other causes than those which develop isobars; and for the sake of classification we will call them generically "non-isobaric winds." They are most probably connected with what we have before alluded to as non-isobaric rains.

We cannot say what is the origin of the wind in thunderstorms and non-isobaric winds, but it is certain that the cause is quite different from that in cyclones. We must therefore take care, in talking about wind, not to mix up two kinds which really have little in common. From all this we see the very fallacious results which come of trusting blindly to instruments, and also that any statistical values which are derived from mixing up various sorts of wind can only give rise to discordant deductions.

We may also remark that merely saying that a storm blew with such a force or velocity tells us very little either of the true character of the wind or of the amount of destruction which the gale might cause. An instrumental record of forty miles of wind in an hour may be made up either of a steady weight of wind, or of violent gusts alternating with quieter intervals. The damage done in the latter case would many times be greater than in the former.

of

Then there are many minute differences in the way of blowing which instruments cannot even detect. We all know that most chimneys smoke more with an east wind than with a west one. We have also just shown that the velocities of these two winds is not the same for the same gradients.

It has been suggested, with a great deal of probability,

that the difference may be due to the wind not blowing horizontally, and that east winds are perhaps directed a little downwards. Another very striking phase of wind is the difference between the kind of sea raised by the south-west gale in front of a cyclone and the north-wester in rear. The first raises a high sea with only a moderate amount of white water, while the latter rakes the surface of the ocean into long streaks of foam. There are other reasons for believing that in front of a cyclone the wind is rising, while in rear the air-currents have a slant downwards. If so, the cold, dry clearness of north-westers is readily explained. The whirls of dust that precede some kinds of rain are also familiar instances of the specific character which belongs to different winds.

RELATION OF DIRECTION TO GRADIENT.

We will now consider the details of the relation of the direction of the wind to gradients and to the lie or trend of the isobar conjointly.

When we talk of gradient only, we get no indication of the direction of the wind, for the barometric slope may face in any direction or have any aspect. Following the analogy of barometric gradients to hill-slopes, we will call the direction towards which gradients slope the aspect of the gradient, so as to keep the word direction for wind. For instance, if isobars run east and west they may slope either north or south, or we might say that the aspect of the gradients was either towards the north or the south, just as we should talk of a hill; or, to take the analogy of geological terms, we might say that the strike of the

isobars was east and west, but that the gradients dipped either north or south.

But by combining the idea of gradient with that of aspect, and both with the Buy Ballot's law, we see at once that if the isobars run or strike east and west, the general direction of the wind will be westerly if the aspect of the gradients is towards the north, but easterly if the aspect is to the south. We therefore say that in the former case we have gradients of such a value for westerly winds, and in the latter gradients of such an amount for easterly winds. This holds for every direction. In Fig. 36 we have, as before explained, a gradient of ten between A and B for velocity, and now we can say that the gradient is also for north-westerly winds; at E there is a gradient of 3.3 for south-westerly winds. By this simple method of expression, whenever we see a synoptic chart, we can calculate at once both the probable direction and force of the wind.

INCLINATION OF WIND TO ISOBARS.

Buy Ballot's law does very well for the general sweep of the wind, but the subject is capable of much greater refinement. The acute angle between the direction of the wind and the lie of the isobar is called the inclination of the wind to that isobar. Taking all kinds of winds and all kinds of isobars, Whipple has found that the inclination amounts to 52° at Kew; while Loomis has deduced an angle of 42° in the United States.

But, by taking the inclination of the wind in different shapes of isobars and different portions of each shape,

Ley, Loomis, Hildebrandson, and others, have arrived at a series of remarkable generalizations as to the general circulation of the atmosphere. They find that the wind is much more inclined and incurved in the right front of a cyclone than in any other portion; and that in the rear the inclination is very small, if not occasionally reversed —that is to say, a little outcurved.

We have examined the details of these cyclone surface-winds, as well as of those in an anticyclone in our chapter on Clouds. There we treated each shape separately, but we can connect both in a very striking manner if we call attention to some general values obtained by Loomis from observations over the Atlantic Ocean. Taking an ideal cyclone, with an adjacent anticyclone, he finds that, starting from the anticyclone, the inclination of the wind to the isobar begins at about 52°, and then gradually decreases to 25° near the centre of a cyclone. Of course this is a generalized case, for we have shown that the inclination is not the same on different sides of a cyclone. The great thing to remember is that in every shape of isobars each part has a wind velocity and direction of its own relative to the gradients.

The only other material source of variation is diurnal. We have already sufficiently explained, in our chapter on the Meteograms, that, whatever the inclination due to any part of any shape of isobars may be, the diurnal variation imposes a modification on that, but does not alter the direction due to general causes. Land and sea breezes we shall discuss in our chapter on Diurnal Variations of Weather.

CALMS.

We have already stated that calms are the product of no barometric gradient. The most persistent calms are found in the "doldrums," or the col of low pressure near the equator between the north-east and south-east trade winds all over the world.

In temperate regions the most persistent calms are near the centres of stationary anticyclones; but more short-lived calms are found in the centres of cyclones, along the crest of wedges, and in cols.

We do not think it necessary to give any special examples of either gales or calms, for they are abundantly illustrated by numerous charts in the course of the work; we need only call attention to Figs. 65 and 66 of south-westerly gales in Great Britain, to Figs. 77 and 78 of easterly gales, and to Figs. 22 and 24 of calms.

WINDS IN THE SOUTHERN HEMISPHERE.

So far we have confined our attention to winds in the northern hemisphere only; now, however, that we thoroughly understand the nature of wind in that hemisphere, we can easily follow the modifications which occur south of the equator.

The great general principles-that every shape of isobars has a distinctive wind; that cyclones incurve, while anticyclones outcurve; that the velocity is mainly determined by the gradient, and also the relation of diurnal to general winds-are the same in both hemi