THE RAINFALL OF THE BRITISH ISLES M.DE CARLE S. SALTER THE RAINFALL OF THE BRITISH ISLES SOME RECENT PUBLICATIONS GEOGRAPHY AND WORLD POWER By JAMES FAIRGRIEVE, M.A., F.R.G.S., Lecturer at the London Day Training College. New and Enlarged Edition. $s. net. A lucid and fascinating account of the influence which geographical facts have exerted upon the develop- ment of mankind. LAND FORM MAP BOOK By JAMES FAIRGRIEVE, M.A., F.R.G.S., Lecturer at the London Day Training College. 2s. net. By working through this book pupils are trained in the use of contour lines to represent shapes. By mapping first simple forms and then more and more complete ones, they are led by a graded series of exercises to make maps from photographs. THE NEW REGIONAL GEOGRAPHIES FOR SECONDARY AND HIGH SCHOOLS By LEONARD BROOKS, M.A., F.R.G.S. Inspector of Schools under the London County Council ; Formerly Geography Master at William Ellis School, and Lecturer in Historical Geography, Birkbeck College. Book I. THE AMERICAS, v. II. ASIA AND AUSTRALASIA. 4*. ,, 111. AFRICA AND EUROPE. 6s . 6<*. Parti. AFRICA. With Maps and Plans, zs. 6d. Part 2. THE BRITISH IbLES. aj. 6rf. Part 3 . EUROPE, including the BRITISH ISLES. 5 s. Book IV. THE WORLD. 7 s. 6rf. With new Maps and Plans. LONDON: UNIVERSITY OF LONDON PRESS, LTD. 17, WARWICK SQUARE, E.C-4 THE LATE GEORGE JAMES SYMONS, F.R.S. FOUNDER OF THE BRITISH RAINFALL ORGANIZATION. [Frontispiece THE RAINFALL OF THE BRITISH ISLES M. de CARLE S. SALTER SUPERINTENDENT OF THE BRITISH RAINFALL ORGANIZATION (METEOROLOGICAL OFFICE). JOINT-EDITOR OF THE METEOROLOGICAL MAGAZINE. LATE VICE-PRESIDENT OF THE ROYAL METEOROLOGICAL SOCIETY. ASSOCIATE OF THE INSTITUTION OF WATER ENGINEERS "... like the phenomena of nature^ like the sun and the sea, the stars and the flowers, like frost and snow, rain and dew, hailstorm and thunder, which are to be studied with entire submission of our own faculties, and in the perfect faith that in them there can be no too much or too little, nothing useless or inert, but that, the farther we press in our discoveries, the more we shall see proofs of design and self-supporting arrange- ment where the careless eye had seen nothing but accident ! " DE QUJNCEV, '' On the Knocking at the Qate in Macbeth." LONDON UNIVERSITY OF LONDON PRESS, LTD. 17 WARWICK SQUARE, E.C-4 1921 PREFACE I HAVE frequently been struck by the need for a book summarizing the results of many years of labour in compiling and studying statistics of the rainfall of the British Isles. A very large number of papers and articles on various aspects of the subject have been published from time to time, and it has been dealt with in a more or less cur- sory manner in many books on meteorology and physiography; but on being asked by engineers, teachers, and students for reference to any volume giving general information on the subject, I have constantly found myself at a loss. In endeavouring to meet this need I am aware that I am drawing freely on the work of others, more especially of my predecessors in the direction of the British Rainfall Organization. I have tried in every instance to acknowledge this, but after twenty-four years of active participation in the work of the Organization, dating back to the lifetime of its 'founder and covering the administration of his successors, Mr. Sowerby Wallis and Dr. H. R. Mill, it is difficult always to know exactly how lines of research originated. I am in an especial degree indebted to Dr. H. R. Mill for constant encouragement and teaching during the nineteen years when I had the privilege of working under him. I have to acknowledge the permission of the 4982-16 vi PREFACE Institution of Water Engineers to reprint portions of my recent paper on "The Relation of Rainfall to Configuration," and to thank the Royal Meteoro- logical Society for a similar courtesy in respect of my paper on " The Measurement of Rainfall Duration." Acknowledgments are also due to the Director of the Meteorological Office, the Royal Meteorological Society, Messrs. Casella & Co., Messrs. Negretti & Zambra, and Mr. J. Baxendell for the loan of blocks and drawings. The remainder of the illus- trations have been prepared by Mr. A. T. Bench. Throughout this volume I have used the English inch as a unit of measurement of rainfall in prefer- ence to the millimetre, believing it to be more familiar to the majority of readers. I have endeav- oured to avoid the use of unnecessary technical terms, and have assumed only an elementary acquaintance with physics and meteorology. CARLE SALTER. 62, CAMDEN SQUARE, LONDON, N.W.I. July 1921. CONTENTS PREFACE . y* 9 * . -+ pp. v-vi CHAPTER I Introductory Methods of study and sources of information PP. i-5 CHAPTER II THE PHYSICAL PROCESSES OF RAIN FORMATION Laws governing water-vapour in the atmosphere Condensa- tion Dew and hoar-frost Nuclei Mist and fog Cloud Snow and sleet Impurities in rain Atmospheric cooling Conduction Admixture Thermo-dynamic cooling The temperature-lapse Ascending air-currents Convectional rain Hail April showers Cyclonic rain The mechanism of atmospheric depressions Line squalls Converging currents Orographical rainfall pp. 6-28 CHAPTER III RAIN GAUGES Principles of the rain gauge Evolution of the standard pattern The Snowdon funnel Capacity of gauge Thermal insulation The measuring glass The " Camden " measure Bradford and Seath- waite gauges Rain gauge experiments Obselete patterns of rain gauge and their defects ...... pp. 7~53 CHAPTER IV THE EXPOSURE OF RAIN GAUGES Experimental observations Effect of elevating gauges Con- ditions of exposure Protectional devices The Nipher shield The Wild fence Natural shelter Over-shelter . . pp. viii CONTENTS CHAPTER V MECHANICAL RAIN GAUGES Function of the mechanical gauge Dial gauges The tipping- bucket gauge The counterpoised bucket gauge The Casella Recorder Halliwell's Patent The Hyetograph The Fernley Simple syphon ^gauges The Rainfall -rate recorder pp. 67-84 CHAPTER VI THE MEASUREMENT OF RAINFALL DURATION Early records Methods of observation Sources of error The personal and instrumental equation Standardization * pp. 85-93 CHAPTER VII THE MAPPING OF RAINFALL DATA Advantages of the cartographical method Detection of obser- vational errors Construction of a rainfall map Various types of rainfall map Determination of general values Exterpolation of isohyetal lines . . \ . . . , pp. 94-104 CHAPTER VIII THE INCIDENCE OF DAILY RAINFALL Diurnal range of amount and duration of rainfall Intensity of rainfall Heavy falls in short periods . . pp. 105-117 CHAPTER IX THE FREQUENCY OF DAILY RAINFALL Definition of the " rain-day " Frequency of rain-days and its variations Droughts Rain-spells Their frequency and distri- bution . . j -. > . .... pp. 118-129 CONTENTS ix CHAPTER X TYPES OF REGIONAL DISTRIBUTION 1 Identification of showers Isochronous rainfall maps Distribu- tion of rainfall in thunderstorms Distribution associated with primary cyclones Relation to the tracks of depressions pp. 130-162 CHAPTER XI TYPES OF REGIONAL DISTRIBUTION II Exceptional cyclonic rainfalls Stationary depressions Deflected depressions Rainfall with straight isobars Distribution of oro- graphical rainfall ...... pp. 163-187 CHAPTER XII THE SEASONAL VARIATION OF RAINFALL The rainfall of calendar months The thirty-five years' average Types of seasonal march Isomeric rainfall maps and their appli- cation Regional distribution of average monthly rainfall Driest and wettest months Deviations from the average Varia- tions of general monthly rainfall Rainfall of winter and summer pp. 188-211 CHAPTER XIII THE FLUCTUATIONS OF ANNUAL RAINFALL The civil year as a unit of time Characteristics of annual distri- bution of rainfall Variations of annual rainfall, 1860 to 1919 Long period fluctuations The mass curve Choice of period for determining annual normals Fluctuations of general rainfall Cycles Regional distribution of variation from normal pp. 212-231 CONTENTS CHAPTER XIV THE RELATION OF RAINFALL TO CONFIGURATION Frequency and seasonal incidence of orographical rainfall The easterly type Approximation of annual rainfall to the oro- graphical type The relation of average annual rainfall to configura- tion Rainfall at or near sea-level The increase of rainfall on windward slopes The effect of shelter The decrease of rainfall on leeward slopes Rain-shadow Rainfall in narrow valleys pp. 232-268 CHAPTER XV THE ECONOMIC APPLICATION OF RAINFALL DATA Volumetric determination of amount of rainfall The construc- tion of a rainfall map for water-supply purposes Evaluation of compensation water Evaporation and percolation Effective rain- fall Relation of rainfali to flow of streams Underground water Need for & hydrometric survey , . . pp. 269-286 GLOSSARY pp. 287-290 INDEX pp. 291-295 LIST OF ILLUSTRATIONS GEORGE JAMES SYMONS, F.R.S. . , * . Frontispiece PAGE STRUCTURE OF HAIL STONE . . . . . 21 LINES OF FLOW IN A MOVING CYCLONE . . . 24 RAIN GAUGE USED IN 1695 . . . . . .. 3 PRINCIPLE OF THE RAIN GAUGE . - . . . . 31 SHALLOW RAIN GAUGE FUNNEL . . . , . 33 GLAISHER FUNNEL . . ... . . . . 34 SNOWDON FUNNEL . ... . . . . -34 BRASS RIM FOR RAIN GAUGE , ; . . 35 METEOROLOGICAL OFFICE GAUGE . . . .36 SNOWDON GAUGE . . . . . . . 37 INSULATED SNOWDON GAUGE . ... . . -37 FLAT-BOTTOMED MEASURING GLASSES . . . .40 " CAMDEN " MEASURING GLASS ..... 40 MEASURING GLASS NORWEGIAN PATTERN .... 40 MENISCUS AND FALSE MENISCUS ..... 41 BRADFORD GAUGE . . ' .. "' . . . . . 42 SEATHWAITE GAUGE ....... 44 BRITISH ASSOCIATION GAUGE . '. . . -47 HOWARD GAUGE . . . . . . . .48 Box GAUGE . '..'.' ^ * . 48 GLAISHER GAUGE . . ..,, 5 FLEMING GAUGE . . . . . . 5 SIDE-TUBE GAUGE ...... . 52 xi xii LIST OF ILLUSTRATIONS PAGE TAP GAUGE . . . . . . . . . 52 RECTANGULAR GAUGE . . . . .' . 53 FUNNEL OF JAGGA RAO GAUGE . . . . -53 EXPERIMENTAL GAUGES . . . . . . 57, 60 NIPHER SHIELD NORWEGIAN PATTERN . ... 63 BILLWILLER'S DESIGN FOR NIPHER SHIELD ... 63 GAUGE SUNK IN PIT ....... 63 SECTION OF WILD'S FENCE , . , . . . .64 DIAL RAIN GAUGE. ....... 68 TIPPING-BUCKET RECORDING GAUGE .... 69 CASELLA STANDARD RECORDING GAUGE .... 70 BECKLEY RECORDING GAUGE ...... 71 CASELLA IMPROVED RAINFALL RECORDER .... 73 HALLIWELL RECORDING GAUGE ..... 75 HYETOGRAPH . . . . . . . 76, 77 FERNLEY RECORDING GAUGE . . . . . 80, 8 1 CASELLA'S SIMPLE SYPHON GAUGE ..... 82 NEGRETTI & ZAMBRA'S RAINFALL RATE-RECORDER . . 83 MEASUREMENT OF RAINFALL DURATION .... 87 CONSTRUCTION OF A RAINFALL MAP ... 98, 99 CLASSIFICATION OF INTENSE RAINFALLS . . . 112, 113 DISTRIBUTION OF RAIN DAYS . . . . . .123 DISTRIBUTION OF THE SPRING DROUGHT OF 1893 . . 127 SNOWSTORM OF DECEMBER 25-26, 1906, ISOCHRONOUS LINES 133 ISOCHRONOUS RAINFALL MAPS, AUGUST 26-27, J 9 12 J 34? *35 RAIN-FIELDS OF MAY 31, 1911 . . . 137, 138, 139 CLOUD AND RAIN-FIELDS, MAY 31, 1911 .... 140 RAIN-FIELDS, JUNE 14, 1914 142, 143 RAINFALL DISTRIBUTION : THUNDERSTORM TYPE . 145, 146 RAINFALL DISTRIBUTION : MODIFIED THUNDERSTORM TYPE . 147 RAINFALL DISTRIBUTION : CONVECTIONAL-CYCLONIC TYPE . 149 PARALLEL TRACTS OF THUNDERSTORM RAIN . . .150 LIST OF ILLUSTRATIONS xiii PAGE ALIGNMENT OF RAIN-SPLASHES IN THUNDERSTORM . . 153 ALTERNATING DRY AND WET AREAS IN THUNDERSTORM . 153 SECONDARY CYCLONIC RAIN SIMULATING THUNDERSTORM DISTRIBUTION . / *54 RAINFALL DISTRIBUTION WITH PRIMARY CYCLONE . 155, 157 RAINFALL WITH ABNORMAL CYCLONE TRACK . . .158 ABNORMAL CYCLONIC RAINFALL DISTRIBUTION . . . 158 PRECIPITATION WITH CURVED CYCLONE TRACK . . .159 RAINFALL DISTRIBUTION WITH LOOPED CYCLONE TRACKS 160, 162 RAINFALL DISTRIBUTION WITH NEARLY STATIONARY DE- PRESSIONS ....... 165, 166 RAINFALL DISTRIBUTION WITH DEFLECTED CYCLONE TRACKS .... 169, 172, 173, 175, 177 PRESSURE CONDITIONS FOR OROGRAPHICAL RAINFALL . .181 OROGRAPHICAL RAINFALL PRODUCED BY PASSAGE OF DEPRESSION 1 82 OROGRAPHICAL RAINFALL DISTRIBUTION . . . 184, 1 86 CURVES OF AVERAGE MONTHLY RAINFALL. . 191, 194, 195 ISOMERIC RAINFALL MAPS . . . 197, 198, 199, 200 DISTRIBUTION OF AVERAGE MONTHLY RAINFALL 203, 204, 205, 206 FLUCTUATIONS. OF ANNUAL RAINFALL . . 216, 219, 220 FLUCTUATIONS OF ANNUAL RAINFALL IN IRELAND . . 226 TYPES OF MONTHLY RAINFALL DISTRIBUTION 235, 236, 237, 239 MAPS OF ANNUAL RAINFALL DISTRIBUTION . 241, 242, 244 DISTRIBUTION OF AVERAGE ANNUAL RAINFALL . . . 246 RELATION OF RAINFALL TO CONFIGURATION CROSS-SECTIONS 261 AVERAGE ANNUAL RAINFALL IN UPPER FORTH VALLEY . 277 RAINFALL AND WELL-DEPTH AT CHILGROVE . . .285 CHAPTER I INTRODUCTORY THE study of the distribution of rainfall has commonly been regarded as a branch of Meteorology, or more particularly of Climatology, and most of the attention which has been given to the subject has therefore come from those who have made these sciences their special concern. In certain respects the classification is undoubtedly justified : precipi- tation of aqueous vapour from the atmosphere, in all its forms, is one of the central phenomena round which much of the science of meteorology revolves ; and the seasonal and regional distributions of this all-important climatic element are among the main considerations of Climatology. It is of course impossible rigidly to delimit the spheres of Climatology and Geography, which have much that is common ground ; but it should be noted that it is very largely by geographical methods that the most important additions to our knowledge of the natural control of rainfall distribution have been made. It is necessary to point out, however, that any empirical method of studying natural phenomena, however complete and systematic, is at best capable of yielding only a one-sided view of the subject as a whole, so that, in the present instance, the aid of experimental physics is an essential adjunct to direct 2 RAINFALL OF THE BRITISH ISLES observations in attempting to arrive at an under- standing of the natural laws which they illustrate. The problem of atmospheric precipitation as it has presented itself to the student of pure meteor- ology, by itself as limited as that approached solely from the geographical side, is one of exceptional complexity. The atmosphere, the theatre of the operations which he is examining, is an enormous and almost unconfined mass of elastic fluid in a state of perpetual movement along complicated paths and in three dimensions. It is constantly varying in respect of its temperature, pressure, transparency, humidity, electrical condition, and probably also in other ways of which we are not even cognizant. These variables are intimately correlated by laws not always fully understood, in some cases not even known, and nearly all have an important bearing on the condensation of moisture from a gaseous to a liquid form which constitutes the phenomenon of precipitation. Owing to our position at the bottom of the ocean of air, and to the vast extent of that ocean, direct observation of the variations in the conditions enumerated is limited in its scope. Some of the factors can be more or less accurately measured in the portions of the atmosphere accessible by human or mechanical means ; others can be observed only by rough-and- ready methods. As an example, the vital question of obtaining records of atmospheric humidity, by which its variations in space and time may be deter- mined, has never been completely mastered ; whilst the variations of temperature with which it is closely bound up are so great and so rapid that they can be followed only in the broadest outline. In apply- ing mathematical reasoning to problems with so INTRODUCTORY 3 many unknown, or only partially known, factors, it is necessary to introduce large assumptions and sometimes to ignore important side-issues in order to arrive at the required simplicity. In spite of these difficulties, very considerable additions to our knowledge of the dynamics of the free atmosphere have resulted from research along geophysical lines, and the student who desires to approach the subject of rainfall distribution from the standpoint of the geographer will do well to acquaint himself as a preliminary with the work of the modern schools of meteorology. Among the more eminent of British meteorologists to whose writings attention may be drawn in this connection should be mentioned the late Dr. John Aitken, Mr. W. H. Dines, and Sir Napier Shaw, whose studies of the thermo- dynamics of the atmo- sphere have played a leading part in clearing away much of what was fundamentally wrong in the hypotheses of former schools of meteorology. Among foreign workers along similar lines perhaps the first place should be given to Professor V. F. Bjerknes, of Christiania, whose original methods of research into the complex dynamical reactions of the free atmosphere have been extremely illuminating. In the more direct field of geographical research the student of rainfall problems, so far as these apply to the British Isles, is deeply indebted to the late Dr. Alexander Buchan, and in a still higher degree to the organizing ability and lifelong industry of the late George James Symons, who laid the foundation for the systematic observation of the rainfall in all parts of the country on a scale un- paralleled in any part of the world, and in so doing 4 RAINFALL OF THE BRITISH ISLES made possible the cartographical studies of Dr. H. R. Mill. It is from the very numerous maps of rainfall distribution constructed by Dr. Mill during the first twenty years of this century, and from his clear scientific deductions therefrom, that practically all our existing knowledge of the rainfall of the British Isles from the purely geographical side must be drawn. Work on similar lines in the north of Europe has been carried out with conspicuous success by Dr. G. Hellmann, of the German State Meteorological Service. In referring to the co-ordinating work of the leaders of meteorological science it wduld not be just to omit reference to the large and devoted band of helpers in all parts of this country who, by the patient accumulation of data, achieved often only at considerable personal sacrifice, have been and still are constantly providing the raw material so essential to the progress of any observational research. Among the five thousand rainfall observers in the British Isles at the present day, there are, of course, a consider- able number of paid officials, but by far the greater number are voluntary and self-equipped. Their individual interests are exceedingly diverse, and their willingness to work together on a uniform system, suggested from without, for the good of a common cause, implies not merely successful organization, but also a mutual bond of disinterested attachment to the science which they all unite to serve. The great mass of statistical information bearing on the rainfall of the British Isles which has been brought together by the British Rainfall Organiza- tion, and is still being added to year by year, offers a rich field of material to any who care to delve in it. A good deal has, no doubt, been done, especially in INTRODUCTORY recent years, but that far more awaits the doing is certain. What lines future research may take it is impossible to say, and whether any purely mathe- matical treatment will yield generalizations of value remains still to be proved. The work so far attempted may be briefly summed up as a partial study of the subject from two broad points of view, the variations of amount of rainfall in Time and its variations in Space. In regard to the actual weather experienced day by day the variability in both these respects is very great, and without close study the fall of rain appears to be among the most capricious and unregulated of all natural phenomena. Meteor- ology has shown, however, that, in spite of its seeming waywardness, rainfall is controlled, like everything else in nature, by fixed and comprehensible laws ; and that as we widen our purview the limits of variability in time and in space become narrower and narrower. Whether the period covered by trustworthy obser- vations has as yet been long enough to enable us to fix the limits of variability with certainty is doubtful, but it appears to have been sufficient for us to obtain some insight into the operation of the main controlling factors. The object of this book is to bring together some of the general conclusions so far deduced ; to suggest, rather than formulate, a working hypothesis which future students of the subject may build upon or amend, and to draw attention to the economic utility of a more complete knowledge of our resources in respect of one of Nature's greatest gifts. CHAPTER II THE PHYSICAL PROCESSES OF RAIN FORMATION IT is principally in connexion with the physical and dynamical processes operating in the atmosphere that the student of the geographical aspects of rain- fall must look to the physicist and mathematician for collaboration in order to gain that second view- point so essential in obtaining a clear conception of the natural laws underlying the phenomena which his observations illustrate. It is difficult, in some cases impossible, to repro- duce in a laboratory the precise conditions of nature, the more so when these conditions occur in a theatre so vast as the free atmosphere, and are complicated by factors which are in many cases unknown and inaccessible to direct observation. On this account there is still much to be learnt of the physical reactions which take part in the production of observed weather phenomena, but by a process of repeated simplification and constant readjustment of hypothetical reasoning to fit with observed facts certain invariable laws have been formulated which there is every reason to regard as representing the fundamental basis of rainfall formation. It is from work along these lines that the following brief sum- mary of the physical processes of rainfall has been drawn. All the varied phenomena of aqueous precipitation PHYSICAL PROCESSES from the atmosphere, of which rainfall is the most important, depend upon the fact that water, under certain conditions of temperature and pressure, is capable of existing in the form of vapour. Water- vapour obeys the same physical laws as any other gas, and if added to a fixed volume of dry air operates in increasing the pressure exerted by that air, i.e. its weight, in exactly the same way as if an equal quantity of any other gas were introduced. The limit to the capacity of water to remain in the form of vapour is determined entirely by temperature, and therefore the amount of water-vapour which can exist in a definite volume of air varies in accordance with the temperature of the air. The existence of air is not, of course, theoretically essential for the existence of water-vapour, but in natural conditions air is always present, and the above statement holds good. Air which is more or less dry will readily take up moisture from any available source by evaporation, and, provided sufficient water is available, evapora- tion will continue until a certain vapour-pressure is attained (varying with the temperature), after which no further evaporation will take place unless the temperature rises. Temperature in Percentage of normal l degrees Fahr. . air-pressure. 