire this spiral circulation for two reasons-(a) all straight line, gusts or jets in fluids, subject to any form of resistance, necessarily break up into rotating spirals whenever the velocit exceeds a certain limit, because the resistances deprive some particles of the fluid of a little more of their original velocity and energy than the other particles near by them, and thus the whole series is drawn away from linear into curvilinear paths; (b) in addition to their rectilinear motions the particles of air have a rapid circular motion in common with the whoe atmosphere diurnally around the earth's axis. Therefore every particle of moving air comes under the influence of a set of forces depending on its own rate of motion relative to the earth's surface and its position relative thereto. If the rticles are moving eastward, viz. in the same direction as the eartiiéls diurnal rotation, then the result is as though the atmosphere were rotating more rapidly than does the earth at present; conse uently the particles of wind push toward the equator as thou h flue atmosphere were trying to adopt a more flattened spheroidal figure corresponding to its greater velocity of rotation. If the particles are moving westward, on the other hand, it is as though the atmosphere were revolving less rapidly than the earth, and as though the flattened apheroid of revolution due to the present rate of rotation were more ecidedly flattened than need be; consequently the particles of air push towards the poles. If the winds blow toward either pole, then their initial moment of inertia about the earth's axis, due to the initial radius and the eastward movement of the air, must be retained; consequently, as the air advances into higher latitudes and to smaller circles of diurnal rotation its velocity must increase, and must carry the particles to the east of their initial meridians. If the wind blow towards the equator its initial moment of inertia must be a plied to a larger radius, and its velocity correspondingly diminished so that it is left behind or falls away somewhat to the west. “ The reasoning of those who in attempting to ex lain the trade winds assume that the atmosphere in moving toward) or from the equator has a tendency to retain the same original linear velocity is erroneous "(Ferrel's Movements of Fluids, 1859). In general the winds tend to retain their moments of inertia, and in the northern hemisphere must necessarily always be deflected continuously toward the right hand. The exact amount of this deflection was first distinctly stated by Poisson,^[1] as applied to the movements of projectiles; it was also announced b Tracy of New Haven in 1843, ut was first applied to the atmospiiere by F errel, who deduced its meteorological consequences. This law is not to be confounded with that of Buys Ballot, who in 1861 deduced from his observations in Holland the rule that the gradient of pressure between two stations for any day would be followed in twenty-four hours by a wind perpendicular to that gradient, and having the lower pressure on the le t hand. Buys Ballot's law was in the nature of a rule for prediction, and was modified by Buchan 1868, who enunciated the ollowing: “ The wind blows towards the regions of low pressure, but is inclined to the gradient at an angle which is less than 9o°.” In this form Buchan's law was an improvement upon the laws current among cyclonologists, who had assumed that, in a rough way, the wind blew in circles around the low centre, and was therefore sensibly at right angles to the gradient. It ought, however, to be said t at Redfield throughout the whole course of his studies, from 1831 to 1857, never gave adherence to this view, and in fact for the severer portions of hurricanes determined the average inclination of the movements of the lower clouds at, New York City to be about 7° inwards as compared with the truly circular theory. Now Ferrel's law explains mechanically the reason why the winds do not blow either radially or circularly, and gives the means for determining their inclination to the isobars in all portions of the cyclone and for various degrees of resistance by the earth's surface. The general proposition that the barometric gradients on the weather map are not those that cause the wind, but are, properly speaking, the result of the combined action of the wind, the rotation of the earth, and the resistances at the earth's surface, as first explained by Ferrel, seems to have been neglected by meteorologists until brought to their attention repeatedly by Professor Abbe between 1869 and 1875, and especially by Professor Hann in a review of Ferrel's work (see Met. Zeit. 1874). The independent investigations of Sprung, Koeppen, Finger, and especially Guldberg and Mohn, confirm in general the correctness of Ferrelfs law.

It is quite erroneous to imagine that the low pressures in storm areas and in the polar regions, and especially the belt of low pressure at the equator are due simply to the diminution of the density and weight of the air b the action of its warmth or its moisture, or to the abundant rainflaill as relieving the atmosphere of the weight of water. It has been clearly shown that none of these operations can directly affect the barometric pressure to any appreciable extent, but that high and low pressure areas, as we see them on the weather map, owe their existence entirely gto the mechanical interaction of the diurnal rotation of the earth and the motions of the atmosphere. The demonstration of this point by F errel in 1857 is considered to have opened the wail for modern progress in theoretical meteorology. Both Es and ann have abundantly shown that the formation and downtgll of rain do not produce any low barometric pressure unless they produce a whirling action of the wind-that, in fact, the latent heat evolved by the condensation of vapour into rain may so warm ptp the cloud as to roduce a temporary rise in pressure even at the su ace of the ground? due to the outward push produced by the sudden expansion of the cloud. [The details of the thermodynamics of, this operation have been elucidated by Wm. von Bezold.] The .force with which the wind presses to the right or tends to be deflected in that direction is ${\displaystyle \scriptstyle 2nv\sin \phi }$, while the curvature of the path of the wind is measured by its radius of curvature, which is ${\displaystyle \scriptstyle v/2n\sin \phi }$, where ${\displaystyle \scriptstyle v}$ is the velocity of the wind, ${\displaystyle \scriptstyle n}$ is the equatorial velocity of the earth's rotation, and ${\displaystyle \scriptstyle \phi }$ is the latitude. It will be seen from this that there is no deflection at the equator; therefore, as Ferrel stated, there is no tendency to the formation of great whirlwinds at the eqfuator, hence hurricanes and typhoons are rarely found within 10° o the equator.

Ferrel fre uently speaks of an anti-cyclone, whereby he means the area of (liigh pressure just outside of a strong cyclonic whirl; the expression perl-cyclone would have been more a propriate and is sometimes substituted. The term anti-cyclone, as lilrst introduced by Galton in 186% is applied to a system of winds blowing out m a central area of igh pressure, and this is the common usage of t e term in modern meteorolo . The term cyclone among meteorologists and throughout English literature, except only a few cases in the United States, is equivalent to the older usage of whirlwind, and it is unfortunate that misunderstandings often arise because local usages in America apply the word cyclone to what has for centuries been called a tornado. The mechanical principles discussed by Ferrel led him to an algebraic relation between the barometric gradient ${\displaystyle \scriptstyle \mathrm {G} }$, the wind velocity ${\displaystyle \scriptstyle v}$, the radius of curvature of the isobar ${\displaystyle \scriptstyle r}$, and the inclination ${\displaystyle \scriptstyle i}$ between the wind and the isobar, which is

1Recherche sur le mouuement des projectilerdans l'air en ayant égard d Vinjiuence du movement diirrne de la terre; dated 1837, printed Paris, 1839.