S.S.W. is short at full moon, and continues longer at other times.

W.S.W. winds may set in at almost any time, but continue longer if they commence a few days previous to the new

moon.

For the West wind I have found no certain rule. W.N.W. and N.W. winds may occur at any time, but seldom follow E. or N.E.

I

I have no doubt many rules might be found in reference to wind from one quarter following another, and it would be desirable to get average results from a still longer period than that from which Mr. Glaisher has taken his averages. Mr. Glaisher does not attempt to account for the prevalence of these winds according with the position of the moon. venture to propose an explanation. I have already said that it is certain that tides exist in the upper portion of the atmosphere, caused by the attraction of the sun and moon, like the tides of the sea. Heat, however, is so powerful a disturbing force in the atmosphere, that currents and storms in the lower portion of the atmosphere often overcome these tidal movements.

Still, these tidal movements remain a constant force, although not a very powerful one; but in the long run it is probable that the winds of the lower part of the atmosphere have a general tendency to follow these tides. Take the instance of S. and N. winds. I have said that if S. wind sets in at full moon, it will only last a short time; but N. wind at same period will last a long time.

I imagine that this shows that the tidal current of the upper atmosphere at full moon is from N. to S. If a wind sets in against this, it lasts but a short time; but if it is in unison with it, it lasts longer.

I suppose I need hardly say that rain is the result of air charged with vapour meeting or mixing with colder currents of air. The most casual observer connects rain with particular winds; and if we are able to predict changes of wind, it is a fair advance towards predicting rain. The rules I have laid down are by no means infallible, but I believe they will be found true in a large proportion of instances.

This is all I have to record in reference to the predicting of immediate changes of the weather; but there have lately been discoveries to enable us to predict some months beforehand the general character of the ensuing weather. For these observations I am mostly indebted to Mr. G. Bramham, and various communications of his to the Meteorological

Society. By the study of registers kept over many years it is observed that extremes of heat and cold are generally preceded by several months of uniform temperature. This sometimes enables observers at or a little after the equinoxes to predict with considerable certainty the general character of the ensuing summer and winter.

Thus I believe in this present year that the very uniform temperature which we had from the middle of January to March 6th was the precursor of the long continuance of dry weather with high temperatures that we have lately experienced.

The rules enunciated by Mr. Bramham are as follows:"When the mean temperatures of the first three months have been so nearly uniform that the range of monthly mean temperature in the first quarter of the year has been only 1.2° or less, the succeeding summer will be characterized by extreme heat.

"When the mean temperature of all the months, from November to March, are above the average, the ensuing summer will also be above the average.

"When the mean temperature of June is below May, or if there is no progressive increase of temperature in June, a cold and rainy July and August may be expected.

"When the mean temperature of December is more than 2° above November, January, February, and March will have a temperature above the average, and January and February will be wet and rainy."

I have spoken of mean temperatures, that being the best form of registering observations for the purpose of these predictions. A uniform mean temperature may be looked upon as identical with settled and calm weather with little air disturbance. These rules have been discovered entirely from a comparison of the registers of passed years, not from the study of the complex physical forces causing the changes of weather. I will, however, attempt some explanation of them.

Immediately after our shortest day, of course, the sun begins to approach our northern hemisphere. If at that time up to about the period of the equinox (the sun being advancing towards us then) there is a uniform mean temperature, which accompanies calm settled weather, then there will also be very little intermingling of air currents, and consequently a great accumulation of heat about the northern boundary of the tropic. This accumulation of heat is afterwards sure to affect the temperature of the ensuing summer. On the same principle, when after our longest day there is

calm, settled weather during July, August, and September, when the sun is receding from us, there will be a great accumulation of cold in the northern hemisphere, which will make the ensuing winter unusually cold.

Our ideas of the extent of the atmosphere have of late years been much enlarged, and there seems no doubt that it extends some forty-five miles above us. It is, therefore, conceivable that vast accumulation of hot or cold air should occur which have this subsequent effect on the temperature.

It may be remarked that these rules are somewhat incomplete. In fact, they amount only to this-that if at certain times of the year weather of a certain character prevails, then we are enabled to predict the character of the weather of the ensuing months.

Still, I think the discoveries already made are a step in the right direction, and I wish to suggest to observers that the careful registering of maximum and minimum temperatures is the best means of getting data to enable them to predict weather.

In the state of our present knowledge of the subject, many years might occur in which the weather might give us no opportunity of making these predictions; but I believe extremely hot summers and extremely cold winters are always preceded by weather which will enable us to predict

those events.

1

ON

HILL FORTRESSES, SLING STONES, AND OTHER ANTIQUITIES IN SOUTH-EASTERN DEVON.

BY PETER ORLANDO HUTCHINSON.

IN the summer of 1861 the Archæological Association of London visited Exeter, and on the 22nd of August in that year I read before them a paper on "The Hill Fortresses, Tumuli, and some other Antiquities of Eastern Devon." Since that time I have had the opportunity of looking up and examining several other objects of interest scattered over this portion of the county, not noticed in my former paper, and it is to these that I wish now to call your attention.

DUMPDON.-The first place to which I will advert is the great camp of Dumpdon. It will do well to begin with, as it lies only two miles and a half northward from the town in which we are assembled. I am not aware that any plan of this camp has been published. In figure and size it very much resembles Hembury Fort, though not quite so long it is a little broader. The form of the hill on which it stands is very like that of Hembury, being a sort of promontory with the point tending to the south. The north end in both is defended by bold earthworks cut right across the ridge of the hill. This is the broadest part of each camp, and from which they gradually contract to a rounded point. About one-third of the pointed end of Dumpdon is planted with beech trees, the space being shut in by a modern hedge run transversely across the area. Near the middle of the camp, namely, at 450 feet from the south point, and 128 from the west agger, is a mound which might be taken for a tumulus, but I understand it was thrown up a few years ago by the officers of the Ordnance Survey, as an object to assist them in the triangulation of the country, similar ones having been erected on several of the neighbouring hills. Across this mound the width of the camp is 361 feet; the whole length of the area is 825 feet; the elevation of the hill is 879 feet above the sea level. The