32 0-6 50 1-2 78 3-2 96 5-9 114 8-3 211 100 1 These values are adapted from Sir Napier Shaw's Forecasting Weather, p. 153, 8 RAINFALL OF THE BRITISH ISLES The vapour-pressure attainable at certain specific temperatures may be expressed in percentages of normal atmospheric pressure, taken for convenience as 1,000 millibars, equivalent to 750 millimetres, or 29*5 inches of mercury on the ordinary barometer scale. The slight difference between the last temperature given (211) and the boiling- point of water (212) arises from the pressure being taken for the purpose of the calculation at 750 mm. instead of 760 mm. or 29-92 inches, which is about the mean value for sea-level. Air which contains the largest possible quantity of water-vapour for its temperature is said to be satu- rated, and the temperature at which any body of air becomes saturated is known as the dew-point. Any additional quantity of water-vapour introduced could not remain in the gaseous form and, except under conditions which will be mentioned later, must necessarily condense. It follows further from this law that if air containing water- vapour be cooled to the temperature of its dew-point, it will ipso facto become saturated. Further cooling will bring about condensation of part of the water- vapour until equilibrium is re-established, partly by the lowering of the dew-point occasioned by the diminished quantity of water- vapour remaining, and partly by the release of heat in the process of condensation. Such condensation by cooling is the only means known in nature by which water is withdrawn from the atmosphere, and atmospheric cooling is therefore essential for all precipitation. Before considering these temperature changes in detail it is convenient to draw attention to the conditions under which condensation of water- vapour takes place in the atmosphere. In order that PHYSICAL PROCESSES this may occur it has been shown to be necessary that some nucleus should exist upon which the condensed water-drop may form. If such is not present, cooling may proceed beyond the dew-point without condensation, the air passing beyond saturation to the abnormal condition known as super-saturation. Aitken has demonstrated this clearly under artificial conditions, but it is highly probable that such a phenomenon never occurs in nature. In the case of atmospheric cooling brought about by conduction of heat from the air to any solid body such as the earth, a building, or plant, this body itself forms the required nucleus, and when cooling has proceeded to the stage of condensation, dew or hoar-frost is deposited. In the free air the nucleus is readily provided by dust-particles and other minutely subdivided matter held in suspension. Aitken's investigations of the dust-content of the air have shown that at all elevations within the reach of water- vapour sufficient dust exists to provide the necessary nuclei, 1 and the conditions of condensation in the absence of dust, although of great scientific interest, are not immediately applicable. It is, however, noteworthy that even in the complete absence of solid nuclei condensation will take place, after a high degree of super-saturation has been attained, on the electrified ions produced by the dissociation of atoms, as has been proved by Mr. C. T. R. Wilson. The first stages of normal condensation in the free atmosphere may be pictured as the coating with moisture of the floating dust-particles. These particles themselves, at any rate in the upper atmo- 1 Proc. R. Soc. Edin.j xvii, 1891, pp. 193-254. io RAINFALL OF THE BRITISH ISLES sphere, are extremely minute, 1 but in spite of this an optical effect readily perceptible to the naked eye is produced. When dry, dust-particles reflect the light which falls on them, and do not, except by diffusion of light, interfere with the transparency of the air ; when damp, they absorb a large pro- portion of this light, and if condensation brings this about, the air immediately becomes opaque. An example of this phenomenon is provided when steam is ejected into the air. On first issuing from the funnel of an engine steam is invisible, but directly it mixes with the cooler air. condensation takes place and a cloud is formed. It is a common error to speak of seeing a cloud of steam ; as a matter of fact, the cloud is composed of liquid water deposited on the dust in the air, and when after a short exposure the droplets re- evaporate and return to the form of water-vapour, they again become invisible. When natural condensation in the air occurs in contact with the earth, the phenomenon is commonly referred to as a mist, when in the upper air as a cloud. The two are similar in all respects, except that there may be a difference in the cause of the cooling which has given rise to the change from water-vapour to liquid form. True fog differs from mist in some respects which are not of importance in this con- nexion. It is important at this point to observe that dust- particles, as has been pointed out by Aitken, have to a certain extent hygroscopic properties, so that when present in sufficient numbers they will induce 1 As an example of the minuteness of dust-particles, Aitken computed that a single puff of cigarette smoke consisted of about 4,000,000,000 particles. PHYSICAL PROCESSES n condensation even though the air is not completely saturated with moisture. This explains the persist- ence under certain conditions of haze or fog in comparatively dry air. Rain is, however, never formed in this way. In the upper atmosphere the conditions under which clouds are formed are numerous, and the cloud may appear in a great variety of forms, some of which are extremely beautiful. It is foreign to the subject now under discussion to consider these varieties of cloud in detail, but it is of interest to notice the three important types classified by Luke Howard the Stratus, or sheet- cloud, analogous to a high fog, the Cumulus, or heap- cloud, and the Cirrus, or hair-cloud. The two former are com- posed of water-droplets formed in the manner just described, and the Cirrus of ice-particles. The Cirrus is the most lofty of all clouds and is known to consist of ice on account of the peculiar optical effects which it produces. When condensation, after reaching the cloud stage, is continued, the tiny water-particles of which the cloud is built up begin to coalesce, and as they become heavy enough to overcome the friction of the air, fall downwards by gravity. In all probability many of these drops are re- evaporated during their passage earthward, but those which escape are further augmented by collision with others and grow large enough to reach the earth as rain-drops. There is a limit to the size of rain- drops, since, owing to their velocity of descent, drops exceeding certain dimen- sions must be broken up again before reaching the ground. It is not fully understood under exactly what conditions condensation in the atmosphere has the 12 RAINFALL OF THE BRITISH ISLES effect of forming ice-crystals instead of liquid droplets. It is highly probable that the latter are frequently formed, even though the temperature is below the freezing-point of water under ordinary circumstances. It is, however, not improbable that a good deal of our ordinary rain is in the form of ice in its earliest stages and melts during its descent. If this melting is incomplete, the phenomenon of sleet is observed ; and if melting does not take place, the ice-crystals will reach the ground in the form of snow. In very cold climates individual ice-crystals or very small conglomerations fall, but in more temperate conditions masses of entangled and half- melted crystals form snow-flakes. Apart from the difference of temperature in the strata at which condensation occurs, there is not known to be any physical distinction between the conditions under which rain and snow are formed, so that for purposes of study the two may be regarded as merely different manifestations of the same phenomenon. The peculiar conditions giving rise to the formation of hail stones will be more conveniently dealt with later. The conception of the fact that precipitation takes place in consequence of condensation on solid nuclei floating in the free air gives rise to speculation as to the origin of the occasional phenomena known as " black rain," " blood-rain," " milk-rain," etc., viz. rain bringing with it an admixture of foreign sus- pended matter, sufficient to cause discoloration. It appears probable, however, that the presence of such foreign matter may be due less to its having formed the actual nuclei for condensation than to the accident of its presence in the air lying between the condensation stratum and the earth. Such PHYSICAL PROCESSES 13 impurities may consist of soot, plant-pollen, finely divided sulphur from manufacturing processes, sand, or numerous similar substances. The remarkable so-called " blood-rain " which occurred in the south of England and in Germany in February 1903 was shown to have been impregnated with mineral substances carried by air- drifts from the Saharan Desert. 1 The fine dust discharged in volcanic eruptions has been known to remain in suspension in the air for several years. Under normal conditions rain is also found to con- tain a certain amount of salt and other soluble impurities. Observations by Angus Smith 2 and others show that the quantity present is much greater near the sea than inland, justifying the assumption that evaporated sea-spray is the source, but the products of manufactories also contribute a portion. The salt-content of the atmosphere is much greater in winter than in summer. Observa- tions covering ten years at Rothamsted, Herts., gave an average deposit of about 26 Ib. per acre per year. 3 Instances are occasionally met with in which much more remarkable " impurities " are deposited in showers of rains. Among these may be mentioned recorded instances of showers of thousands of tiny fishes, of immature frogs and other small animals. There is no doubt that such phenomena are due to strong ascending air-currents, such as local whiilwinds, carrying these light objects away from the ground and transporting them through the 1 See Q.J.R. Met. Soc., vol. xxx, 1904, p. 57. 2 See R. Angus Smith, Air and Rain, 1872. 3 See Journal of R. Agric. Soc. of England, October 1883 ; and Journal of Agric. Sci., London, October 1919, vol. ix, pp. 309-37. 14 RAINFALL OF THE BRITISH ISLES atmosphere until the force of gravity once more brings them to the earth. 1 Apart from such abnormal cases there is no doubt that rain performs an important function in cleansing the air of impurities and bringing them to the ground both in suspended and dissolved form. The pre- cipitation of ammonia and other nitrogenous matter with rain has an important bearing on its capacity to fertilize the soil. Valuable researches on the quantity and nature of atmospheiic pollution are being carried out by the Committee on Atmospheric Pollution, under the superintendence of Dr. J. S. Owens for the Meteorological Office. 2 When we look for the natural conditions under which the temperature of large masses of free air may be modified in such a manner as to give rise to changes in its vapour-content, these are found to be of several kinds. The variations in the amount of heat received from the sun might at first sight appear to be a potent factor in the situation, but we know that the trans- mission of solar heat to or from the atmosphere is for the most part indirect. The greater part of the sun's rays which reach the earth pass through the air 'without raising its temperature appreciably, and react on the land and sea. Part of the heat thus received is again lost by radiation through the air, but part is transmitted to the lower strata of the atmosphere by conduction. Similarly, when the 1 An interesting account of a large number of phenomena of this nature, brought together by W. L. McAtee, of the U.S. Bureau of Biological Survey, is published in the Monthly Weather Review of the U.S.A. Department of Agriculture, vol. xlv (No. 5), May 1917, p. 217. 2 An account of the work of the Committee on Atmospheric Pollution is given in a lecture by Dr. Owens, published in the Q.J.R. Met. Soc., vol. xliv, 1918, p. 149. PHYSICAL PROCESSES 15 surface of the earth is cooled below that of the air by radiation of heat into space, the layers of air in immediate contact are cooled by conduction. Al- though the atmosphere as a whole contains an almost inconceivably vast quantity of water in the form of vapour, the amount present in the shallow surface layer cooled by conduction is relatively insignificant. In addition to this, the conditions favourable for cooling by this process are those of relative calm, so that when slight condensation has reduced the available moisture supply, it is not re- newed by the advent of fresh moist air. Condensation of this kind commonly takes the form of dew or hoar-frost, or, under certain con- ditions, of mist or fog, and the total amount of water accruing to streams from this source, in temperate climates at any rate, is so small as to be negligible in comparison with true rainfall. Other processes of cooling have been adduced to account for condensation of rain, among the most important of which is the supposed mixing of cold air with warm moist air. It can be shown that, although neither mass of air is in itself in a condition to condense its water- vapour, the mixture may have a temperature below the combined dew-point. As a matter of fact, air currents of different tempera- tures do not mix readily in nature, and if they do by any chance do so, as, for example, when churned together by eddy-motion, the amount of water released is quite small and seldom forms more than a cloud or fog-bank. It is probable, though not certain, that the admixture of air at different temperatures may take place at the planes of contact of superimposed strata, probably arising from eddy- motion induced by friction. In these circum- 16 RAINFALL OF THE BRITISH ISLES stances incipient condensation, giving rise to more or less permanent cloud-layers, but not sufficiently pronounced to cause actual rain, may occur. The subject awaits fuller investigation. Besides the temperature variations brought about in the atmosphere by insolation and conduction, which, as has been seen, are riot sufficiently pro- nounced to account for the phenomenon of rain, an extremely important effect is produced by thermo- dynamic reaction. This process, which in nature probably seldom occurs entirely free from compli- cation arising from the factors previously mentioned, is most easily studied if considered apart from them. It is an accepted physical law that heat and work are different manifestations of energy, and in accord- ance with the law of the conservation of energy they are mutually convertible. If a thermally insulated body of air is compressed by any agency, some of the work expended in compressing it is converted into heat and the air is warmed. Inversely, if air under similar circumstances is released from pressure, being a gas it expands and is cooled. Changes of temperature arising in this way are known as adiabatic, since they occur without transference of heat from any adjacent body or mass of air. The importance of adiabatic temperature changes in meteorology arises from the well-known fact that the density of the air varies in accordance with the pressure of the superincumbent strata, so that samples taken at different elevations above the earth's surface would show a decreasing density with increas- ing height. During recent years the attention of meteorolo- gists has been turned very largely to the study of the upper layers of the atmosphere, and a great mass PHYSICAL PROCESSES of observational data is now available bearing on the vertical temperature gradient, or " lapse-rate," under various conditions. Mr. W. H. Dines has calculated the mean temperature at various altitudes over England for each month of the year. 1 MEAN MONTHLY TEMPERATURE AT VARIOUS ALTITUDES FOR ENGLAND IN DEGREES FAHRENHEIT oi j3 a Sf 1 1 i ^ > i 1 1 o | 1 .a IB ag i > i 1 i ) a H-> 3 June 28, 1917. In some tropical districts daily rainfalls of from 40 to 50 inches have occurred. In the drier parts of the country, with an annual rainfall of less than 40 inches, a gauge to be read once a month would be immune from risk of overflow if capable of containing 15 inches, but in mountain areas at least 25 inches should be provided, and in a few excep- tional places, such as Snowdonia, theEnglish Lakes, and part of the West Highlands from 50 to 60 inches occa- sionally falls in a very wet winter month. In these cases special patterns of gauge are necessary, since the provision of so unusual a capacity where not required renders the instrument unwieldy and expensive. The thermal insula- tion of the gauge, in order to ensure the greatest possible immunity from evaporation and frost, is best secured by sinking the lower part in the ground. In very exposed positions an outer covering of felt or some other non-conducting material is a wise precaution. In the " Meteorological Office " pattern rain FIG. 9. "METEOROLOGICAL OFFICE GAUGE. RAIN GAUGES 37 gauge (see Fig. 9) the receiver is furnished with a copper can slightly tapering in shape, so that in case of frost the joints are unlikely to be strained. In the " Snowdon " pattern (see Fig. 10) there is a straight-sided inner can, and within it a stout glass bottle with a narrow neck into which the LJKJ- FIG. 10. " SNOWDON " GAUGE. FIG. ii. CASELLA'S IN- SULATED SNOWDON. THE STIPPLED PORTION CONSTITUTES THE AIR- JACKET. pipe from the funnel is inserted. An improved gauge recently introduced by Messrs. Casella & Co. presents the additional advantage that the inner can is suspended from a narrow metal collar soldered on the inner side of the outer casing, so that it is entirely surrounded by an air-jacket; a great 38 RAINFALL OF THE BRITISH ISLES , : measure of immunity from evaporation and freezing is secured. The standard rain gauge, of which Figs. 9, 10 and 1 1 are examples, is characterized by extreme sim- plicity of construction, the number of soldered joints being kept to a minimum, and in every possible case being so planned that no unequal strain falls upon them in daily use and that if they should become defective no risk of leakage either inwards or outwards is likely to occur. In the " Meteor- ological Office " gauge the outer can is slightly splayed so that it can be fixed securely into the ground without becoming dislodged in removing the funnel. With the " Snowdon " this can best be done by embedding the gauge in a block of cement or in the interior of a piece of drain-pipe sunk flush into the ground. The old practice of fixing a gauge by means of wooden pegs is not recom- mended. In some gauges a slight elaboration is intro- duced by making the space immediately below the funnel hollow, so that in the event of snow it may be rapidly melted by pouring in hot water. This is a great advantage in districts where snow is frequent, since if the accumulated mass is melted slowly, it is apt to evaporate in the process. A properly constructed gauge of the standard pattern, made of copper, well fixed and carefully used, should be capable of lasting and retaining its accuracy for half a century, and even gauges of galvanized iron, especially when painted on the outside from time to time, should remain in good condition for thirty years. It is undesirable to paint the inside of the funnel, since the surface of the paint in time becomes spongy and absorbent. RAIN GAUGES 39 To make a reading of the gauge it is necessary to remove the funnel, which fits closely over the lower outer can, and to pour the accumulated water into a graduated glass measure. The diameter of the measure being made smaller that that of the funnel of the gauge, any desired magnification of the scale can be obtained, and no difficulty is experienced in obtaining measurements accurate to *oi inch or o-i mm. Any