is found in that peculiar quality that no thermometer can appreciate, but which is readily recognized by our more delicate sensations. In a typical east-going cyclone the neuralgic, pain-producing heat comes with the southeast wind on the right front of the depression; but when a cyclone goes west, the then right front has a north-west wind and the same distressing quality of heat.

FORECASTING DEPENDS ON NO THEORY.

We can now readily understand from all the foregoing remarks that forecasting depends neither on any theory nor on any calculation. The whole science, from beginning to end, rests solely on observation.

The shapes of isobars, and the relation of wind and weather to them, are matters of experience only. We find that certain kinds of weather are associated with different portions of each fundamental form of isobars, and we classify accordingly. We give each shape of isobars a conventional name, but that does not bind us to any theory of atmospheric circulation. In like manner, we see that no averages or mean values are of any avail in forecasting weather. Cyclones may usually take a certain path, but they need not do so; the greater portion of the rainfall of any country may come with a south-west wind, but that does not prevent many fine days with the wind from that quarter. On an average, in England, three days out of four may be cloudy, and the forecaster who always announced a cloudy day would have seventy-five per cent. of success. Still, in an anticyclonic period his calculations would totally fail; he could never

say what kind of cloud would appear, and such a system would have no claim to be called forecasting in the modern sense of the word. It is impossible to suppose that we have yet nearly reached the highest perfection of which forecasting is capable, but still we know enough of the nature of the subject to say with certainty that calculation will never enter much into the science of weather-prevision. Natural aptitude and the experience of many years' study are the qualifications of a successful forecaster. "In fact, meteorology is not an exact, but an observational science, like geology or medicine; and just as, however accurately the symptoms or treatment of any malady may be described, the skill to recognize and the judgment to treat must rest on the ability of the physician, so in meteorology, however carefully the relation of weather to isobars may be defined and the nature of their changes described, the judgment which experience alone can give, to enable a warning to be issued, must ever depend on the professional skill of the forecaster."

DETAIL POSSIBLE.

It may not be out of place to introduce here a few remarks as to the amount of detail which it appears possible to give to daily forecasts. Under various headings, we have already discussed the influence of local obstacles in modifying the appearance or intensity of any kind of weather, and also the powerful diurnal variations of every element in all parts of the world. When to these we add the tendency of cyclones to form secon

daries, so small as not to show in an ordinary synoptic chart, then we may easily understand that it is the general character only of weather which a forecaster can ever safely predict. The general character is the quality of weather which we have taken such pains to show is constant in each portion of every shape of isobars, and that never changes under any local or diurnal variation.

If we live in any place which commands a view over any large tract of country, and we think how often we see both cloud and rain which only affect a very small portion of our horizon, we can readily understand that, even if it were possible to issue minute forecasts, every few square miles of country would require a separate warning.

HOW FAR IN ADVANCE CAN FORECASTS BE ISSUED?

We may also consider how far in advance forecasts can safely be issued. The numerous charts which we have already given will show the reader the amount of change which twelve or twenty-four hours may develop in the distribution of pressure. Sometimes we have been able to trace the changes in either of these intervals quite easily; at other times it has been difficult to say how the first set of isobars has grown into the second. In the United States the observations are taken three times a day, and this appears to be sufficiently frequent for all practical purposes. In most European countries, reports are not sent up more than twice a day; but with this interval, cyclones sometimes form so suddenly that they are not forecast in time to give any warning. We

shall give an example of such a case further on in this chapter.

Thus, from eight to twelve hours seems to be the furthest time for which forecasts can be issued in advance, and even then many local details cannot be given. Some meteorologists are of opinion that a good deal of forecasting will be done in the future, with the assistance of a complete knowledge of recurrent periods of heat, cold, rain, or storm; and we lean strongly to that view, if these periods are used in the manner so fully explained in our chapter on Seasonal and Cyclical Periodicities.

TIME OF PREPARATION.

A few particulars of the time necessary for collecting and examining the materials for synoptic charts will perhaps enable the public better to understand the practical conditions of the problem of weather-forecasting and storm-warnings.

In Great Britain, the morning observations are taken at 8 a.m. Even with all the rapid organization of the British Post-office, the majority of the reports do not arrive till between 9 a.m. and 10 a.m. As fast as they arrive, the information is entered on a chart, and a synoptic chart is constructed. If necessary, telegraphic intelligence of storms is immediately sent to the coasts, and in every case information as to the state of the weather, and a forecast for twenty-four hours ahead, is sent to the press.

In practice, storm-warnings can rarely be despatched

before 11 a.m.; that is to say, three hours after the observations have been taken. If we allow at least another hour before the public can have access to the information, we see at once that the day is so far gone that the forecast can have little practical importance for the majority.

The greatest value is when a storm has just begun to show over Valentia at 8 a.m.; then the English coasts can be warned in time. Still, in the three or four hours which must elapse before the storm can be warned, the cyclone will have advanced, perhaps, as much as a hundred and twenty miles, so that, before a telegram can reach the western shores of England, the gale will either have commenced, or the appearance of the sky will have given unmistakable warning.

The whole theory of storm-warnings by means of the electric telegraph is based on the supposition that the message travels faster along the wire than the storm along the earth's surface. But, as the practical organization of collection and distribution of intelligence takes at least three hours, the storm must either move slowly or over a considerable intervening district before any set of stations can be successfully warned. The forecasts which are issued from reports taken at 6 p.m. are of

more use.

The organization of the press enables the public to obtain the office forecast much more quickly than by any other means. The British reports are taken at 6 p.m., while the United States Signal Office obtain their latest about 11 p.m. These five hours are an unquestionable gain. In Great Britain there are, however, difficulties