of the winds round the centre is altogether inadequate, on the principle of centrifugal force, to bring about the low pressure at the centre. Thus, if the wind blows round a circle 300 miles in diameter at the rate of 70 miles an hour, the centrifugal force would only depress the barometer at the centre a little more than 0.02 inch ; whereas in the Bahama hurricane the difference of the pressure at that distance from the centre was 2 inches, or 1000 times greater than that due to centrifugal force.

441. But the spiral rotation instead of the circular rotation of the winds in storms completely alters the whole complexion of the question affecting the theory of storms. For since it follows from it that enormous quantities of air are constantly being poured all around into the area of the storm, and since, notwithstanding these accessions tending to increase the pressure

, observation shows that the pressure is not thereby increased, but on the contrary sometimes diminished, we are compelled to the conclusion that from a large area within and about the centre of the storm a vast ascending current must arise to the upper regions of the atmosphere. Now all observation shows that in front of storms the air is warm and moist, and it is there that most of the rain accompanying storms falls. Thus far observation ; but may we not go a step further, and say that it is the heat of condensation into rain of the vapour of the warm moist air in front of the storm when it has been drawn up into the higher regions of the atmosphere, which relieves the barometric pressure and thus forms the motive power of the storm ? If this be the case, then the onward course of the storm will follow that of the upper current within which the condensation of the vapour into rain takes place, –a conclusion which observation appears at present to point to. The theory is also in accordance with the observation that storms generally contract in area as they traverse a dry climate, and die out when they meet or come against the dry polar current, because under such circumstances less vapour being condensed, less latent heat is given out and the barometric disturbance is consequently diminished.*

* This view of the part played in storms by the heat of condensation was first proposed by Brandes, and since adopted by Espy, Hare, Russell,

and some others, as part of the theory of storms. It cannot be doubted that the heat given out in the condensation of vapour into rain must play an important part in storms. We have already seen in art. 76, that wherever a copious rainfall prevails over any considerable portion of the earth's surface, there the mean atmospheric pressure is low. We have stated that Espy entered the barometer on his storm-charts on a defective and misleading principle; and from this he may have been led to maintain that the

442. The part of the track of the West Indian hurricanes after turning to the N.E., is in accordance with this theory; but how account for the first part of their course, which is at right angles to the prevailing trade-winds of that region ? The usual place where the vapour brought by the trade-winds is condensed, is the region of calms, where heavy rains and thunderstorms daily occur. But since this condensation takes place simultaneously over a somewhat broad belt of the earth's surface, which for the time is stationary, it follows that the storm is neither rotatory nor progressive, the only effect of the condensation being the flow of the regular trade-winds towards the belt where it takes place. When the condensation is more copious than usual, the effect will be the acceleration of the speed of the trade-winds. This is the most probable explanation of the harmattan which occurs at certain seasons on the coast of Africa ; it always blows in one direction from the land, and does not increase quite to the violence of a gale, and then dies away. Copious rainfall in the belt of calm opposite that part of Africa, and at some distance, is the probable cause of these winds. Similar in some respects to these are the tornados of western Africa, which blow invariably off the land, first with little force, but ultimately they rise to the strength of a heavy gale, and then after an hour, or sometimes two hours, die away. The direction of the wind remains all the time unchanged; and the barometer varies little, if any, during their continuance. They are generally accompanied with rain and thunder, though sometimes they are quite dry winds. They are probably caused by very heavy rains in the region of calms, falling over a limited region and lasting only for a short time.

443. Let us suppose that the atmosphere of the West Indian Islands has from some cause become exceptionally warm and moist, and that, at the same time, a high barometer is interposed between that region and the belt of calms. In such circumstances, as the trade-winds cannot flow towards the belt of calms, the usual provision for draining them of their moisture is taken away, and a rapid accumulation of aqueous vapour, to the north of the high barometer, takes place, ready to burst in any instant in rain and

direction of the wind in storms is centripetal—an assertion which neglects
theoretically the influence of the earth's rotation on its axis in changing the
direction of the wind, and practically the evidence to the contrary furnished
by his own storm-charts. But this is no reason for disregarding that part
of the theory which refers to the heat of condensation. The centripetal
theory has long been abandoned as impracticable by the Meteorological
Department of the United States, and the spirally in-moving course of the
winds adopted instead.

storm. This would appear to have been the case before and during the Bahama hurricane of October 1866, Plate V. The following table shows the atmospheric pressure and winds at St Croix, and on board H.M.S. Buzzard, which sailed from Barbadoes on the 26th September for England. St Croix is in N. lat. 17° 44' 29" and W. long. 64° 41', and the mean height of the barometer at sea-level is 30.080 inches.

BAROMETRIC PRESSURE AND WINDS OBSERVED AT ST CROIX, WEST INDIES,

AND ON BOARD H.M.S. BUZZARD, FROM 26TH SEPTEMBER TILL 4TH
OCTOBER 1866.