further refinement of the measure- ment is unnecessary for ordinary purposes, and is, as a rule, meaningless. There should, however, be no difficulty in graduating the glass to an accuracy of ooi inch, ancT this is desirable, although the readings are only taken to -01 inch. In order to give an open scale, the glass measure for a 5-inch gauge should not exceed about 1-5 inch in internal diameter and for an 8-inch gauge about 2-0 inches. Glasses with these diameters respectively, would be inconveni- ently long and liable to be broken if made to contain more than '50 inch of rain, and this capacity is usually adopted. In case of a greater fall of rain it is therefore, of course, necessary to fill the glass to the -50 inch mark as many times as required, and then measure the residue, subsequently adding together all the amounts measured. Since the units of measurement for readings in inches and millimetres are respectively -oi inch and o-i mm., the correct practice in measuring very small falls is to ascertain whether they reach half- unit measurement, i.e. -005 inch and 0-05 mm., respectively ; if less than this amount, the few drops of water should be thrown away and the day entered as rainless ; if more than this and less than the unit amount, the reading should be entered as 'Oi inch or o-i mm. respectively. The number of days on 40 RAINFALL OF THE BRITISH ISLES FIG. 12. FLAT-BOT- TOMED MEASURING GLASSES. which rain is recorded thus depends in an important measure upon the accuracy of the lowest graduation mark. In glasses with flat bottoms, as shown in Fig. 12, it is often very difficult to de- termine this point with certainty and a very slight personal bias on the part of the observer may make a serious difference in the number of days with rain recorded. In the " Camden " pattern of measure (Fig. 13), this difficulty is to a large extent overcome by tapering the glass at the bottom, thus spacing out the lowest graduations. An additional graduation mark for the half-unit is added below the unit graduation, enabling the de- cision as to ^i whether a small fall should be entered or not to be made with ease and cer- tainty. This pattern of glass appears to have been introduced *< first in Germany, and the same p r i nc i pi e is adopted in a somewhat different form in the Norwegian meteoro- FIG. 13. " CAMDEN " MEASURING GLASS. FIG. 14. MEASURING GLASS NORWEGIAN PATTERN. RAIN GAUGES 4 1 logical service, by the use of a measuring glass in which the magnification of the lowest graduation mark (in this case PO mm.) is made by means of a small glass cone at the bottom (see Fig. 14). The glass measure should preferably be con- structed of crown glass of good quality and of even transparency. With very thick or slightly opaque glasses a distortion of the line of vision renders the exact position of the surface of the water difficult to determine. In a narrow glass the true water- surface, of course, takes the form of a meniscus, or slight concavity, and the measurement should always refer to a line tangential to the bottom of the curve. Care must be exercised, however, in cases FIG. 15. MENISCUS. FIG. 1 6. FALSE MENISCUS. of slight distortion to detect the true bottom of the meniscus : this is often seen as a black line between two fainter apparent surfaces. The importance of carefully making allowance for distortion is apparent when it is remembered that a systematic error of, say, "005 inch in every daily reading will give an error of about I *oo inch in the course of a year. As a check upon the accuracy of the daily readings it is desirable whenever possible to place a second gauge beside that read each day and to make a reading once a week or once a month. For this purpose, or for monthly readings at stations where a daily visit is impracticable, especially if in regions of heavy, rainfall, gauges of larger capacity than the " Sn'ow- 42 RAINFALL OF THE BRITISH ISLES don " are usually required. Two standard patterns are in use, the " Bradford " and the " Seathwaite," the latter being designed for use only in exception- ally wet local- ities. The " Brad- ford " rain gauge (Fig. 17), origin- ally designed for use at the moor- land stations of the Bradford Corporation Waterworks in Yorkshire, is identical with the " Snowdon " in respect of its funnel, but the lower can is lengthened to give any desired capacity. The gauge is sunk in the earth so that it projects I foot, and the depth to which the lower part is buried usually secures FIG. 17." BRADFORD " GAUGE. immunity fro m freezing. This gauge contains no bottle, but an inner metal recep- tacle covered by a diaphragm pierced to allow the pipe from the funnel to enter and also to allow the RAIN GAUGES 43 water to be poured out for the purpose of measure- ment. A cedar-wood rod tipped with brass is pro- vided, graduated suitably to the diameter of the inner can, and when a reading is desired the first step is to make a preliminary measurement by dipping the rod into the water and reading off the amount by noting the portion wetted. This reading is taken only to the nearest *io inch. The water is then poured into a measuring glass and measured as in the case of the daily gauge. The glass is usually made to contain I'oo inch and graduated to *io inch, but accuracy to oi inch is easily obtained, if desired. The " Seathwaite " gauge (Fig. 18) was specially designed by Dr. H. R. Mill for use in remote mountain areas where exceptional rainfalls may be looked for, and where, owing to snow, it is some- times impossible to make visits even every month. The funnel is 5 inches in diameter at the orifice and is splayed to a diameter of 8 inches where it fits over the lower can. A locking device prevents it from being tampered with by any unauthorized person removing the funnel. The lower can, which is about 14 inches deep, is double,' forming a casing, and the space be- tween the sides is packed with felt to minimize the risk of damage by frost. Within this double outer body is a second can which holds the accumulated water. The gauge is provided with a metal " dipper," suggested by the late Mr. Gethin Jones, consisting of a narrow-necked and flat-bottomed vessel with a long handle holding exactly 5 inches of rain-water when full to the brim. Before reading, a preliminary rough measurement is taken with a rod, as in the " Bradford " gauge. In making the exact reading the dipper is immersed in the water and withdrawn as often as it is completely filled, each fill repre- 44 RAINFALL OF THE BRITISH ISLES senting, of course, 5 *oo inches. When it is no longer possible completely to fill the dipper by immersion, it is laid aside and the inner can drawn out, the small residue of water being measured by a glass in the FIG. 1 8. " SEATHWAITE " GAUGE. usual way. It is advisable to test the capacity of the dipper from time to time, as any dent will affect its accuracy. One of these gauges has been in use on the Stye, Cumberland, for some years, a district RAIN GAUGES 45 in which as much as 50 inches of rain sometimes falls in a single month, and where heavy snow and severe frost are of frequent occurrence, and has given satisfactory results. The adoption of the standard rain gauge for use in the British Isles is largely due to the exhaustive experimental work carried on under the direction of Mr. Symons during a period of nearly thirty years between 1860 and 1890. Observations were made by Colonel Michael Foster Ward at Castle House, Calne, in Wiltshire, commencing in 1863, and simultaneously by Rev. J. Chadwick Bates at St. Martin's, Castleton Moor, near Manchester. In 1865 Mr. R. Chrimes set up an elaborate col- lection of experimental gauges on the flat roof of the Boston Reservoir, Rotherham, and in the following year Rev. T. E. Crallan undertook the observation of a series of gauges at Hurst Green, Sussex, the latter being designed to test the material best suited for the construction of the gauge. About a year later Mr. Crallan's set was handed over to Rev. C. H. Griffith, of Stratfield Turgiss, near Reading. To him afterwards were sent the various experimental gauges which Mr. Symons had had in use since 1863, an d four years later, when Colonel Ward left Calne, the whole of the gauges in his charge were also sent to Stratfield Turgiss. A separate series was established at Aldershot in 1869 in the care of Sergeant Arnold. In 1870 the Stratfield Turgiss gauges were handed over to Rev. F. W. Stow, of Hawsker, near Whitby, in order that the experimental readings might be repeated in the more exposed climate of the north, a precaution which experience showed to be of great importance, though from the point of 46 RAINFALL OF THE BRITISH ISLES view rather of exposure than of the pattern of gauge. These observations will be mentioned later. In 1872 the Rotherham experiments were com- pleted, and in 1875 the Rotherham Corporation undertook the charge of the gauges and had them erected on the bank of the Ulley Reservoir ; these gauges, like those at Hawsker, also gave invaluable information rather from the point of view of expo- sure than any other. The records were kept up with certain modifications until 1890, and form one of the most conclusive pieces of deliberate scientific research on rainfall observation ever carried out. Further experiments on the size and position of gauges were commenced in 1877 by the late Mr. George Dines, and in 1881 a complete discussion of the results, together with those of previous obser- vations, was undertaken. The detailed description of the various series and the conclusions to which they led are published in the annual volumes of British Rainfall. OBSOLETE PATTERNS OF RAIN GAUGE The several series of investigations pointed conclu- sively to the superiority of the " Snowdon " type of gauge, and it is a matter for regret that many of the older patterns are still to be found in use, and even more unfortunate that they are still sometimes manufactured and sold to the uninstructed. The defects of these gauges are of importance, because it is often necessary to utilize records made with them when no others are available, and because nearly all the records of more than fifty years ago must be accepted with caution as liable to be to some extent vitiated. As will be shown later, the magni- tude of the error in results caused by the use of RAIN GAUGES 47 non-standard instruments depends largely upon the conditions under which they are exposed, and infor- mation on this point is too often lacking. In comparing records of early periods with modern observations, great care must be exercised on this account. The principal types of obsolete rain gauge are : (i) the " Howard " gauge, invented by Luke Howard, the well-known student of cloud forms, and compiler of early statistics of the climate of London ; (ii) the " Fleming " and other float-gauges ; (iii) the " British Association ' gauge, sometimes known as the " Symons " gauge, although discarded by Symons at an early stage of his investigations ; (iv) the side-tube gauge ; (v) the Glaisher gauge, de- signed by James Glaisher; (vi) rectangular gauges ; and (vii) tap-gauges. In practically all the early patterns, as already mentioned, the shallow funnel was a fundamental defect. In the " Howard " and " British Associa- tion " gauges a further disadvantage was inadequate size, not only leading to overflow in case of unusually heavy rain, but precluding the sinking of the lower part of the gauge in the ground, thus involving risk of evaporation by over-heating, or of bursting by frost. Howard's gauge, which consists of a simple funnel fitted over the neck of a bottle by a metal FIG. IQ. " BRITISH ASSOCIA- TION " GAUGE. 48 RAINFALL OF THE BRITISH ISLES collar, presents a peculiar defect. This collar not infrequently becomes partially detached owing to the strain of constant removal, especially when frozen, and when this occurs, raindrops running down the outside of the funnel find their way through the crack (see point marked in Fig. 20) and enter the receiver, thus unduly increasing the amount of water caught. In the Jagga Rao gauge (see p. 53) this risk was obviated by the provision of a little metal hood (see Fig. 27). A similar source of error affects FIG. 20. 'HOWARD" GAUGE. FIG. 21. BOX GAUGE. certain patterns of box- gauge (see Fig. 21), and is particularly troublesome in the " Glaisher " gauge, which is otherwise a very good pattern. It is not at all easy to detect the first symptoms of leakage in the Glaisher gauge, and numerous instances have been brought to notice in which serious errors have arisen through this ca.use. In extreme cases the collar fitting over the receiver falls away altogether, so that the funnel rests upon the lip of the receiver, and any slight distortion of the metal, in either the funnel or the can, forms an aperture for the illegiti- mate entrance of rain which has run down the outer RAIN GAUGES 49 face of the funnel, and of blown raindrops which strike the gauge at the point of leakage. Dr. Mill, in his paper " On the Best Form of Rain Gauge," l gives an interesting example of the effect of the progressive deterioration of uncared-for Glaisher gauges. A number of Glaisher gauges which had been condemned on inspection as defective were continued in operation for some years together with the Snowdon gauges, which had been set up alongside to test the extent of their inaccuracy. Three gauges of the group, A, B, and C, were sound, but one of these (A) gave small readings for another reason. The remainder, D, E, F, G, and H, exhibited the defect described. The records during six years are given in the following table : Year. i9< >i. IQC >2. IQC >3- i 9 ( H. xg< >5- * B a o Ij O C 8 | a o 53 H S3 8 Gauge. a I 1 T) I H 1 "M 'M .a 1 o I 1 1 O en o w 1 in. in. in. in. in. in. in. in. in. in. in. in. g rA 27-5 28-1 26-1 28-0 43-3 45-5 32-2 34-7 31-3 34-3 35-8 39-o s -j B 26-4 26-3 25-4 25-4 47-3 48-8 29-6 29-6 27-2 26-8 31-2 30-7 w ^C 3i-9 31-2 28-3 27-6 55-2 54-6 32-5 32-2 30-8 30-6 36-7 36-6 o> (D 33-7 3i-7 32-6 30-0 62-6 57-8 34-5 3i-7 33-2 29-9 38-6 36-4 '-C I E 32-6 30-9 29-2 257 62-7 46-8 41-6 28-3 34-o 24-3 38-2 28-7 ** FIG. 31. NIPHER SHIELD FIG. 32. BILLWILLER's DESIGN FOR NORWEGIAN PATTERN. NIPHER RAIN GAUGE SHIELD. in April 1885. In his experimental work in Russia Wild observed a series of elevated gauges and found FIG. 33. GAUGE SUNK IN PIT. the diminution of rainfall with elevation to be far greater than had been found in England. He 64 RAINFALL OF THE BRITISH ISLES attributed this result to the greater frequency of He therefore established a gauge sunk snow. in a pit, as had been previously tried by Colonel Ward at Came, and was led to the opinion that, provided such a gauge can be efficiently protected from snow-drift, it will indicate the true fall more readily than an exposed gauge. At a later stage of his experiments Wild attached Nipher shields to the whole series of gauges (except the pit gauge), and carried out a rigorous comparison between this form FIG. 34. SECTION OF WILD'S FENCE. of protector and a simple fence enclosure. The latter was composed either of solid woodwork or interwoven wicker, and when constructed to the dimensions proposed (about 16 feet square and 8 feet high) was found to be extremely efficient. It should be noted in reference to the great height (8 feet) of the Wild fence, that the normal rain gauge exposure in Russia was at least I metre, pro- bably in many cases 1-5 metre, as a precaution against risk of burial in snowdrift, so that the angle at which driving rain would be intercepted was far smaller than if the gauge had been at the height of I foot. The EXPOSURE OF RAIN GAUGES 65 counterpart of the Wild fence which has been suggested by Dr. Mill for use in wind-swept districts in this country is therefore a turf wall 2 feet high, forming a ring about 6 feet wide, with the gauge in the centre. Great as is undoubtedly the value of artificial wind protectors in ensuring the proper functioning of a rain gauge in exposed situations, it should always be borne in mind that natural shelter from wind is preferable. The ideal rain gauge exposure for wind-swept districts is one in which the prevailing wind is tempered by the interference of gently rising land or by a belt of trees. The angle of incidence of such shelter should not be greater than from 10 to 15 with the horizontal, varying with the degree of exposure to strong wind. If no such shelter is available, the gauge should be placed in a slight hollow, not of course a deep hole, so that it is protected from the direct impact of the wind. The gauge itself should always, if possible, be on level ground, and if in a hollow it should be at the bottom, not on the slope. If in the open, land sloping down- wards towards the prevailing wind direction, even if at some distance, is. detrimental. These precautions become less and less necessary with distance from the sea, except at hill-stations; and in sheltered valleys any additional specific shelter is usually unnecessary. In any circumstances, however, it is advisable to avoid sloping land for a gauge site, and even in fairly calm conditions a hillock or terrace makes a bad exposure. Over-exposure of rain gauges is probably the most fruitful source of error in rainfall observing, and far too little attention has hitherto been paid to it in selecting sites for rainfall stations. It is extremely difficult to lay down any simple instruction which will entirely meet the case. A system of inspection 66 RAINFALL OF THE BRITISH ISLES by officials thoroughly conversant with the varying requirements of each locality would do much to remedy the defect, but some time must elapse before any such scheme could be put into effective operation. The opposite pole of danger in respect of faulty gauge exposure viz. over-shelter is much easier to avoid. Whilst it is true that a degree of shelter which would be harmful in one case would be much less so in another, broadly speaking, the conditions are similar everywhere. It is usually safe to suggest that the top of any object, such as a wall or other building, should never subtend an angle greater than 45 with the gauge. In windy positions, where the rain commonly falls at an acute angle, 30 is preferable to 45. In the case of growing plants, shrubs, or trees, the angle should in no case be greater than 30, that is to say, the distance of any such object should be at least twice its height. This allows for growth, which is, apt to be overlooked as it takes place, and even when this precaution has been taken the encroachment of growing trees must never be lost sight of. It should be noted that whilst the error introduced by undue shelter by a wall or building is always caused by the interception of part of the rain, that caused by trees or shrubs may be either positive or negative, interception occurring under certain conditions, whilst at other times water-drops hanging on leaves may be blown into the gauge or drip from overhanging branches. In at least one instance in which a gauge was placed by a careless observer actually under trees, the posi- tive and negative errors practically balanced, and, until the gauge was inspected, no fault in the exposure was suspected. This method of obtaining accurate records is not recommended ! CHAPTER V MECHANICAL RAIN GAUGES SELF-REGISTERING rain gauges in great diversity have been introduced from time to time, but most of them have been short-lived and their defects, if any, have had little effect in putting erroneous readings on record. The automatic rain gauge has its proper uses, and is an important auxiliary to the direct- reading gauge when employed in the legitimate manner. It is necessary, however, to insist that in no respect can it be regarded as an instrument to relieve the observer of trouble ; it should never be used entirely to take the place of an ordinary gauge, if only for the reason that no piece of mechanism, however perfect, can be expected to work with sufficient accuracy and uniformity in all circum- stances when exposed to the vagaries of the British climate. It is therefore recommended that when a self-registering gauge is employed, and this is desirable whenever practicable, a gauge of the ordi- nary pattern should be set up alongside and used as a check upon its indications. The proper func- tion of the self-registering rain gauge is to give a record of the time at which rain has fallen and of its duration and intensity, and for this to be done in a satisfactory manner certain precautions are necessary. The types of automatic rain gauge which do not 67 68 RAINFALL OF THE BRITISH ISLES supply these data and which are primarily intended for labour-saving devices are on this account to be regarded as objectionable. They have all the disadvantages of mechanical gauges without any of their advantages. The pattern most commonly met with is the dial gauge. In this instrument the rain, passing through the funnel, falls into a twin bucket balanced on a knife-edge and made to tip up and empty itself when rilled to a certain weight, usually corresponding to -01 inch. The other half of the bucket is then pre- sented for filling, and in its turn tips up and empties, bringing the balance back to its former position. The movement of the bucket is made to turn an escapement wheel which actuates a pointer on a dial, FIG. 35. DIAL GAUGE. so that the observer is able to read off the amount of rain which has fallen. In theory, there appears to be no great objection to the instrument, but any slight rusting of the parts or unequal wear- ing of the mechanism introduces an error so dispro- portionately large that after use for a year or two the indications of dial gauges are seldom to be trusted. A gauge depending on a somewhat similar device but so constructed that the movement of the buckets actuates a cam and by means of a lever raises a pen- arm and traces a line on a revolving drum has attained MECHANICAL RAIN GAUGES 69 a certain measure of popularity. This is enhanced in the view of some by the introduction of an electrical appliance whereby the recording part of the appa- ratus may be indoors. There is a certain fascination in watching the movements of a pen-arm tracing a line on a chart on a library table while the rain beats against the windows without, but this type of gauge can be regarded less as a scientific in- strument than as an interesting toy. Besides suffer- ing from the errors intro- duced by friction already men- tioned, auto- graphic tipping- bucket gauges recording by a pen-line fail to give a true record of the duration of light rain. The trace is, of course, in the form of a series of steps, each, as a rule, representing -01 inch, and this amount is recorded without any indication of the time in which it was falling. A considerable improvement on the tipping- bucket principle is that of the counterpoised bucket. In gauges of this kind, of which the " Casella Standard " is the most important FIG. 36. TIPPING BUCKET RECORDING GAUGE. FIG. 37. CASELLA STANDARD GAUGE. 