in which the Monsoons change, and that the absolute maximum is observed at the change from the SW. to the NE. Monsoon.

As regards the velocity of progression, it appears that this is occasionally so slight in the case of the Typhoons,. that Capper seems to have considered them to be stationary and not progressive.

*

The most remarkable instance of slow rate of motion of a cyclone is given by Piddington in the 13th 'Memoir with reference to the Law of Storms, being the "Charles Heddles" storm of the Mauritius from the 22nd to the 27th of February 1845.' The progressive motion during each of these five days respectively was 70, 100, 115, 89, 85 English miles. This gives an average motion of 92 miles per day, or 3.8 per hour; while the ship itself, at a distance of about 50 miles from the centre, was carried round and round the cyclone in such a manner that she sailed 1,300 miles, and yet found herself after the five days at a distance of only 354 miles from the port from which she had started, as the storm had made five complete revolutions with her. This is, as Piddington says, a sort of sea-romance, or, to speak like a good Mussulman, it is something which might have happened, if it had so pleased the Prophet.

In the year 1840 I published † an attempt to explain the origin of these phenomena, which is as follows:

Let a b represent a series of material points, in the northern hemisphere, parallel to the equator, which are set in motion, by any cause, towards the pole in the direction a c. If the space d b h were a vacuum, these points would move to the position g h, inasmuch as they

*Observations on the Winds and Monsoons. London, 1801.

+ Berichte der Berliner Akademie. (Reports of the Academy of Berlin.) 1840, p. 232. Pogg. Ann. lii. p. 1.

с

d

е

travel from greater to smaller circles of latitude. If, however, the space d b h contain air which has not been set in motion in the same direction, the particles at b in their motion towards d will constantly be brought into contact, in that space, with particles of air which are rotating with a less

α

velocity than they are themselves, and, consequently, their velocity of progression towards the east will be diminished. For this reason the point b will move to f, instead of to h. The particles at a are differently circumstanced, as they are in contact, in the direction of b, with particles which possessed originally a velocity of rotation equal to their own, and, accordingly, their motion takes place as if in vacuo: i.e. they move towards g. It follows from this that if a b be a mass of air travelling from south to north, the direction of the motion of the air on the eastern side of the mass will be much more nearly that of a south wind, than that of the air on the western side, where it is more of a west wind, and the effect of this difference will be to generate a tendency to a cyclonical movement in the direction S., E., N., W. Such a tendency as this could not exist if the space d b h did not contain a resisting medium, and will therefore increase in proportion as this resistance interferes with the force which ought to deflect the wind towards the west point. Hence the storm will rotate with greater velocity the less the path along which it travels varies from the original direction in which the motion took place (i.e. towards c.) In the zone of the Trade-winds the space d b h is filled with air, which is moving from NE. towards SW. In this case the

greatest resistance which is possible will be presented to the motion of the mass a b, and it will become possible that the particles at b should be prevented to such an extent from taking the direction of a west wind, that they will actually move towards d, while particles at a tend to move towards g. Accordingly, in this district, the violence of the rotatory motion will be greatest, but the cyclone will advance in a straight line, and its diameter will not increase. As soon as such a storm enters the temperate zone the circumstances are altered, the space d b h is filled with air which is already in motion from SW. towards NE. The resistance which had been experienced by the particles at b, either undergoes a sudden and considerable diminution, or else disappears entirely: the direction of motion. along the line b d is changed for one along bh, so that the path of the cyclone changes so as to be nearly at right angles to its former course; and its diameter increases rapidly, owing to the disappearance of the obstacles which produced a difference between the motions of the particles at a and b respectively.

The explanation in the case of the southern hemisphere follows from the same principles, as may be seen from the diagram. The direction of the rotation of the cyclone

is the reverse of what it is in the northern hemisphere, and the change of the direction of the path at the edge of the torrid zone is analogous.

The preceding explanation of the origin of the rotatory motion is of course only applicable to the cases in which large masses of air of considerable lateral extension are set in motion: smaller whirlwinds (trombs), waterspouts, &c.,

arise from other causes, and we may therefore expect that, as regards the two hemispheres, their motions will exhibit neither conformability to a law, nor a distinct contrast. In fact, Gen. Reid saw, from the Government House at Bermuda, a waterspout which was rotating in a direction contrary to that of the great cyclones, while Mr. Redfield observed a small tornado which rotated in the same direction as the cyclones. The observations of Akin, at Green Bush near Albany, of Dwight, at Stockbridge, Mass., and of Dr. Cowles, at Amherst, on the violent whirlwinds which arise during forest conflagrations when the air has been calm, prove that a strongly developed courant ascendant is capable of generating a rotatory motion in the surrounding air.

The West India hurricanes arise at the inner edge of the zone of Trade-winds, in the so-called Region of Calms, where the air ascends and flows away in the upper strata in the direction opposite to that of the Trade-wind. This renders it probable that the primary cause of the cyclones is the intrusion of a portion of this upper current into that which lies underneath it. The reason that the cyclone, in the earlier portion of its course, travels from SE. towards NW., may probably be assigned by the following consideration. According to our theoretical explanation of their origin, a direction similar to that described will be most favourable to the generation of a cyclonical movement. If, as may sometimes happen, the first impulse takes place in the direction from SW. to NE., the north-east Trade, which is blowing in the contrary direction, will offer the same amount of resistance to all portions of the line (the front of the mass of air) which is advancing, and no tendency to a cyclonical movement can arise.

It is well known that, during the eruption of Coseguina on January 20, 1835, when the isthmus of Central

America was disturbed by earthquakes, volcanic ashes were carried by the upper counter-Trade-wind not only as far as Kingston in Jamaica, a distance of 800 English miles against the direct Trade-wind, but also 700 English miles to the westward, where they fell on the ship Conway in the Pacific Ocean. This fact proves that, in the higher regions of the tropical atmosphere, the air does not always move regularly from SW. towards NE., but that the regularity of this movement is interrupted by currents which flow from E. to W. It appears that there is some connection between these facts and the fall of dust, which has come from Africa, in the North Atlantic Ocean, an occurrence which is frequently observed. We must attempt to ascertain the cause of such abnormal currents.

Their origin is to be found in the tables of barometrical fluctuations which have been given (see pp. 54 sqq.). If we compare the annual barometrical curves for Hobarton, Port Jackson, Cape Town, Rio Janeiro, Santiago, Monte Video, and Buenos Ayres with those for Northern Asia, Eastern Europe, and Hindostan, which are represented on pp. 61, 62, we see that the pressure in the southern hemisphere increases at the period at which that in the northern decreases. The most obvious explanation of this would be to suppose the existence of a periodical transfer of the air from the one hemisphere to the other. However, we find that the mean annual tension of aqueous vapour at Barnaoul is 0.191 ins., while at Banjoewangie it is 0.897 ins., so that the supposition of an exchange would be a tacit assumption that air can be converted into water, and vice versa. We are therefore compelled to return to the action of the dry air. A glance at the tables (pp. 52-60) will show that the aggregation of air in the southern hemisphere is insufficient to explain the rarefaction on the northern. It is therefore clear that the