From this table we learn that during the hurricane, which in all probability began a little to the north of Barbadoes on the evening of the 26th or morning of the 27th September, the atmospheric pressure was above the average to the S.E. and E. of the region traversed by the storm. This being the case, the air would naturally flow from the high toward the low barometer—that is, a general movement of the atmosphere would set in to the N.W., which was the course pursued by the storm for the first four days, a course at right angles to the usual trade-winds. During this part of its course, the rate at which it travelled was slow, not exceeding fifteen miles an hour. Such a slow, and often a much slower, progressive motion is common to these storms until they enter the region of the return trades, when the course is changed to the N.E., becomes greatly accelerated, and the storm itself expands over a wider area.

444. Newspaper accounts of the weather before the hurricane describe it as close and sultry, the whole atmosphere being ab

normally heated and excessively loaded with moisture. At Ber-
muda, at 9 A.M. of the 4th October, the temperature of the air was
799.4, and the dew-point 739.6, the humidity, therefore, being 83:
and this kind of weather had been prevailing for fully a week
before. On the following day, the 5th, at 9 A.M., when the storm
had just passed, the temperature was only 71°, the dew-point 56°,
and the humidity 59o. Next night the temperature fell to 56°,
whereas, before the storm, it did not fall on any night lower than
72°. It would be interesting to know if West Indian hurricanes
occur simultaneously with a high pressure to the south, being
thus interposed between that region and the usual position of the
belt of calms at that season. If so, a considerable step will be
gained toward understanding the causes of their origin, and how
it is that they are, happily, phenomena of very rare occurrence.
The barometric pressure in the south-eastern States of America
would have possessed peculiar interest; for if the pressure had
been low there, it would have still further accounted for the
course of the hurricane to the N.W. But the meteorological ob-
servations, which had been discontinued there during the war,

had
not then been resumed.

445. Perhaps no writer in meteorology denies the influence of low pressures in drawing the winds toward them ; but the effect of this influence in accounting for the rotation of storms round their centres is in general tacitly subordinated to a theory, by which the storm is conceived to rotate from some force quite distinct from that arising from differences of pressure, and in virtue of this rotation the air to be driven to the outside of the whirl, thus diminishing the pressure at the centre; much in the same way as, in whirling round a pail of water, part of the water leaves the centre and rises up against the sides of the vessel. In applying this theory to cyclones, the wind at every point within the storm is stated to be under the influence of two forces, the one arising from the rotatory motion of the cyclone, and the other from its progressive motion. To illustrate this by an example: Suppose a storm advancing over England to the eastward at the rate of twenty-five-miles an hour, and taking the simplest case of the wind whirling round within the storm at the rate also of twenty-five miles an hour, then both the rotatory motion of the winds and the progressive motion of the storm are from the west, at the extreme south point of the storm; hence there the wind ought to be due west. At the extreme east point of the storm, since the direction of the wind due to the rotatory motion is south, and by the progressive motion west, the wind ought to be seen blowing there from the south-west. At the extreme north,

since the rotatory motion is east, and the progressive west, and both equal, there should be no wind at all. Lastly, at the extreme west point of the storm, the rotatory motion being north, and the progressive west, the wind should be north-west. Further

, since in the south half of the storm the direction of the two motions is generally the same, and in the north half generally opposite, the gale ought to be most severely felt in the southern half of the cyclone, attaining its height at places immediately south of the centre, and least in the northern half. When the rotatory motion much exceeds the progressive, the deviations from the true circular course of the winds when laid down on a synchronous chart would be less than those stated above ; but in all cases, unless when the cyclone was stationary, the winds in the front part of the storm ought to be observed blowing out of the circular area of the storm. I am altogether at a loss to discover a storm to which this theory applies. If we look at the West Indian cyclone of October 1866, Plate V., we do not see any of the winds observed at 8 P.M. show even a tendency to blow in a direction in accordance with the theory, but, on the contrary, they all blow round and in upon the centre. To the European storm of the 2d November 1863, Plate IV., the same remark applies, as well as to every storm in tropical and temperate regions which I have examined on synchronous charts. The light winds which often prevail in the northern half of European storms are sometimes referred to in illustration of this theory; in such cases it will be found, on examination, that there the isobarometric lines are considerably apart (see art. 419). But when it happens that the isobarometric lines are much crowded together on this side of the storm, the winds are strong and violent. The parts of a storm where probably the winds are not generally proportioned to the difference of pressure, are, that part of the front where the ascending current is strongest, thus diminishing the force of the surface-current, and that part of the rear where the air is coldest and driest, and, as the clouds often show, descending currents prevail.

446. It should be kept in mind that this theory was not arrived at through the slow and tedious process of a rigid induction-viz., by collecting the observations of the barometer and winds on synchronous charts, from which, their mutual relations having been observed, a theory was then constructed in strict accordance with the facts; but from observations, necessarily at the time few and scattered, it was inferred that the rotation of storms is circular, and this inference necessitated another-viz., that the rotation is caused by some force or forces acting on the storm from without.

;