70 . JIG.. 38. BECKLE.Y GAUGE.. 7* 72 RAINFALL OF THE BRITISH ISLES representative used in this country, the water is conducted into a receptacle which sinks uniformly with the increased weight until a fixed amount, usually -20 inch, of rain has accumulated. A simple device at this point overturns and empties the bucket, which is brought back to its original position by a counterpoise weight. The movement of the bucket is made to actuate a pen recording on a drum. The " Beckley " rain gauge is somewhat similar in principle, but in this case the bucket floats in mercury, and is emptied by a syphon. The objection to the tipping-bucket gauge, that friction interferes with the action, is of course applicable also to the counterpoised-bucket type ; but the error introduced is much smaller, owing to the larger amount of water dealt with in each operation. There is no doubt, however, that both the Casella Standard and the Beckley are sluggish in action and inferior in light rain to more modern automatic gauges. It is of importance to appreciate this in view of the use of the Beckley gauge at the official observatories of the Meteoro- logical Office, and of the long record kept at the headquarters of the British Rainfall Organization at Camden Square, London, by a Casella Standard gauge. Most records of rainfall duration kept before 1908 must be accepted with reserve on account of the want of sensitiveness in the instruments used in comparison with later patterns. In 1911 Messrs. Casella introduced an improved model (Casella's Recorder) (Fig. 39), overcoming most of the objections to the old Standard by the use of lighter parts and anti-friction bearings, and this appears to have given satisfactory results. In this model the water is conducted from the funnel to MECHANICAL RAIN GAUGES 73 a balanced bucket, which is suspended from a two- armed pivoted lever and counterpoised by an adjustable balance- weight. As the bucket fills, it drops, and on reaching the lower end of a guide- plate turns over and empties itself, when the counterpoise weight causes it to return to its original position. The pen is attached to the end of the lever farther from the bucket, and is hung FIG. 39. THE CASELLA IMPROVED RECORDER. at right angles, being carried behind the revolving drum so that the time co-ordinates are rectangular. A great step in the improvement of self-registering rain gauges was the re-adaptation of the float principle in Halliwell's Patent. The prime difficulty, which no previous design, save possibly Hellmann's gauge, had overcome, was that of bringing the float back to zero when the receiving chamber was full, a simple syphon being unsatisfactory for 74 RAINFALL OF THE BRITISH ISLES many reasons. The Halliwell gauge, designed for use at the Fernley Observatory, Southport, for Mr. J. Baxendell, requires care in handling, and is not entirely free from minor defects, but in the hands of a conscientious observer who will not grudge a little trouble in keeping the working parts in order, it gives results of great accuracy. The float system was also made use of in the " Hyetograph," also designed by Halliwell, on prin- ciples suggested by Dr. Mill. The descriptions of the mechanism of the Halliwell gauge and of the " Hyetograph " given in Dr. Mill's article in British Rainfall, 1908, are so clear that it seems best to reprint them. " The Halliwell Patent Recording Rain Gauge. This instrument gives the best and clearest trace of any I have tested. It is made to register on the drum seven days' rain or twenty-four hours' rain ; and the latter form is the only one which I recommend, as the time- scale on any drum running for a week is too contracted for measuring the intensity of short falls. "The large receiving funnel .... conducts the rain by a curved pipe to the bottom of the float-chamber, shown in the middle of the central diagram of Fig. 40. The float carries a rod, the top of which rests against the lower side of the pen- carriage (shown enlarged on the left-hand side of the diagram) and pushes this up, sliding between the rigid metal guides until the float-chamber is full and the pen has reached the top of the chart. The float-chamber is filled by half an inch of rain, and a side tube projecting on the right has hanging over it, on a delicately adjusted catch, a tube forming the short arm of a syphon, which branches above into an inverted U-tube, forming the long arm of the syphon (shown separately on the right-hand side of the central section in Fig. 40). As the lower part of the float-rod emerges above, on the float-chamber being filled, a triangular piece of brass attached to it throws over a small rocking weight, which de- taches the suspended syphon, and this falls with the central short arm into the side tube of the float-chamber, starting the syphon full-bore, and emptying the chamber in five seconds. The float FIG. 40. HALLIWELL GAUGE. 75 76 RAINFALL OF THE BRITISH ISLES falls and the pen comes to zero on the chart. The water syphoned out of the float-chamber rushes into another chamber, at a lower level, with a very small orifice below, and fills it, raising a second float in that chamber, a rod from which, running through the axis of the syphon, lifts the syphon up and hangs it on its catch, ready to act again. The water then slowly trickles out of the syphon- FIG. 41. HYETOGRAPH, GENERAL VIEW. float chamber and the whole operation repeats itself for every half-inch of rain that falls. " The Halliwell gauge is rather complicated, and as mercury is required to act as a joint for the syphon, it requires some care in fitting up. The adjustment of the catch holding up the syphon is a delicate matter, as if it is too strong it may not be released, if too sensitive some accidental tremor may set it off before its time ; but we are bound to say that as a rule it acts admirably." MECHANICAL RAIN GAUGES 77 " The Hyetogra-ph. The ordinary rainfall observer has neither the time nor the technical skill to tend a piece of delicate mechanism, nor does he care to spend as much as .17 on an instrument. 1 I had long thought over some way of cheapening the recording gauge by simplifying the mechanism without sacrificing the open scale usually given by frequent discharges of the water collected. -O COOO 00 M Tft^M t^MOOOO O O> N O N OO HI c\O < ^- c cocococo-^t-ioo o 6 M O MOO CONOO O lOt^M M 53 Ti-TfTj-TfO t^r^oio>ooio "OOOOOOOOOOOO oq o cooo O coOOO OO 53' CO TJ- co >O O 00 ... 909999999999 po o M r^t^t-ioo^ r>.M >. o ir-. iOO CO 122 67 69 119 49 1906 156 II 7 94 68 158 77 60 85 43 140 114 92 1907 57 86 89 133 152 153 78 9 6 32 140 86 119 1908 82 96 156 122 104 70 96 106 118 48 64 96 1909 '74 52 154 141 90 106 117 77 81 141 47 138 1910 119 170 69 131 103 109 109 141 30 76 119 126 igil 62 123 77 108 83 107 58 70 79 87 133 175 1912 118 107 167 47 80 182 112 161 57 IOI 94 145 1913 M3 70 164 165 130 94 3^ 43 92 94 1 20 78 1914 79 1 60 162 81 85 68 I0 7 88 67 56 132 1 80 1915 124 185 60 96 79 60 I 44 76 58 77 80 167 1916 107 169 IO2 117 143 102 82 91 65 162 130 97 1917 72 47 92 94 105 TOO 73 173 87 146 108 58 1918 112 138 57 72 95 60 140 84 211 IOI 93 128 1919 128 77 141 112 60 92 55 87 95 58 90 153 the whole range of possible variation, but it is sufficient to give a fail idea of the conditions. In 27 months out of the 204 the general rainfall was 50 per cent., or more, in excess of the average, but in only n months was there a deficiency of 50 per cent. The fall was as much as twice the average only once, and was never more than 70 per cent, below the average. In the particular period under consideration, September showed the greatest range, including both the driest and wettest months relatively to the average ; but a general divergence from the average of 50 per cent, or more was far more frequent in February and March than at any other time of year. November showed the smallest range, and July and November were the only months which never experienced a general excess of 50 per cent. ; whilst January, March, May, 14 210 RAINFALL OF THE BRITISH ISLES and June in no case showed a deficiency of 50 per cent. One of the most remarkable features of the table is the fact that September had less than its average fall in 14 out of 17 years, this occurring in 9 consecutive years from 1909 to 1917. On the other hand, there was only one dry April from 1904 to 1911, inclusive. The longest run of consecutive dry months was 7, from May to November, 1919 ; and the longest run of consecutive wet months 7, from December 1915 to June 1916. The period from January to October 1903, 10 months, with only one falling to as much as 5 per cent, below the average, was, however, probably the most remarkable period of sustained high general rainfall. Broadly speaking, wet winter months over the the country as a whole may be regarded as indicative of an abnormal tendency to orographical rains, probably brought about by an increased intensity in the great Atlantic wind-drift. Wet summer months are also sometimes due to an unseasonable prevalence of south-westerly weather, but more often to cyclonic rains, and sometimes to thunder- storms, which are, however, seldom sufficiently widespread to affect large areas. It is, therefore, possible that by examining the general rainfall of the summer and winter separately, some clue may be obtained on general lines to the fluctuations from year to year in the prevalence of rainfall of the different types. The general percentage values for the six months October to March and for the .six months April to September for the last 13 years show that during this period the winter rainfalls were, relatively to the average, greater than the summer falls, suggesting that in recent years orographical rains have been more prevalent than is SEASONAL VARIATIONS 211 normally the case. The run of eight consecutive wet winters from 1911-12 to 1918-19 is a striking SEASONAL GENERAL RAINFALL OF THE BRITISH ISLES AS PERCENTAGE OF AVERAGE Year. Per cent, of average. Year. Per cent, of average. Winter. Oct.-March. Summer. April-Sept. Winter. Oct.-March. Summer. April -Sept. 1906- 7 1907- 8 1908- 9 1909-10 1910-11 1911-12 1912-13 1913-14 97 "3 78 114 97 131 120 III IO2 I0 3 97 104 82 no 87 83 1914-15 1915-16 1916-17 1917-18 1918-19 123 116 104 103 112 86 97 108 H3 83 Average of 13 years . 109 97 feature. The series is not long enough to show whether such sequences are abnormal. CHAPTER XIII THE FLUCTUATIONS OF ANNUAL RAINFALL IN studying the distribution or incidence of rainfall in days or months we are dealing with periods of time which individually have no special significance. It is true that a slightly marked diurnal range of rainfall is distinguishable, but it is so utterly swamped by the enormous variability of daily rainfall as to be meaningless in relation to individual days. The division of time into calendar months, although extremely useful in some respects for meteorological work, is purely artificial, and even if the calendar month coincided with the lunar period, which it of course does not, there would still be no reason for regarding it as a natural period from any climato- logical point of view. The year, on the other hand," is in every way suitable for the grouping of climatological observations, nearly all of which exhibit a pronounced annual term. It is probable that some improvement might be effected by ending the year on some date other than December 31, which has the effect of throwing parts of each winter into two separate years; but the difficulty of deciding precisely when winter ends and spring begins, and the fact that this occurs at varying times in different years and in different places, makes it impossible to hit on any precise date for terminating the climato- logical year which would be entirely free from disadvantages. Attempts have been made to obtain a more natural yearly period by tabulating records 212 ANNUAL FLUCTUATIONS 213 from October I to September 30, but by far the greatest bulk of the available statistics are grouped in civil years from January I to December 31, and to adopt any other would be extremely inconvenient. If the annual totals of rainfall at a large number of places for any individual year are examined, they will be found to exhibit a wide range of variation, the amount recorded at the wettest stations being sometimes nearly ten times as great as that at those of least rainfall. Generally speaking, it will be observed that the more elevated stations have a considerably larger rainfall than the low-lying ones, and further that, in the British Isles, the westerly stations are as a rule wetter than those in the east. If the total rainfall of a year is mapped, in order to study the distribution more closely, the records will thus be found to indicate a distribution of a strongly marked orographical type. If now we take the records for any .other year and deal with them in the same way, the orographical type will be reproduced with only relatively minor differences in the run of the isohyets ; but if the two years were dissimilar in regard to their total rainfall, the isohyets will be found to be shifted in places. In a map of the average annual rainfall over a number of years the run of the isohyets is very much like that in any individual year, and their positions are somewhere about midway between the extremes shown in the years of highest and lowest fall. The most extreme shifting of the isohyets in individual years from their average positions will not necessarily occur in all parts of the map in the same years that is to say, the greatest departures from average conditions may occur in one district in one year and in another district in another year. 214 RAINFALL OF THE BRITISH ISLES Although the variations from place to place are considerable, the departures from the average at any one place in the most extreme years appear to have definite limits. The most convenient way of studying these departures is to express the total rainfall of each year as a percentage of the average annual fall at the station. There are few completely trustworthy records extending back beyond about 1860, and as examples of the annual fluctuations since that date we may examine four long records, selected to represent different parts of the British Isles. These are at London ; Haverfordwest, in Pembrokeshire ; Glengyle, near the head of Loch Katrine on the borders of Perthshire and Stirling- shire ; and Belfast. The four are thus situated re- spectively in England, Wales, Scotland, and Ireland. Year. London. Haverfordwest [Pembrokeshire) . Glengyle (Perthshire). Belfast (Antrim). Year. Annual rainfall. Per cent, of average. Annual rainfall. Per cent, of average. Annual rainfall. Per cent, of average. Annual rainfall. Per cent. of average. in. in. in. in. 1860 32-24 126 56-99 119 94-20 103 38-23 ill 1860 1861 22-27 87 51-80 108 112-50 122 34-02 98 1861 1862 27'57 108 38-30 80 105-10 114 39-18 H3 1862 1863 21-59 84 45-13 94 105-50 U5 36-92 107 1863 1864 16-93 66 40-06 83 80-60 88 29-49 85 1864 1865 29-48 H5 50-77 106 72-20 79 32-02 93 1865 1866 31-60 124 54-97 114 100-70 no 35-56 103 1866 1867 26-29 103 55-87 116 98-90 108 32-68 95 1 867 1868 23-40 9i 56-01 117 118-30 129 3I-58 9i 1868 1869 25-42 99 54-69 114 91-00 99 32-57 94 1869 1870 21-32 83 40-01 83 71-30 77 30-14 87 1870 1871 25-02 98 46-73 97 90-10 98 31-91 92 1871 1872 33-86 132 69-78 145 127-80 139 44-46 129 1872 1873 22-67 89 45-67 95 95-6o 104 3I-I3 90 1873 1874 18-82 74 5I-I5 106 106-60 116 34-78 101 1874 1875 28-44 in 58-43 122 91-20 99 31-98 93 1875 1876 26-16 102 53-49 III 93-70 102 39-89 116 1876 1877 28-17 110 64-18 134 128-50 140 42-28 122 i877 1878 34-08 133 54-05 H3 82-00 89 29-14 8 4 1878 1879 33-82 132 49-69 103 87-00 95 33-52 97 1879 1880 30-28 118 40-76 85 69-00 75 28-76 83 1880 1881 1882 27-92 27-14 109 1 06 45-i8 63-39 94 132 80-00 104-90 87 114 38-47 39-32 in 114 1881 1882 ANNUAL FLUCTUATIONS 215 T f\rAf\n Haverfordwest Glengyle Belfast i^ondon. Pembrokeshire) . (Perthshire). (Antrim). Year. Annual ainfall. Per ent. of verage. Annual ainfall. Per ^nt. of verage. Annual rainfall. Per cent, of average. Annual rainfall. Per cent, of average. Year. in. in. in. in. 1883 24-40 95 50-75 106 100-60 no 33-96 98 1883 1884 20-35 80 43-60 91 107-40 117 33-28 96 1884 1885 26-64 104 50-27 105 84-10 92 29-57 86 1885 1886 27-01 106 57-64 120 81-10 88 36-88 107 1886 1887 19-21 75 35-23 73 67-00 73 23-45 68 1887 1888 27-74 109 47-08 98 89-40 97 32-80 95 1888 1889 23-85 93 37-31 77 76-30 83 31-20 90 1889 1890 21-23 83 42-82 89 95-10 104 32-58 94 1890 1891 28-15 no 51-13 106 94-40 103 31-88 92 1891 1892 22'6l 88 37-45 78 89-60 98 31-21 90 1892 1893 19-80 77 35-55 74 91-90 100 25-92 75 1893 1894 27-94 109 49-75 104 101-70 in 31-63 9i 1894 i895 21-47 84 38-79 81 74-20 81 33-01 96 1895 1896 23-52 92 40-69 85 76-20 83 32-83 95 1896 1897 22-86 89 50-98 106 96-10 105 35-73 103 1897 - o o 1898 17-69 69 42-56 88 112-20 122 30-26 88 1890 1899 22-54 88 42-26 88 IO6-2O 116 34-91 101 1899 1900 23-28 9i 50-06 104 IO4-OO H3 40-56 117 1900 1901 22-17 87 45-50 95 76-70 83 32-10 93 1901 1902 20-84 81 40-72 84 65-90 72 30-41 88 1902 1903 38-10 149 56-69 118 I29-50 141 42-34 123 1903 1904 20-65 81 42-72 89 88-80 97 31-84 92 1904 1905 22-97 90 39-23 83 77-10 84 31-80 92 1905 1906 24-26 95 50-40 105 81-70 89 36-15 104 1906 1907 23-01 90 44-17 92 85-10 93 38-04 no 1907 1908 23-67 92 44-91 93 88-10 96 38-75 112 1908 1909 26-75 105 41-04 85 75-70 82 36-88 107 1909 1910 25-36 99 45-61 95 88-80 97 40-57 117 1910 1.911 24-79 97 49-83 104 94-80 103 35-70 103 1911 1912 27-88 109 56-59 118 106-60 116 44-43 129 1912 1913 22-41 88 53-16 in 84-00 9i 37-68 109 1913 1914 25-72 101 50-53 105 95-8o 104 34-98 101 1914 1915 32-18 126 49-27 103 75-40 82 36-63 106 I9 J 5 1916 34-01 133 43-33 9i 96-85 105 37-6o 109 1916 1917 30-04 117 41-84 87 76-25 83 36-00 104 19*7 1918 29-69 116 52-49 109 96-35 105 37-99 IIO 1918 1919 26-21 102 42-99 89 72-85 79 34-64 IOO 1919 Average, Average 18601910 25-59 100 48-03 100 91-84 IOO 34-57 IOO 1860-1919 Wettest year . 38-10 (1903) 149 (1903) 69-78 (1872) 145 (1872) 129-50 (1903) 141 (1903) 44-46 (1872) 129 (18721 U9I2/ Wettest year Driest 16-93 66 35-23 73 65-90 72 23-45 68 Driest year . (1864) (1864) (1887) (1887) (1902) (1902^ (1887) (1887,1 year Mean Mean deviation 3-62 14 6-12 13 ii-95 -t 13 3-43 10 deviation 216 RAINFALL OF THE BRITISH ISLES Fig. 112 shows the percentage variations in the form of curves. Apart from certain prominent Characteristics, such as the high rainfall of 1872 and Rainfall 1860-1919 PerCent of Average. FIG. 112. 1903 and the low rainfall of 1887, which are common to all the curves, it is not easy to detect any definite similarity, beyond the fact that the range of variation at all four stations is of the same order of ANNUAL FLUCTUATIONS 217 magnitude. The above table shows that the mean deviation is greatest, 14 per cent., in London, but that it is only I per cent, smaller at Haverfordwest and Glengyle. At Belfast, however, it falls to 10 per cent. This small range is a characteristic common to all Irish stations. The extreme range is 83 per cent, for London, the wettest year having 2j times as much rain as the driest ; 72 per cent, and 69 per cent, respectively at Haverfordwest and Glengyle, where the wettest year had approximately twice as much rain as the driest ; and 61 per cent, at Belfast, where there was rather less difference between the extreme years. These values represent the range of annual rainfall in the British Isles generally. Another way of examining the variation factor is to group the annual departures in order of magnitude. Departures from average. London. Haver- fordwest. Glengyle. Belfast. Per cent. No. of years. No. of years. No. of years. No. of years. o to 9 . 26 26 24 37 10 to 19 . * 21 25 24 I 7 20 to 29 6 6 9 5 30 to 39 . . 6 2 I i 40 or more . i I 2 It will be observed that departures exceeding 30 per cent, of the annual average occur with greater frequency in London than at the other stations, and that departures of 10 per cent, or more are far more rare in Ireland than elsewhere. These data illustrate the generalization that, broadly speaking, the percentage range of annual rainfall rises to a maximum in the east and falls to a minimum in the west, but that in Great Britain the 2i 8 RAINFALL OF THE BRITISH ISLES regional difference is not pronounced. The varia- tion is probably a measure of the insularity of the climate, since in continental climates, where oro- graphical rain is not so frequent as it is on the western sea-board of Europe, the annual rainfall exhibits a considerably wider range. Although at first glance the curves in Fig. 112 suggest that the variations from year to year are entirely capricious, a closer scrutiny shows certain broad tendencies overlying the incidental irregulari- ties. For example, in the curves for London and Haverfordwest the earlier years of the period con- sidered were distinctly wetter than the later years, though there is some indication of a recovery during the last decade. At Glengyle the alternation is less marked, but still apparent, and at Belfast the wet years are more pronounced in the last two decades than at any other period, though the wet spell in the seventies common to all the curves is clearly shown. The evidence for a long-period alternation of relatively high and low rainfall is much more conspicuously demonstrated if, instead of plotting the annual percentages, the values are expressed as successive overlapping ten-year averages, thus smoothing out minor or short-period fluctuations. This is done in Fig. 113. Here the curves exhibit a definite alternation, the rainfall rising to a maximum in about 1875, falling to a minimum in about 1890, and afterwards rising to a second maximum at the end of the period. The four curves do not precisely synchronize, and that for Glengyle is less definite than the other three, this station showing some indication of a shorter term, but all show the same tendency for alternate spells of preponderat- ANNUAL FLUCTUATIONS 219 ingly wet and dry years. It is interesting to observe in passing that the abnormally wet year 1903, common to all the curves, and indeed common to the whole of the British Isles, fell in a period of general dryness. Whatever, therefore, may have been the cause of the unusually great fall in this year, it was apparently not a phase of any long-period fluctuation, but a strongly marked exception to the general tendency prevailing at the time. The year Annual Rainfall I86Q-19I9. Overlapping 10 gear Means. London. no 100 90+ ' Glengyle. FIG. 113. 1887, on the other hand, the year of lowest rainfall over the country generally, was merely the driest of a long spell of years of deficient fall, and 1872 was the most conspicuous of a run of wet years (see Fig. 112). The alternation of wet and dry spells is even more markedly brought out by the use of the method of " residual mass curves " suggested by Mr. A. A. Barnes 1 for the analysis of rainfall records. The residual mass curve is constructed' by plotting 1 See Q.J.R. Met. Soc. t vol. xlv, 1919, p. 209. 220 RAINFALL OF THE BRITISH ISLES the percentage departure from the average as a cumulative value from the beginning to the end of the series. Thus the year 1860 for London had a Rainfall l86CH9l9._Fkr Gent df Average. Residual Mass Curves. werage values. lOqear Means. FIG. 114. rainfall 26 per cent, in excess of the average of the 60 years and the following year a deficiency of 13 per cent., the two years taken together thus exceeding the average by 13 per cent. ; the third year of the ANNUAL FLUCTUATIONS 221 series, 1862, had 8 per cent, above the average, and the aggregate departure for the three years was therefore 21 per cent. Continuing this process throughout the whole period, we get the curves shown in Fig. 114. The smooth lines in the same diagram represent the residual mass curves of the successive ten-year totals 1860-69, * 870-79, etc> They serve to bring out the general tendency more readily than do the crude annual curves. The position of the curves relatively to the zero, or average line, obviously depends upon the accident of the relative wetness or dryness of the earliest years, and may, therefore, be disregarded. The important point to notice is the effect of the pre- ponderatingly wet years in producing a tendency for the curve to rise and of the preponderatingly dry years in causing it to fall. In London, for instance, the eight years 1875 to 1882, inclusive, had an aggregate excess of 121 per cent., or 15 per cent, per annum, whereas the 20 years 1883 to 1902 had an aggregate deficiency of 190 per cent., or 9-5 per cent, per annum. The recovery from 1914 to 191 9, with an aggregate excess of 93 per cent., or 15-5 per cent, per annum, is well shown. At Haverfordwest the contrast between the wet and dry spells is still more striking. From 1865 to 1886 the aggregate excess was 209 per cent., or 10 per cent, per annum, and from 1887 to 1910 the aggregate deficiency amounted to the same total, 209 per cent., or 9 per cent, per annum. The peak of the curve at Haverfordwest is four years later than in London, whilst at Glengyle there is a double peak in 1877 and 1884, a less pronounced fall, and no recovery at the end. The Belfast curve is somewhat indefinite as to the precise date of the 222 RAINFALL OF THE BRITISH ISLES maximum, and the minimum is much earlier than at the other stations. The run of dry years from 1882 to 1893 and the run of wet years from 1905 to 1919 are very well marked. The relative values given for the individual years and groups of years of course depend for their magnitude upon the average of the whole period of sixty years considered. It would appear to be a justifiable assumption that some agency is operating to cause a periodic fluctuation in the rainfall, and although the time available for observation is too short to enable a definite term to be fixed, the data appear to indicate that during the sixty years 1860 to 1919 at least one crest and one trough are included. In all probability the amplitude and term of the fluctuation varies in different parts of the country and the epochs of maximum and minimum also probably vary in different places, though in very broad outline there is some tendency for a similarity in the type of fluctuation in all parts. We have no evidence at present to show with any certainty whether similar fluctuations have occurred in previous years, nor whether, if they occurred, their amplitude and term were the same as here shown. The consideration of this fluctuation is of great importance in selecting a suitable period for computing normal annual rainfall values. It has been shown that an average of 10 years may exhibit an excess or defect of as much as 12 per cent. ; and even if the period be as much as 50 or 60 years, should it be chosen so as to include two maxima and only one minimum, the average will be higher than the true normal, and if two minima and only one maxi- mum are included it will be too low. If the term ANNUAL FLUCTUATIONS 223 of the fluctuation can be fixed, it is not improbable that the average for some shorter period may be hit upon which will, by including only one maximum and one minimum, be closer to the normal than the average of the whole period. Sir Alexander Binnie l and others have examined this point with great detailandhaveconcluded that, broadly speaking, the average of 35 years will in nearly every case represent the true normal rainfall at least as closely as any longer period for which trustworthy obser- vations are likely to be available. It is possible to test the accuracy of this assumption conclusively only if two successive periods of 35 years can be compared, and there are certainly too few completely trustworthy rainfall records of 70 years' duration to enable this to be done. It is not without value, however, to examine the 26 successive overlapping periods of 35 years which occur within the 60 years 1860 to 1919 in their relation to the average of the whole period. The table on page 224 shows that by far the greatest number of 35-year periods had average values below that of the 60 years, a result which would be com- patible with the suggestion made above that these 60 years include two wet spells and only one dry spell, and, therefore, presumably show too large an average. The hypothesis of a constant value for 35 years is, however, shaken by the considerable range shown, the extremes differing by as much as 8 per cent, at Haverfordwest. This range is not appreciably affected by the question of the proximity of the 6o-year average to the normal. The percentage values in the table refer of course to individual stations. There is little doubt that a 1 Rainfall, Reservoirs, and Water-supply, by Sir A. R. Binnie. 224 RAINFALL OF THE BRITISH ISLES OVERLAPPING 35-YEAR AVERAGES AS PERCENTAGES OF AVERAGE OF 60 YEARS, 1860 1919 Period. London. Haverford- west. Glengyle. Belfast. 18601894 101 I0 3 IO2 97 1861 1895 100 IO2 IO2 96 1862 1896 100 IOI IOO 96 18631897 99 IO2 1.00 96 18641898 99 IOI IOO 96 1865 1899 99 102 IOI 96 1866 1900 99 102 IO2 96 1867 1901 98 IOI IOI 96 1868 1902 97 IOO IOO 96 1869 1903 99 IOO IOI 98 1870 1904 98 99 IOI 97 1871 1905 98 99 IOI 97 1872 1906 98 IOO IOI 98 18731907 97 98 99 97 1874 1908 97 93 99 98 18751909 98 97 98 98 1876 1910 98 97 98 98 1877 1911 93 96 98 98 1878 1912 9 96 97 98 18791913 96 96 97 99 1880 1914 95 96 98 99 1881 1915 96 97 98 IOO 1882 1916 96 96 98 IOO 18831917 97 95 98 99 1884 1918 97 95 97 IOO 18851919 98 95 96 IOO Mean value . 98 99 99 98 Extreme range 6 8 6 4 smaller range would be found if general values for areas were considered, and, speaking from general experience, to say that there is a range of about 6 per cent, between the largest and smallest 35-year averages which have occurred within the range of observation would probably be a fairly accurate statement. This may presumably be interpreted to mean that the average of any period of 35 years is not likely to differ from the true normal by more than 3 per cent. The wide adoption of the period ANNUAL FLUCTUATIONS 225 of 35 years as a basis for computing average rainfalls makes this range of variation a matter of some moment, and it must also not be overlooked that all rainfall measurements are only approximations to the true fall, and the limit of ordinary observational error, even under the most favourable circumstances, can hardly be put at less than about 3 per cent. In other words, if our inference is justified, the average of any period of 35 years will represent the true normal fall within the limits of observational error. As has been mentioned, the range of variation of annual rainfall is perceptibly reduced if, instead of dealing with individual stations, it is possible to refer to the general values for any large area. This is of course due to the fact that the extreme measure- ments do not occur simultaneously in all parts of the area, and thus the larger the area the more limited will the fluctuations be. For the purpose of demon- strating this we may examine the general rainfall for the whole of Ireland, arrived at by meaning a large number of interpolated values of annual rainfall from maps for the years 1865 to 1919 inclusive. There are unfortunately not sufficient data to carry the series back to 1860, but the average of the 55 years is probably within about 0-2 per cent, of that for the 60 years, so that the difference is immaterial. The table on p. 226 gives the general rainfall for each year and the percentage which it bears to the average. The mean deviation will be seen to be no more than 6-4 per cent, or about half that for an individual locality, and the extreme range only 47 per cent. The deviation from the average was less than 5 per cent, in 26 years and less than 10 per cent, in 43 years, reach- ing 20 per cent, only three times during the 55 years. 226 RAINFALL OF THE BRITISH ISLES GENERAL RAINFALL IN IRELAND, 1865 1919 Year. Rainfall. Per cent. Year. Rainfall. Per cent. Year. Rainfall. Per cent. 1865 43-i 99 I88 3 47-0 107 I9OI 42-4 97 1866 44-3 101 1884 43-2 99 IQO2 39-1 89 1867 41-9 96 1885 41-3 94 1903 54-4 I2 4 1868 44'8 IO2 1886 48-2 no 1904 44.4 IOI 1869 42-8 9 8 1887 33-7 77 1905 37'5 36 1870 39-3 9 1888 42-6 97 1906 41-0 94 1871 43' 98 1889 4 J> 5 95 1907 44-2 IOI 1872 54'4 124 l8qo 42-9 98 1908 43-5 99 1873 41-2 94 1891 43-2 99 1909 40-4 92 1874 43-1 99 I8 9 2 43-4 99 I9IO 47'4 108 1875 42-4 97 1893 37'3 85 I9II 42-5 97 1876 46-5 1 06 1894 45-8 105 1912 45-8 105 I8 77 51-9 119 1895 42-0 96 1913 45-3 104 1878 41-7 95 1896 41-7 95 1914 45'4 104 1879 4!-3 94 1897 47-8 109 1915 42-9 98 1880 40-9 94 1898 43-3 99 1916 47-5 109 1881 44'3 IOI 1899 44-1 IOI 1917 42-1 96 1882 49-6 H3 1900 48-8 112 1918 47-6 109 , 1919 39-2 90 Average 1865 1919, 43*8 inches. Wettest years, 1872 and 1903, 54-4 inches or 124 per cent. Driest year, 1887, 33'7 inches or 77 per cent. Mean deviation, 2-8 inches or 6-4 per cent. General Rainfall of Ireland. Percentage of Average. 1865- 1919. no- 100 90- 80-- 70-- A \/ v A A 120 --no 100 90 70 +30-- 20- + 10- -10-- 20-- -30 Residual Mass Curve. -30 FIG. 115. ANNUAL FLUCTUATIONS 227 Taking the overlapping decadal means, the wettest period is from 1 868 to 1877 with 102-7 per cent, of the average, and the driest. 1887 to 1896, with 94-6 per cent., the extreme range being thus 8-1 per cent. Of the 46 decadal means included in the period, 30 showed values within 2 per cent, of the average ; and the 17 overlapping 35-year averages were all within I per cent, of that for the whole period. The range of rainfall is certainly smaller in Ireland than in Great Britain, and the above generalizations must not, therefore, be applied to England or Scotland, but' it is nevertheless certain that the general values for these countries would show much smaller variations than are found at individual stations. Reverting to the table on pp. 214-215, attention may be drawn to the regular sequence of one wet year following two dry years at Haverfordwest throughout the twenty years 1889 to 1908. No such tendency is apparent at the other three stations, but nevertheless this temporary periodical recurrence was sufficiently widespread in England and Wales to be reflected in the general percentage of the average for the whole country from 1889 to 1909, as has been pointed out by Dr. H. R. Mill in British Rainfall. ENGLAND AND WALES. GENERAL PERCENTAGE OF AVERAGE ILLUSTRATING THREE-YEAR PERIODICITY 1889 92 1896 . Qi 1903 . . 129 1890 8 9 1897 . ^ 101 1904 . . 89 1891 110 1898 . 87 1905 . . 85 1892 89 1899 . 93 1906 . . 101 1893 83 1900 107 1907 . . 99 1894 106 1901 . 88 1908 . . 90 1895 92 1902 . 83 1909 .* . 104 228 RAINFALL OF THE BRITISH ISLES This sequence, so far as the general values for England and Wales indicate, was not apparent before 1889 and entirely broke down after 1909, but its reality between those two dates is beyond doubt. Mr. A. P. Jenkin : has pointed out that there is some evidence for the existence of a three-year period in rainfall over the whole of Europe, but that at intervals the order of the sequence is reversed, one dry year alternating with two wet years. He does not consider that the periodicity is confined to the twenty years commented on by Dr. Mill, though there can be no question, as far as England and Wales are concerned, that its existence was very much more marked at this time than any other. Probabably a period of a fraction less or more than three years would be found to carry the sequence further, the 21 years being the phase of apparent coincidence, and if this is the case the phenomenon should sooner or later be repeated. It is attractive to regard the period as likely to prove one-third of the sunspot cycle. Other real or apparent sequences in rainfall records have been pointed out from time to time, and the number of different periods in which re- currences have been observed is extraordinarily large, varying from a few weeks to 41 years. 2 For the most part they have proved to be either temporary or local. The problem of detecting and measuring weather recurrences is now being attacked in a systematic 1 See Q.J.R. Met. Soc., vol. xxxix, p. 29. 2 J?or further particulars of work on this fascinating subject, the reader is referred to a series of papers by Professor H. H. Turner, F.R.S. (see Q.J.R. Met. Soc., vol. xxxvii, p. 209 ; vol. xli, p. 315 ; vol. xlii, p. 163). ANNUAL FLUCTUATIONS 229 manner by harmonic analysis with Schuster's periodigram, a method which yields results of the utmost interest. It is to be noted that in cases where periodicities have been established their utility in foreseeing the future is limited by the fact that their occurrence is of too general a character to be applied with certainty to individual localities or definite epochs. The principal value of this line of research is to link up meteorological events or sequences with collateral extra-terrestrial pheno- mena, by which means a knowledge may be gained of their relationships, if any. One of these extra-terrestrial phenomena to which attention has been given is the sunspot cycle, as suggested on p. 228. Prof. Turner has recently pointed out also that there is a well-marked correlation between rainfall and temperature variations and the periodical movements of the earth's axis. With more especial reference to the rainfall of the British Isles, a method of approaching the subject of annual rainfall fluctuations which has not yet received the attention it appears to merit is that of concentrating rather upon the distribution than upon the general excess or deficiency of the fall. Since 1906 maps have been published annually in British Rainfall showing the regional distribution of each year's rainfall as a percentage of the average. The data from which similar maps might be constructed for earlier years are available. The first and most striking fact which appears from the examination of these maps and data is that it is extremely rare for the total rainfall of any year to be above or below the average in all parts of the country at the same time. In most years some districts exhibit pronounced excess and others 2 3 o RAINFALL OF THE BRITISH ISLES pronounced defect, and in every year the geographical range of the deviation is of the order of 30 or 40 per cent, of the average. The distribution of the deviation is sometimes of great interest. In some years it indicates a general excess in the west and deficiency in the east, suggesting an unusual preva- lence of orographical rain. In these years the winter months will be found to have been wetter relatively to the average than the summer months. In other years the rainfall will be found to have exceeded the average by a wider percentage margin in the east than the west, this excess having occurred principally in the summer. These are obviously years with a relative excess of cyclonic or thunder- storm rain. In two years exhibiting these contrasted conditions the general rainfall may be* the same, whereas the circumstances which determined it were obviously different. A point of great interest which has been observed in studying these annual percentage maps is the frequency with which parallel belts of relatively high rainfall occur separated by tracts of relatively low fall. These belts invariably extend from south- west to north-east, i.e. in a direction parallel to the track of the prevailing rain-bearing winds. It is clear that they represent the lines of highest frequency of low-pressure systems or the predomin- ating lines of convergence in the air-drift. Wet belts occasionally occur in the same localities in several successive years, giving rise locally to runs of high annual values, but more often they shift somewhat in position from year to year. The result of these geographical variations in the locale of high rainfall is to introduce an irregularity of incidence into the sequence of values for an ANNUAL FLUCTUATIONS . 231 individual station which effectually masks any periodicity which may be in operation. The method of research which the above consider- ations suggest can only be successfully prosecuted, on the lines hitherto attempted, if a large and well-distributed set of records of rainfall can be referred to, covering a period sufficiently long to yield trustworthy and comparable average values, and until recent years the number available was inadequate. When this has been improved and the number of years extended, further generaliza- tions should be possible which will point the w r ay to the next stage. It appears, however, already to have been established that any investigation which aims at the elucidation of the causes of the periodical rainfall fluctuations of the British Isles must primarily involve an inquiry into the causes which operate to bring about variations in the intensity and position of the North Atlantic wind- drift and of the general atmospheric circulations of which it forms a component part. CHAPTER XIV THE RELATION OF RAINFALL TO CONFIGURATION ALMOST the first broad generalization which emerged from the systematic study of rainfall observations, and one which indeed might almost be said to have been formulated before it could be confirmed by scientific observation, so obvious was it, was that, as a general rule the amount of precipita- tion increases wdth elevation above sea-level. This statement has been found to hold good, though with some important modifications and exceptions, for all parts of the world, and it is true in a very special sense for the British Isles. Coming to closer quarters with the problem, however, it soon becomes evident that, although in the long run the relationship between land elevation and rainfall almost always exists, its numerical expression is not always the same. With close study the apparent exceptions and anomalies prove, as is invariably the case with natural phenomena, to throw more light on the true nature of the problem than the instances which follow the rule. It will be clear from the preceding chapters that on any individual day almost any conceivable rain- fall distribution may occur, and the orographical features of the land surface may or may not have any association with it. It is only when longer periods are dealt with that the relative frequency of the 23? RAINFALL AND CONFIGURATION 233 different types begins to tell. Since 1908 there have been published in British Rainfall, maps showing the distribution of the total rainfall of each month over the British Isles. A cursory examination of each map is sufficient to indicate whether on the whole the rainfall distribution was dictated by the land configuration or not, that is, whether the orographical rains predominated over the con- vectional and cyclonic rainfalls in combination. In a large number of cases there is quite distinct evidence of the occurrence of considerable orographical rain during the month, but insufficient to make it the dominating factor in the distribution. In the following summary these cases are placed in a separate class : CLASSIFICATION OF MONTHLY RAINFALL DISTRIBUTION, 1908-1919 Number of instances. Jan. Feb. Mar. Apr. May. June. July. Aug. Sep. Oct. Nov. Dec. Well - marked oro- graphical influence T2 TT TO TO 4 ! 6 2 7 6 12 12 Slightly-marked oro- graphical influence o I I I 7 4 IO 3 4 5 O O No apparent oro- graphical influence o O I I i 2 O o i i O It should be noted, in the first place, that in the four winter months, January, February, November, and December, well-marked orographical influence occurred on all but one occasion. These months form the period when south-westerly winds are at their greatest frequency and greatest average velocity. Convectional rains are probably completely absent, and cyclonic rains, though not infrequent, are 234 RAINFALL OF THE BRITISH ISLES commonly less heavy than in summer, and seldom occur except in association with orographical rain. In the period available for consideration an almost equally great preponderance of orographical rains occurred in March and April, but this may not be normal. During the remaining six months, May to October inclusive, orographical influence was apparent in 67 out of the 72 months, but in 33 cases it was not well marked, leaving 34 months, or 50 per cent., with well-marked influence, and it may be remarked that even these were seldom so definitely characteristic as was the case in the winter. July, in 10 out of the 12 years, exhibited a weak relation- ship between the rainfall distribution and the configuration. This is the month of greatest fre- quency of convectional rains, and purely cyclonic rains are also frequent, whilst orographical rains are at a minimum. In the whole twelve years there were only 7 months, or 5 per cent, of the total number, which exhibited a rainfall distribution wholly devoid of orographical influence. Of these, two occurred in June and one each in March, April, May, September, and October. The period is not long enough to indicate the normal seasonal distribution, but sufficient to make it clear that the winter rainfall is most strongly influenced by the configuration of the land, its distribution even over so short a period as one month being practically entirely determined by it ; but during the summer half-year, whilst oro- graphical influence is nearly always apparent, in a large number of cases it is not overwhelmingly strong, and in a small percentage of months it is entirely wanting. An example of prominent orographical distribu- tion during a winter month is given in Fig. 116. RAINFALL AND CONFIGURATION 235 This month, February 1914, was one of persistent south-westerly winds with frequent depressions FIG. 1 1 6. MONTHLY RAINFALL DISTRIBUTION OROGRAPHICAL TYPE. travelling on unusually northerly tracks. The excess of rainfall on the high land in the west, and in a less degree in the south, is well brought out, as is 236 RAINFALL OF THE BRITISH ISLES also the marked decrease in the fall towards the east coast and in the sheltered estuaries in the west. FIG. Iiy. MONTHLY RAINFALL DISTRIBUTION, SHOWING WEAK OROGRAPHICAL EFFECT. Fig. 117 gives an instance of weak orographical influence, the distribution having in this case been dominated by thunderstorms in England and RAINFALL AND CONFIGURATION 237 cyclonic rains in the east of Scotland. In the hilly regions of the west, especially in the north-west FIG. 1 1 8. MONTHLY RAINFALL DISTRIBUTION WHOLLY DEVOID OF OROGRAPHICAL EFFECT. Highlands, there is, however, a distinct tendency for the isohyetal lines to fall into the positions typical of 238 RAINFALL OF THE BRITISH ISLES orographical rain, so that the month's distribution as a whole is of an intermediate type. Among the few cases in which orographical in- fluence is so weak as to be completely overwhelmed by rain of other types, no better example is likely to be found than that which occurred in June 1903. Mention has already been made of the great cyclonic rains in the Thames Valley during this month (see pp. 160-161, ante). In the south of the British Isles these rains formed so large a proportion of the whole fall of the month in question that they ' completely dominate the map, and any orographical rain which may have occurred, at any rate south of lat. 54 N., is not in the least apparent in the run of the isohyets. A partial inversion of the normal orographical distribution sometimes occurs, especially in the spring months, when, on account of abnormal pressure-conditions, easterly or north-easterly winds persist for any long period. Such winds may be derived from the southern flanks of an anticyclone lying to the north, in which case they usually bring dry, cold weather. In some circumstances, how- ever, easterly winds are associated with considerable rainfall in the east of Great Britain and deficiency in the west. It appears probable that such rains may be attributed to convergence of the easterly drifts on the north side of southerly low-pressure areas with those of the anticyclone. If the duration of any protracted spell of such conditions happens to concide more or less with the calendar month, a distribution like that shown in Fig. 119 results. In Fig. 119 the isopleths indicate the percentage relation of the rainfall in various parts of the country to the average of 35 years, and bring RAINFALL AND CONFIGURATION 239 out the pronounced concentration of the rainfall in the east as compared with the west. Although MARCH, 1916. j PER CENT. OF AVERAG FIG. Iig. MONTHLY RAINFALL IN RELATION TO THE AVERAGE. THE EASTERLY TYPE. the type of distribution is entirely different in this case from the common orographical type associated 240 RAINFALL OF THE BRITISH ISLES with normal south-west winds, and shows no strongly marked relation to the configuration, it is probable that, physically speaking, the rain is allied to the orographical as well as to the cyclonic and must be regarded as an intermediate variety. If the period considered is longer, say one year instead of one month, orographical rainfall of the south-westerly type invariably preponderates over all other types. From the records brought together by the British Rainfall Organization, a consecutive series of annual rainfall maps has been constructed going back to 1865, the majority of which have not yet been published. On examining these maps one finds a strong resemblance in the type of distribution throughout the whole series, and a marked corre- lation between the rainfall and the elevation of the land, giving abundant proof of the truth of the above statement. Figs. 120 and 121, representing respectively the distribution of rainfall in 1903 and 1887, nearly, if not quite, the wettest and driest years of the whole period, show that even in the most extreme variations in the amount of the total rainfall in the year the run of the isohyets does not vary greatly, a line of one denomination in one map merely taking the place of one of another denomi- nation in the other. Whilst this tendency is indisputable so far as the type of distribution is concerned, the appearance of similarity between one map and another is probably greater than the reality, being enhanced by the disproportion between the rainfall totals in the rainy regions and those in the districts of smaller fall. It must not, therefore, be supposed that one year's rainfall observations in any sense fix the true normal distribution. An individual year's rainfall may at any place fall as much as 40 per RAINFALL AND CONFIGURATION 241 cent, below or rise as much as 50 per cent, above the average, and the amount of abnormality varies FIG. 120. DISTRIBUTION OF ANNUAL RAINFALL. WET YEAR. largely from one part of the country to another every year. Thus the maps of the relation of each year's 16 242 RAINFALL OF THE BRITISH ISLES rainfall to the average, published yearly in British Rainfall, since 1905, show that it is not at all in- FIG. 121. DISTRIBUTION OF ANNUAL RAINFALL. DRY YEAR. frequent for an excess of 20 per cent, in the year's rainfall in one part of the country to occur in the RAINFALL AND CONFIGURATION 243 same year with a deficiency of 20 per cent, in another part, and a range of over 30 per cent, of the average from place to place in the same year may be expected every year, though not, of course, at adjacent stations. Fig. 122 gives an example of the distribution of a year's rainfall in relation to the average of 35 years. On account of the relatively small number of stations for which records over so long a period as 35 years are available, it is possible that the map cannot be accepted as so minutely accurate as an ordinary isohyetal map, where every record can be utilized, but there is no doubt that isopleths indicating percentage variation from an average, if they refer to any period as long as, or longer than, a month, are almost certainly to be expected to be much smoother than isohyetal lines, and it is prob- able that the addition of a great many more values in plotting the map would introduce very little change in the run of the lines. The year chosen as an example was on the whole wet, though small areas in England and a fringe of the north and west coasts of Scotland had less than their average rainfall. In the tinted area each line indicates a successive departure of 10 per cent, from the average, and the tint is made darker as the fall is shown to be greater relatively to the average. If the year had been one of abnormal preponderance of orographical rains, it is clear that these areas of excess would have been found in the mountain districts, and the fact that this is not the case may be taken to indicate that orographical rains were less prominent than usual in this particular year, and that non- orographical rains of more than ordinary magnitude or frequency occurred in three large areas respectively 244 RAINFALL OF THE BRITISH ISLES RAINFALL 1916. PER CENT. OF AVERAGE. FIG. 122. EXAMPLE OF RELATION OF ANNUAL RAINFALL TO THE AVERAGE. in the centre of Ireland, the east of Scotland, and the south-east of England. The fact that the relatively wet areas indicated are large and homogeneous suggests that the non-orographical rains in question RAINFALL AND CONFIGURATION 245 were cyclonic rather than convectional, since the latter, as has been shown, give rise to falls of a very local nature. An examination of the daily and monthly rainfall maps for the year 1916 bears out the correctness of this inference. 1 It is thus clear that whilst orographical rainfall always preponderates over other types in the course of a year, the extent to which it preponderates is variable. The range of this variability diminishes if, instead of studying the rainfall of a single year, we consider the average during a number of years, and the average of 30 to 40 years z probably gives nearly a constant for amount, and consequently for distribution also. Since 1900 a very large number of maps of the average rainfall for 35 years in various parts of the British Isles have been constructed. Some of these have been published in the Water-Supply Memoirs of the Geological Survey, 3 and a few in other publications, but the majority were constructed for the purpose of private Bills for water-supply or water-power schemes, and are still unpublished. It is not possible in a book of this size to repro- duce any large-scale maps of this nature, but a fragment of the detail is given as an illustration (Fig. 125, p. 277) in Chapter XV, and a greatly reduced copy of the isohyetal lines for the whole country 1 See British Rainfall, 1916. 2 See A. R. Binnie, Proc. Inst. C. E., vol. cix, pt. iii, p. 89. H. R. Mill, Proc. Inst. C. E., vol. civ, pt. i ; British Rainfall, 1910, p. [143]. See also ante, p. 222-225. 3 Geological Survey of England and Wales. Water-supply Memoirs for Bedfordshire, Essex, Hampshire, Kent, Lincolnshire, Northamptonshire, Nottinghamshire, Oxfordshire, Suffolk, Surrey, Sussex, Yorkshire (East Riding), published under the authority of the Board of Education. 246 RAINFALL OF THE BRITISH ISLES FIG. 123. DISTRIBUTION OF AVERAGE ANNUAL RAINFALL. partly generalized is shown in Fig. 123. In spite of its small scale, this map will repay close study in conjunction with a physical map. The large-scale_average maps and the data upon RAINFALL AND CONFIGURATION 247 which they were based form a veritable mine of information on the subject of the relation of average annual rainfall to the configuration of the land, and it is instructive to state a few of the generalizations which appear to be justified by their study. The field covered is so wide that it is, however, manifestly impossible to give more than a few selected examples, which must be taken rather as illustrative of the underlying principles than in anyway comprehensive. In all that follows the expression " prevailing wind " must be taken to mean a wind from some quarter between south and west. The words " windward " and " leeward " must be construed in the same sense. The records quoted are in every case the average annual rainfall at the station in question for a period of 30, 35, or 40 years, in the great majority of cases 35 years, and they may, as a rule, be taken as representing the average of a very much longer period within about 3 per cent. In cases where no complete record of the required duration was available, the average of a shorter period has been corrected by means of one or more adjacent long records, so that all figures quoted may be re- garded as comparable. 1 RAINFALL AT OR NEAR SEA-LEVEL No satisfactory rainfall records exist of the -average amount of rain which falls annually over the sea itself in the neighbourhood of the British Isles. If we could obtain trustworthy information on this point we should presumably have a measure of the amount of non-orographical rainfall which falls over the land 1 Reference to large-scale contoured maps is necessary in order to follow the discussion in detail. 248 RAINFALL OF THE BRITISH ISLES generally, since, except possibly quite close to the shore, no orographical rainfall could occur over the sea. There are, however, numerous rainfall records at stations at or near sea-level, and it is instructive to examine the variations of average rainfall under these conditions. On the east coast of Great Britain the smallest average annual rainfall invariably occurs on the shores of the sheltered estuaries. Station. County. Altitude. Average annual rainfall. ft. in. Gravesend Kent 24 20-4 Shoeburyness . Essex 13 19-3 Boston .... Lincoln ii 23-4 Fearn .... Inverness 95 23-2 There is a small but perfectly distinct increase in the rainfall at similar elevations at less sheltered stations on the coast. Station. County. Altitude. Average annual rainfall. ft. in. Walmer Kent 20 25-9 Southwolcl Suffolk 44 24-1 Hornsea Yorks, E.R, 30 25-8 Wick * . . Caithness 81 29-9 The only other conspicuous variation near sea- level on the east coast appears to be a general increase from south to north for stations with similar exposures. This fact is possibly connected with the decrease of temperature in higher latitudes, but it may also be associated with the greater frequency of RAINFALL AND CONFIGURATION 249 the passage of cyclonic depressions in the north of the British Isles than in the south. On the east coast of Ireland the sea-level rainfall values are higher than on the east coast of Great Britain. Station. County. Altitude. Average annual rainfall. ft. in. Courtown Wexford Co 35' Dublin . ... Dublin 54 27-7 Greenore Louth 20 32-3 Donaghadee . Down 4 3i-7 On the south coast of England the average annual rainfall at sea-level exhibits a decided increase from east to west. Station. County. Altitude. Average annual rainfall. ft. in. Dymchurch Kent 12 25-4 Bognor . Sussex 15 26-0 Ventnor Hants, I. of W. 81 29-2 Wareham Dorset 18 31-4 Dcvonport Devon 20 3<>-7 There are, however, conspicuous exceptions to this otherwise nearly uniform increase, relatively smaller falls being observed at sheltered stations Station. County. Altitude. Average annual rainfall. ft. in. Emsworth (Thorney) Hants 23 26-7 Exmouth Devon 51 27-7 and relatively larger falls at one or two exceptionally 250 RAINFALL OF THE BRITISH ISLES placed stations. In the case of the two quoted the larger fall is undoubtedly connected with the proximity of high land. Station. County. Altitude. Average annual rainfall. ft. in. Hythe .... Eastbourne Kent 12 Sussex 12 28-0 3'9 The west coast of Great Britain differs in some essential respects from the east and south coasts. The most notable differences from the point of view of sea-level rainfall are the much more indented coast-line, especially in Scotland, where sea-lochs extend inland in many places to distances of 20 or 30 miles, and, over a considerable portion of the coast, the existence of high mountains within a few miles of the sea, especially in the north. The sea- level rainfall is everywhere higher than in the south and east, and exhibits a striking, though not uniform, increase from the south to north. Station. County. Altitude. Average annual rainfall. ft. in. Scilly . Cornwall 40 32-7 Barnstaple Devon 25 37' 1 Castle Malgwy ti Pembroke 30 43-7 Aberdovey Cardigan 22 37-8 Holyhead Anglesey 48 35-o Southport Lanes. 38 32-7 Barrow M 36 38-1 Whitehaven Cu nberland 21 41-7 Auchencairn Kirkcudbright 50 46-8 Eallabus (Islay) Argyll 68 48-8 Quinish (Mull) M 35 56-6 Arisaig Inverness 3 61-8 Scalpay Island M 4 74-4 Bendamph Ross -25 86-5 RAINFALL AND CONFIGURATION 251 The sea-level rainfall on the west coast of Ireland is higher for the same latitude than that in Great Britain, and shows, on the whole, the same tendency to increase from south to north. Station. County. Altitude. Average annual rainfall. ft. in. Darrynane Kerry 13 49-9 Valentia Island y t 12 56-0 Ennistymon . Killybegs Clare Donegal 37 65 4$-5 60-6 The rainfall near sea-level at places situated on the shores of sheltered estuaries and sea-lochs on the west coasts shows wide variations from that at more exposed coast-stations. The reasons for these variations involve a consideration of circumstances which will be more conveniently dealt with later. Except in the case of certain peculiarly situated spots, the average annual rainfall at inland stations at or near sea-level does not differ appreciably from that on the nearest coast ; the tendency being on the whole for it to be slightly lower. The most conspicuous low-lying plain in the British Isles is the Fen district, over the whole of which the average annual rainfall is 'less than 25 inches, falling to 22-1 inches at Cambridge (35 feet) and 23-0 inches at March (10 feet). A consideration of the facts enumerated above appears to justify the conclusion that even at elevations only a few feet above sea-level considerable orographical influence affects the amount of rainfall. This is shown by the pronouncedly larger falls on the west coasts, and slightly larger falls on the south coasts than on the east, an effect which seems only 252 RAINFALL OF THE BRITISH ISLES possible on the hypothesis that the rain-bearing westerly and south-westerly winds begin to deposit their moisture freely immediately on coming into contact with the land. It is not impossible that some allowance must be made for the frictional effect of the land surface in impeding the free movement of wind. It is well known that wind- velocities are normally higher over the sea than the land owing to this cause, and the general result of such interference has been pictured as the formation of a sort of cushion of relatively slow-moving air over the land above which the wind moving from the sea would tend to mount, giving rise to ascending currents and consequent rainfall. After passing across the British Isles the available moisture must be diminished in an important degree, accounting for the relatively small rainfall on the east coast. Such orographical rain as is derived from the North Sea and carried by easterly or north- easterly winds is certainly very much smaller in quantity than that from the south-west quarter, but it is probably to this effect that the relatively high rainfall on the elevated land in eastern Aberdeenshire, the North and East Ridings of Yorkshire, and Norfolk is due. The diminished fall on the shores of estuaries on the east and to some extent on the south coast is ob- viously due to shelter from rain-bearing winds. It will be observed that the fall is smallest when the shelter is on the south or west side. A similar diminution is to be noted in sheltered estuaries on the west coast : Station. County. Altitude. Average annual rainfall. Minehead Burnham Somerset ft. 50 18 in. 31-0 RAINFALL AND CONFIGURATION 253 In these cases the shelter is distinctly on the south- west. The following are exposed to the south and west and sheltered on the north and east : Station. County. Altitude. Average annualrainfall. Morecambe . - . Westport * * ; Lanes. Mayo ft. 35 in. /}0-O 46-2 THE INCREASE OF RAINFALL ON WINDWARD SLOPES Except in a few special cases the average rainfall increases with increased altitude, but the rate of increase per 100 feet varies within wide limits. It is of interest to study the conditions under which such variations occur. We may first take the simple case of rising land facing the sea, with a gradient approximately parallel to the track of the prevailing wind. Such conditions occur along the slopes of the South Downs in Sussex and part of the eastern extremity of the North Downs in Kent : Station. County. Altitude. Average annual rainfall. Increase per 100 ft. from base station. ft. in. in. Bognor . Sussex 15 26'0 Eartham 230 29-5 1*6 Selhurst 300 32-21 2-2 Bepton . ,, 554 37' 2-0 Brighton Sussex 32 28-7 Patcham 207 3i-8 1-8 Pyecombe 392 35'6 1-9 Hythe . Kent 12 28-0 Paddlesworth 612 38-2 1-7 254 RAINFALL OF THE BRITISH ISLES The above seem to suggest a fairly uniform increase of from 1-5 to 2-0 inches per 100 feet, but the rate may be reduced to a smaller figure if the hill is less continuous, for example : Station. County. Altitude. Average annual rainfall. Increase per 100 ft. from base station. ft. in. in. Birling Gap . Sussex 4 26-9 Willingdon 599 3*'4 o-S Portslade Sussex 167 29-O Poynings , 680 34-3 I-O The smaller rate of increase in these cases may be attributed to the facility with which the wind may find its way round an isolated hill rather than being forced to ascend over the summit. A still more remarkable instance will be mentioned later. For exactly the same reason it is usual to find the rainfall lower, for the same elevation, on the slopes parallel to the prevailing wind at the points where the South Downs are intersected by gaps than on the con- tinuous hillsides at right angles to the prevailing winds, that is to say, each isohyet rises to a higher elevation in the gaps than on the exposed slopes. Comparison may be made with similar data for the inland ridges of the south-eastern countries : Station. County. Altitude. Average annual rainfall. Increase per 100 ft. from base station. ft- in. in- Cranleigh . Surrey 175 29-2 Malquoits . . M 400 31-4 I-O Ewhurst . ... " 600 3^-2 0-7 RAINFALL AND CONFIGURATION 255 Station. County. Altitude. Average annual rainfall. Increase per 100 ft. from base station. ft. in. in. Edenbridge Kent 161 25-9 Ide Hill / 700 30-7 0-9 Botley Hill . " 870 33-5 i-i Maidstonc Kent 30 2 4 -2 Detling . t 9 336 2 7 -3 T *2 Harrietsham . 620 31-5 1-2 It will be observed that the rate of increase is smaller than that on the South Downs, an effect probably due to the diminished humidity of the air, and the more pronounced the shelter afforded by the more southerly ridges the smaller does the increase per 100 feet become. An example is given by the slight rate of increase on the southern side of the Forest Ridges in north Sussex : . Station. County. Altitude. Average annual rainfall. Increase per 100 ft. from base station. ft. in. in. Buxted Park . Sussex 94 30-7 Maresfield 247 31-2 o-3 Nutley . 336 31-6 -3 Crowborough . >, 777 35-i 0-6 Warbleton Sussex 182 3i-4 Tottingworth " 500 32-4 o-3 The falling-off of total rainfall at the same altitude with increasing distance from the sea, coupled with the shelter of the intervening hills, is illustrated by the following. The distance from the sea is measured approximately in a south-westerly direc- tion : 256 RAINFALL OF THE BRITISH ISLES Station. County. Distance from sea. Altitude. Average annual rainfall. miles. ft. in. Bepton . Sussex 15 554 37' Worth . 25 558 34'4 Chipstead Surrey 40 550 30-5 Ash Kent 50 540 27-0 The rates of increase above cited do not cover the whole range of variation, and one interesting anomaly may be mentioned. In the case of fairly steep ridges close to the sea it is commonly found that the maximum rainfall occurs slightly on the leeward side of the crest, irrespective of the altitude of the spot on which it falls. This phenomenon often gives to an individual station, which happens to be so situated, a rainfall greatly in excess of that otherwise proper to its altitude. An example may be seen in the South Downs at Lavington, immediately to the north of the highest part of the Downs, which rise to over 800 feet, but itself at an altitude of only about 250 feet. The average rainfall is 40- 1 inches, probably the largest observed in any part of the south-east of England. Similarly at Alciston, near Polegate, Sussex, 33-9 inches falls at an altitude of only 172 feet, whereas at Willingdon, only 6 miles distant and situated 599 f ee "t above the sea, only 31-4 inches falls. It is clear from a general survey of the distribution of rainfall in the south-east of England, that the average fall is wholly governed, in the first place by the capacity of the land in the district to produce an uplifting effect on the prevailing winds, and, secondly, by the amount of condensable moisture remaining available in these winds after passing over the land 'between the station and the sea. RAINFALL AND CONFIGURATION 257 The conditions which obtain in the south-east of England hold good in the west also, but in this case we have to deal with not only a very much larger sea-level rainfall, but very much more elevated and continuous hill-slopes. A fairly good example may be found in central Wales : Station. County. Altitude. Average annual rainfall. Increase per 100 ft. from base station. ft. in. in. Aberystwyth . Cardigan 15 35-6 . Goginan M 290 46-2 3-9 Cwmsymlog . 800 55-5 2-5 Waenbwll jt 1,380 67-4 2-3 Plynlimon " 1,74 94-0 3'4 Aberaeron Cardigan 50 35-6 Abermeurig . >f 300 46-0 4'2 Tregaron >t 520 50-0 3-1 Maes-y-bettws it 910 66-6 3-6 Towy-fechan . t> 1,330 72-1 2-8 Cors-yr-hwch . Radnor *.775 87-5 3-o There is a fairly close agreement in the rate of increase in these cases, which, it will be noted, is about double that observed on the South Downs. With the steeper gradients of the Snowdon range the rate of increase is still further augmented : Station. County. Altitude. Average annual rainfall. Increase per 100 it. from base station. ft. in. in. Glynllivon Carnarvon 100 44-1 Nantlle . 450 61-2 4'9 Llanllyfni 751 73-9 4'6 Cwmsilian Lake I,IOO 81-0 3-7 Moel Hebog \ 1,50 II2-2 4'9 Lluchfa , ... ' 2,500 161-0 4.9 The last-named station is one of a group situated 17 258 RAINFALL OF THE BRITISH ISLES in the vicinity of Llyn Llydau on the immediate east of the great escarpment of Snowdon, and presents an undoubted example of the phenomenon already referred to, viz. the occurrence of the maximum rainfall immediately on the lee side of the highest land. If we regard the rainfall at Lluchfa, which is fairly representative of Cwm Llydau, as the result of the obstruction to the prevailing wind caused by a ridge of 3,500 feet altitude, the rate per 100 feet works out at 3-4 inches. The English Lake District presents a parallel instance to the Snowdon group, but the seaward slopes of the Lake District mountains are singularly devoid of rainfall stations at representative altitudes, the great majority being situated in the deep-cut valleys which radiate from the centre towards the sea. We may, however, quote : Station. County. Altitude. Average annual rainfall. Increase per 100 ft. from base station. Ravenglass Sea Fell Pike . Cumberland ft. 80 3,200 in. 42-3 137-8 in. 3'i Sprinkling Tarn 1,985 I28-I 4'5 The Stye 1,070 176-9 13-6 The last-mentioned is another famous example of the shifting of the maximum rainfall to the leeward, and if we regard the mass of the Great Gable, rising to 2,900 feet, as the controlling factor, we get an increase of 4-8 inches per 100 feet from sea-level, showing a fair agreement instead of the very large and certainly deceptive value of 13-6 inches shown by calculating from the station elevation. RAINFALL AND CONFIGURATION 259 The rate of increase of rainfall per 100 feet on the windward slopes of the Helvellyn and High Street ridges, which lie in the direct shelter of the central Lake District mountains, is in very striking contrast to the high values quoted above : Station. County. Altitude. Average annual rainfall. Increase per 100 ft. from base station. ft. in. in. Dale Head Hall Cumberland 620 78-5 Whiteside 2,100 85-0 0-4 Grasmere Westmorland 553 83-3 Fairfield >. 2,860 95-5 0-4 Ambleside Westmorland 1 80 76-2 , Kirkstone it 1,500 98-3 1-7 Grey Crag " 1,750 98-0 1-4 Too much weight must not be given to the exceptional values at Whiteside and Fairfield, on account of the uncertainty which always attaches to rain records on high and exposed moorlands. Except where westerly winds find free access the rate of increase on the western slopes of the Pennines is small. At the point selected as an example the Pennines are sheltered by the Snowdon range, and the conditions may be taken as repre- sentative of the less exposed parts of the ridge : Station. County. Altitude. Average annual rainfall. Increase per 100 ft. from base station. ft. in. in. Chelford Cheshire 250 29-7 Macclesfield . 501 34'0 1-7 Leek . Stafford 750 38-2 Axe Edge Derby 1, 600 1-9 260 RAINFALL OF THE BRITISH ISLES The changes in the rate of increase of rainfall per loofeet undervarying conditions of shelter, gradient, and distance from the sea are well brought out in the cross-section through the Lake District and Pennines in Fig. 124. It is significant to note the ready response of the rainfall curve to the changes in the altitude curve in the west, and the diminishing effect of high land to the leeward of the most westerly hill- barriers. This diminution in the amplitude of the rainfall curve is particularly noticeable on the Pennines, where the slopes of Marton Fell and Cross Fell, directly facing the prevailing wind, experience a rainfall very much inferior to that of the Lake District. In order to illustrate the absolute dependence of the average rainfall upon the capacity of the land to exercise an uplifting effect upon the prevailing wind, it is of interest to refer to a single example of low rate of increase which may be explained upon physical grounds. The hill in question, Pendle Hill, rises to a height of 1,800 feet, and enjoys an almost uninterrupted exposure to the rain-bearing winds blowing up the Ribble Valley. Records on the south- western slopes give the following values : Station. County. Altitude. Average annual rainfall. Increase per 100 ft. from base station. ft. in. in. Gawthorpe Lancashire 3 I6 41-4 Sabden . n 500 43-3 I-O Ogden Reservoir " 935' 46-2 0-8 The shape of Pendle Hill, which is an isolated mass, may be compared to that of an overturned boat, with the bows pointing to the south-west. On this 261 262 RAINFALL OF THE BRITISH ISLES account the wind coming up the Ribble Valley is able without hindrance to find its way past the rising land without being forced to ascend over the summit, except possibly to a limited extent. The case of Pendle Hill is no doubt exceptional, but it is certain that, generally speaking, the rate of increase of rainfall per loo feet on hillsides parallel to the prevailing wind is lower than it is in the case of slopes directly facing the wind. This tendency is greater according as the hill is more isolated. This fact has already been mentioned in dealing with the rainfall of the South Downs, but the dis- parity which it brings about is very much larger in the west. An interesting corollary in the case of Pendle Hill is the fact that the rainfall of the valleys lying immediately to the north-west, north-east, and south-east of the ridge appear to have an average rainfall rather higher than might normally be expected, no doubt owing to the convergence of wind-currents deflected from its slopes. DECREASE OF RAINFALL ON LEEWARD SLOPES On the leeward slopes of high land, save in the exceptional cases already referred to, the amount of rainfall diminishes steadily as the prevailing winds gradually descend to lower levels. If these winds have passed over land with an extremely high rain- fall, they will be largely drained of their available moisture ; whereas if they have parted with only a moderate proportion of their water-content, the residue available for precipitation on the leeward slopes will be the greater. It should be noted that descending air is being steadily warmed by the increasing: pressure above, so that its capacity for RAINFALL AND CONFIGURATION 263 retaining moisture is increasing. The tendency for rain to condense is therefore proportionally less than was the case at equal levels on the windward slopes. In the case of the Pennines it is convenient to state the rate of decrease of rainfall per 100 feet in general terms rather than by quoting individual records. From a number of measurements the rate of falling-off in annual rainfall is found to be, broadly speaking, I inch per 100 feet above 1,000 feet, increasing to 2 inches per 100 feet below 500 feet of altitude, measuring in each case from the summit of the ridge. The rate is higher in the districts where the Pennine range is exposed on its western side tothefull effect of the south-west wind. It will be observed that in the case of the Pennines the rate of decrease is more rapid at low levels, on the extreme east, than at high levels. This, no doubt, arises from the cumulative effect of continued desiccation of the air. On the eastern slopes of the Snowdon group of mountains the decrease per 100 feet is larger andmore irregular than on the Pennines. The irregularity shown is possibly due to the difficulty of selecting re- presentative stations uninfluenced by local conditions. Station. County. Altitude. Average annual rainfall. Decrease per 100 ft. from highest station. ft. in. in. Upper Eigiau Carnarvon 2,000 121-5 . Caedryn 1,250 107-2 1-9 Lake Cowlyd . 1,168 86-2 4-2 Lower Brwynog I,OOO 66-8 5'5 Dolgarrog 22 5 '2 3-6 Llyn Dulyn . Carr arvon 1,632 103-6 Eigiau Dam 1,245 84-9 4-8 Frith Rhos . I,IOO 7 J '3 O-I Llanbedr . . ' . 5 10 56-2 4-2 Tal-y-cafn 75 44-2 3'8 264 RAINFALL OF THE BRITISH ISLES Similar high rates obtain to the east of the Plynlimon range : Station. County. Altitude. Average annual rainfall. decrease per 100 ft. rom highest station. ft. in. in. Plynlimon Cardigan 1,74 94-0 Dylive . tt 1,300 71-0 5'2 Carno .. 595 49-9 3-8 Plynlimon Cardigan 1,740 94-0 Pantmawr Montgomery i, 080 65-1 4'4 Dernol . " 850 56-1 4-3 The rainfall at the base of a mountain range is commonly lower on the leeward side than on the windward, and the diminution is intensified in the case of land to the leeward of exceptionally wet districts, giving rise to the phenomenon sometimes known as " rain shadow." A good example is seen in the lower Spean Valley, immediately to the north- east of Ben Nevis, where an average of only 53-7 inches falls annually at Roy Bridge (307 feet). In the unsheltered Lochy Valley a few miles west, at a lower altitude, about 70 inches falls. In the Eden Valley, sheltered by the mountains of the Lake District, the average rainfall is as low as 34-4 inches at Appleby (440 feet) and 32-2 inches at Carlisle (115 feet), compared with about 50 inches at the same altitudes on the south-west of the Lakes (see Fig. 124). A similar effect is seen in Glengarry, near Blair Atholl, Perthshire (420 feet) ; in the Clwyd Valley in North Wales; and on a larger scale in the low rainfall of the Vale of York, and of the central basin of the Shannon. RAINFALL AND CONFIGURATION 265 RAINFALL IN NARROW VALLEYS Many instances are to be found, particularly in the West Highlands, of stations at low elevations, in some cases nearly at sea-level, with rainfall values very much higher than would seem to be appropriate to their altitudes. We may mention, for example : Station. County. Altitude. Average annual rainfall. ft. in. Arrochar . Dumbarton 5 QI-8 Ardlui . . ' . 5 II3-2 Glencoe Argyll 20 84-7 Fort William Inverness 33 75-9 Glenfinnan 50 108-8 Bendamph Ross 25 86-5 An examination of a large-scale contour map shows that these stations are situated, without any excep- tion, in the bottoms of steep-sided valleys of no great width, between lofty hills, obviously carrying a high rainfall. The conclusion is justified that narrow valleys, and, we may add, particularly those transverse to the direction of the prevailing wind, partake of the rainfall of the surrounding hills. This fact alone renders it impossible to compute any general formula for the increase of rainfall with elevation, and makes it necessary to study every record in the light of the configuration of the surrounding country. The degree of completeness with which the rainfall on valley floors represents that on the surrounding plateau diminishes with valleys of greater width. A case in point may be seen in the neighbourhood of Loch Lomond (see Fig. 124), where a number of .stations, about 50 feet above sea-level, on the west bank of the loch, give 266 RAINFALL OF THE BRITISH ISLES values of about no inches under the shadow of Ben Vorlich, and 90 inches under Ben Bhreac, where the loch is about half a mile wide. At Luss, 70 feet, or about the same elevation, but where the loch is 3 miles wide, only 77 inches falls, and at Balloch (91 feet), at the south end of the loch, only 50 inches. The instance may also be mentioned of Fort William, situated practically at sea-level, at the head of Loch Linnhe, and immediately at the foot of Ben Nevis, which rises to 4,407 feet on the east-south-east. The average annual fall at Fort William is 75*9 inches, and that on the summit of Ben Nevis 162-6 inches, giving an increase of 2-0 inches per loo feet. Taking into consideration the very steep gradient of the mountain between Fort William and the summit, and also the westerly position of the stations, the rate of increase is a low one, indicating that the fall at Fort William, though only about half that of the higher station, is larger than the fall proper to its altitude at that latitude. Transverse valleys are undoubtedly more apt to participate in the high rainfall of the surrounding hills than are valleys parallel to the prevailing wind, though the latter do so to some extent, particularly if very narrow. The probability of the phenomenon occurring diminishes with distance from the west coast, and it is rather a striking feature of the narrow valleys of the eastern slope of the Pennines that they carry a low rainfall far up among the hills. Instances are occasionally met with in which the phenomenon of sustained high rainfall over valleys transverse to the prevailing wind with that of the shifting of the maximum fall to the leeward of the highest land, previously referred to, combine to RAINFALL AND CONFIGURATION 267 give a rainfall actually higher over the valleys than over the hills ; this is the case in some of the deep-cut valleys of South Wales, notably in the Black Mountains of Breconshire (see Fig. 124). A point of some importance with regard to the distribution of rainfall in the neighbourhood of narrow valleys, particularly near the west coast, is that a considerable modification in the direction of the surface winds is sometimes brought about by the trend of the valley. Thus, for example, a wind entering the mouth of a valley from a south-westerly direction may pass along lateral valleys as a south- east wind, retaining, however, all the rain-bearing characteristics of a south-west wind. In these circumstances the increase of rainfall per loo feet on the hillsides up which the wind is being forced to rise will be higher than would probably be the case in a valley to which only true south-east wind could gain admittance. It must be repeated that the foregoing statements and examples must be taken only as illustrative of some of the outstanding features of the distribution of rainfall in its relation to configuration. The statements are of necessity somewhat general in character, and the examples frequently somewhat local in application. It is clear that any quantitative generalisations must only be accepted subject to the reservation that the limits of variation therefrom are of great magnitude. There appears to be no way in which to arrive at any clear conception of the physical processes underlying these variations except by the patient study of rainfall records in conjunction with large-scale contoured maps. For this purpose the 2 miles to I inch Ordnance Survey " layer " map 268 RAINFALL OF THE BRITISH ISLES of the British Isles, or its precursor of the same kind produced by Bartholomew long before the Ordnance Survey published this very convenient map, will be found an effective guide to the details of the configuration of the land. CHAPTER XV THE ECONOMIC APPLICATION OF RAINFALL DATA THE pursuit of any branch of natural science is primarily of value not because of any specific practical end to which the knowledge gained may be applied, but on account of the part it plays in the development of the human intelligence. It is a natural and laudable instinct which prompts the pursuit of knowledge for its own sake, and the enunciation of & fact, even though marred by our imperfect appreciation of its true meaning, or our inability to apply the information to our ends, is in itself an achievement worthy of the utmost effort. All scientific knowledge can, however, be utilized for economic purposes, because the more fully we understand natural laws the more surely can natural phenomena be made to serve our needs. The direct importance of information as to the laws governing the phenomenon of rain and as to its incidence and distribution is manifest when it is realized that we are entirely dependent upon rain as the only source of fresh water, of all elements the most indispensable to life. The great abundance of water in the British Isles an advantage which we owe to our geographical position renders this country immune from the devastating droughts which are common in some parts of the world, and it is comparatively rare for any serious or widespread 269 270 RAINFALL OF THE BRITISH ISLES famine to occur merely on account of lack of rain. The British Isles are also greatly favoured in this respect by the fact that, the bulk of the rainfall being orographical, the mountainous districts, which lie principally in the west, enjoy a sufficiently large and continuous rainfall to feed numerous rivers which carry away their superfluous precipitation across the relatively drier plains of the east. The con- tinual renewal of the supply of water in excess of the actual needs of vegetation allows of large accumulations in the permeable underground strata, such as chalk and sandstone, so that even in the event of long periods of rainless weather, supplies can usually be obtained from springs or borings. All applications of rainfall data to economic purposes involve a consideration of certain points beyond those dealt with hitherto. These may be grouped under three heads : (i) The interpretation of rain gauge readings in terms of volume of precipita- tion ; (ii) the losses to which the precipitated water is subject after reaching the ground; and (iii) the relation of the flow of streams and the fluctuations of underground water to the rainfall. The computation of the actual volume of rainfall from observations has been placed on a much more secure basis than formerly by the elaboration of the cartographical method of treatment. Strictly speaking, a rain gauge measurement can only be applied with certainty to the actual spot upon which the instrument is placed. If it is necessary to ascertain the volume of rainfall over an area, it is obviously necessary to know the amount of the fall in all parts of that area and the quantity which has fallen in the ungauged portions must be inferred. Considerable experience is required before this ECONOMIC APPLICATION 271 inference can be made without serious risk of error. In constructing a rainfall map for the purpose of a volumetric determination, it is important to take every precaution that the records plotted are synchronous. The plottings should hot be confined strictly to the area under consideration, but should extend to some distance on every side, since much light may be thrown on the distribution of the fall within the area from that in the adjacent districts. The method of drawing the isohyets differs considerably according to the period which is being dealt with and the nature of the country. It is frequently necessary, for example, to ascertain the volume of water precipitated during a single shower in order to relate it to the magnitude of a flood, and in such a case it is of first importance to ascertain whether the rainfall in question was of the thunderstorm type or whether it was more definitely cyclonic or orographical in origin. For this purpose pressure-maps are of great utility, but the type of rainfall can usually be adjudged by extending the area mapped until the characteristics of the distribution declare themselves. The examples of regional distribution given in Chapters X and XI will enable the reader to form an opinion as to the manner in which a skilled draughtsman may employ analogy in interpreting the measurements over any individual area in terms of a rainfall map. If the distribution shows signs of being definitely oro- graphical, the accuracy of the isohyets may be greatly improved by taking advantage of the details of configuration shown on a contoured map. The use which can be made of the orographical features increases if the period dealt with is considerable and 272 RAINFALL OF THE BRITISH ISLES if the district is definitely hilly. In maps of the average rainfall during a long term of years, such as are required for computing the average water- yield of a gathering-ground for water-supply or water- power purposes, it is imperative that every detail of the orography in its relation to the prevailing rain- bearing wind should be taken into account. In the absence of actual observations it is justifiable to direct the run of the isohyets, or to indicate hypo- thetical dry or wet areas, in the manner which analogy with better represented districts shows to be the most probable. It is important that this prin- ciple should be fully recognized in any determination of the probable yield of a catchment area which is to be utilized for the purpose of constructing water- works or installing hydro- electric plant, since any appreciable error in evaluating the volume of water which can be impounded may involve a useless expenditure of many thousands of pounds or the failure to provide for the essential requirements for domestic or industrial purposes of large centres of population. Fig. 125 gives an example of the method of pre- paring a preliminary determination of the rainfall of areas for water-supply purposes. The map in question was constructed by Dr. H. R. Mill and the author in connexion with the scheme laid before Parliament in 1913 for providing an additional 58 million gallons of water per day for the city of Glasgow. The problem which it was necessary to solve was to ascertain how much water could be abstracted from the catchment area of Loch Katrine and how much additional supply could be obtained by building a dam across the mouth of Loch Voil in the adjacent valley on the north. ECONOMIC APPLICATION 273 At the date when this scheme was prepared, the number of rainfall records available in the immediate neighbourhood of the lochs in question was not great, and a few of those records which did exist were of doubtful accuracy. It was, therefore, necessary to rely to a considerable extent on analogy and hypothetical reasoning in constructing the map. The period dealt with was the 35 years 1878 to 1912. Only a small number of the records available covered the whole of this period, and these few long records were utilized for the purpose of reducing shorter records to their equivalent for the 35 years. The method by which this is done is to ascertain from the long records the percentage of the 35 years' average which fell in any particular short period, and to apply this percentage as a correction to the average value derived from the short period. For example, the record at Corriearklet, 3 miles south of the head of Loch Katrine, covered only the 8 years 1905 to 1912, the annual average being 91-81 inches. An examination of the long records at adjacent stations showed that during these 8 years 101-8 per cent, of the average for 35 years fell, and reducing 91-81 inches in the ratio 101-8 : 100 we get 90-2 inches. All the short records were weighted in this manner before placing them on the map, for which purpose they were rounded off to the nearest inch. In the case of records which critical examination proved to be in error, the totals for the years in which errors occurred were either rejected or amended by the help of neighbouring records. In a few cases of gauges placed on exposed mountain slopes a special test was applied. In these circumstances it is reasonable to expect that the catch during the winter months may be defective, 18 ' 274 RAINFALL OF THE BRITISH ISLES owing to the action of wind. In order to ascertain whether this was so, the recorded fall during the winter half-year was tabulated separately from that during the summer half-year. The same process was followed for several other neighbouring records of known accuracy. If in the case of over-exposed gauges the percentage of catch in the winter was found to be substantially in accord with that at the standard stations, accuracy was assumed ; if, on the other hand, it was found to be deficient in com- parison with the standard stations, a special correc- tion was applied. The following table shows the application of this method : Percentage of rain in winter, half-year. Corrected values. ' Station. No. of years. Annual average. Actual (rejected). Standard (adopted). Average for period. Average for 35 years. in. in. in. Ben Lomond 35 88-26 54 59 g8-20 9 8-2 Ledard 35 59-55 48 59 74-80 74-8 Ben Ledi 6 70-14 55 60 79-47 75-9 As an additional precaution against error, the rain gauges in operation at the time of the inquiry were specially inspected and their accuracy tested, care being taken to note any peculiarities of their exposure. The whole of the actual or computed average values for 35 years were plotted on a copy of the reduced Ordnance Survey map on the scale of 2 miles to I inch, on which the configuration is indicated by tints between the contours of altitude, so that the hills and valleys are thrown into relief. Isohyetal lines were then drawn indicating the distribution ECONOMIC APPLICATION 275 of rainfall, based primarily upon the plotted values, but in respect of their direction guided also to a large extent by the configuration, giving effect to the principles laid down in the preceding chapter. Most of the rain gauges in this region lay in the valleys, and in drawing the lines over the intervening mountains the course followed was that which mountain records in other similar districts showed to be the most probable. Shortly after the construction of this map several additional rain gauges were established at selected spots in and around the Loch Katrine gathering- ground, and when the records for about 18 months were available, they were specially worked up, weighted as described, and added to the map. The computed average values for these gauges are shown on the map within rings. Their indications provided a perfect test of the accuracy of the reasoning upon which the run of the isohyets had been based. Practically without exception they fitted into their places in the scheme of lines, so that no modification was required and the map stood confirmed as originally drawn. This fact does not of course imply that a rainfall map can always be constructed with certainty from meagre data. The fewer the records available, the greater is the risk that an erroneous measurement may be inadvertently accepted as accurate, and it is highly desirable that on any map the number of points of observation should be large enough to enable an inaccurate record to be detected by its want of harmony with the remainder. The map being completed, planimeter measure- ments were made of the areas lying between each successive isohyet, and for each of the zones thus 276 RAINFALL OF T-HE BRITISH ISLES separated an appropriate general rainfall value was assigned by inspection of the map. The actual measurements are given in the following summary. Multiplying the area of each zone by its general rainfall gives the volume of rainfall in square-mile-inches, and the total of all the volumes is the volume of rain falling in an average year over the whole area. The number of square-mile-inches can be readily converted into gallons or any other volumetric unit. CALCULATION OF VPLUME OF RAINFALL Loch Voil Catchment Area Zone. Area. General rainfall. Volume of rainfall. sq. miles. inches. sq. -mile-ins. Less than 80 inches 4-0 77'5 310 80-90 I4-3 86-0 1,230 QO-IOO ,, . 14-5 94-0 1,363 IOO-ITO ,, . 5'i 101-8 519 More than 110 ,, -3 II2-O 34 Total . . : . 38-2 3-455 Loch Katrine Catchment Area Zone. Area. General rainfall. Volume of rainfall. sq. miles. inches. sq.-rnile ins. Less than 70 inches 3-8 69-0 262 70-80 ,, . 10-0 74-5 745 80-90 ,, . 14-3 83-8 i 198 90-100 ,, . 93-8 469 More than 100 ,, 3'i 103-0 319 Total .... 36-2 2.993 If it is desired to ascertain the general rainfall over the areas as a whole, this is arrived at by dividing the total volume by the total area. In the case of Loch Voil this works out at 90 inches, and for Loch 277 278 RAINFALL OF THE BRITISH ISLES Katrine 83 inches. The volume of rainfall in an average year for the two areas is 50,041 million gallons and 43,337 million gallons, respectively. The example which has been given may be regarded as typical of the most accurate method yet devised of utilizing rainfall records for the purpose of computing the volume of water precipitated over any given area in any definite period. It can be applied with the necessary modifications to a large group of engineering problems, including, besides those already enumerated, the design of sewerage and drainage works, the conservation of rivers, and the maintenance of canals. In all cases where Parliament is applied to for authority to impound any area for the purpose of abstracting water, provision is made for a certain fraction of the natural flow to be returned to the stream as so-called " compensation water," in order that its bed may not be dried up. The determina- tion of the amount of compensation water depends entirely upon the rainfall, and the accurate gauging of rainfall, therefore, becomes a matter of public interest. It is highly desirable that the life of streams should be safeguarded in this way, not only on account of milling and fishing rights, which are commonly looked after by zealous owners, but in the interest of sanitation and of preserving the natural beauty of the landscape. It appears, there- fore, to be a public duty for landowners to keep records of rainfall and for an expert public authority to undertake their supervision and interpretation. It has been shown how, within limits, the volume of rainfall can be ascertained. It is necessary, however, to bear in mind that under natural conditions only part of this volume actually perco- ECONOMIC APPLICATION 279 lates into the ground or flows away in streams. A portion is re- evaporated and lost, and in all volu- metric computations some allowance must be made on this account. In dealing with a small area it is sometimes found that in addition to the loss by evaporation, a portion of the water which percolates into the soil will also be lost. This percolated water sinks through the permeable rock and feeds springs, some of which emerge at the outcrop of the lower impermeable strata. Loss occurs on any individual catchment area if the outcrop lies beyond its limits, but the conditions under which such leakage can take place are so variable that it is impossible to generalize as to its amount. The amount of evaporation can be determined by observation by more than one method, but it is difficult to say how far the results of observations under artificial conditions can be safely applied for practical purposes. The most usual observations are those made by exposing a body of water to the air and measuring the change of level, allowance being made for rain by means of an ordinary rain gauge placed alongside. Many early observations on these lines were seriously in error because the vessel used was too small and consequently became unduly heated by the sun. The standard evaporimeter now used in this country is a tank 6 feet square and 2 feet deep, containing about 450 gallons of water, and sunk in the ground to nearly its full depth. Continuous observations have been carried on with a tank of this size at Camden Square, London, since 1885, and at several other stations for shorter periods. " The results appear to indicate that there is a range of only a few inches between the results at different stations. The amount of evaporation 280 RAINFALL OF THE BRITISH ISLES apparently depends chiefly upon temperature and sunshine. The average evaporation per year at Camden Square is 15-5 inches, the highest value recorded being 19-5 inches, or 25 per cent, above the average, in 1911, and the lowest 12-6 inches, or 19 per cent, below the average, in 1888, the fluctuations being thus markedly smaller than those of rainfall. There is a strongly accentuated seasonal variation, 87 per cent, of the total occurring during the summer half- year. A second method of arriving at the amount of evaporation is by means of percolation gauges. The gauge consists of a block of earth, usually I cubic yard in volume, enclosed in a water-tight casing open at the top. The water which passes through the gauge is collected and measured, and the difference between this amount and the total rain- fall gives an indication of the loss by evaporation. A percolation gauge is only practicable in districts where the rainfall is moderate, since in regions of high fall part of the rain runs off the surface of the ground without percolating. The results will obviously differ with soil of varying permeability, and they are thus not of general application, but the records nevertheless yield valuable supple- mentary information. The most complete series of records of this kind are those kept since 1870 at the Rothamsted experimental station of the Lawes Agricultural Trust, near Harpenden. The Rothamsted soil is described by Dr. N. H. J. Miller as a rather heavy loam, with a reddish- yellow subsoil over chalk, both containing flints. The average evaporation deduced from the ECONOMIC APPLICATION 281 Rothamsted observations is 15-3 inches per annum, ranging from 19-1 inches, or 25 per cent, excess", to 11-9 inches, or 22 per cent, deficiency, thus showing a very close agreement with the Camden Square records. The seasonal range of evaporation from soil is, however, somewhat different, as is shown by the following comparison. Jan. Feb. Mar. April. May. June. July. Aug. Sept. Oct. Nov. Dec. Year. Cam. Sq. tank . II 25 66 i-53 2-41 2-8g 3'Qi 2-31 1-33 62 25 OQ I5-5I Rothstd. soil 37 50 88 i'37 1-59 i-77 2-05 2-II 1-77 i-59 81 51 15-32 Neither method of observing reproduces precisely the conditions of nature. In the percolation gauge the water once taken out and measured cannot rise again to thesurf ace, but in natural conditions the roots of plants and especially trees draw large quantities of water from the subsoil and give it to the air by transpiration. This is especially the case when the soil is dry, whereas no evaporation can occur from a dry percolation gauge. The evaporation tank, on the other hand, errs in the opposite direction. In all circumstances it offers an unlimited supply of water for evaporation, whereas when the soil is dry the amount available must be limited. This is undoubtedly the explanation of the difference between the two results for the summer months, and it is reasonable to suppose that the truth lies between the two extremes. The differences in winter are less marked and are probably accounted for by the fact that the percolation gauge, being 282 RAINFALL OF THE BRITISH ISLES covered with grass, offers a larger evaporating surface and is usually in a moist condition at this season. In applying these results generally it is commonly assumed, with some justification, that the amount of evaporation diminishes from south to north. Over the Pennine district empirical observations of the difference between rainfall and the yield of water- supply catchment areas- suggests an annual loss of about 14 inches, and in Scotland the amount probably falls to 10 or 12 inches. Whilst the average loss by evaporation is fairly well established, the actual loss occurring during long dry periods, a matter of great importance to water-works engineers, is still uncertain, and it is desirable that more extensive observations should be made of this important factor. The actual rainfall, less an appropriate allowance for loss by evaporation, and if necessary for percola- tion, has been termed the Effective Rainfall. This represents the best estimate which can be made of the amount of water which actually finds its way into streams and rivers. The actual flow can, of course, be measured by means of weir-gauges. The relation between the flow of water in any individual river and the effective rainfall over its basin is a subject which has long engaged the attention of hydraulic engineers, and many empirical formulae have been suggested for expressing it. The con- sideration of these formulae in detail is beyond the scope of this book, and it is only necessary to make a few general observations. Assuming that the total flow is, in the long run, equal to the volume of effective rainfall, the extent to which the fluctuations in the fall will be repro- duced in the flow of a river varies within wide limits. ECONOMIC APPLICATION 283 It may be. taken as an invariable rule that the fluctuations of the run-off will be smaller than those of rainfall. The extent to which the rainfall fluctuations are smoothed out depends primarily upon the size of the catchment area, being greatest in large river valleys and smallest in mountain streams with steep gradients. The flow-off is more rapid in winter than in summer, and is greatly impeded by the presence of vegetation, particularly by forests. It has frequently been stated that denudation of forest areas has resulted in a diminished rainfall, but it appears to be highly probable that this is a fallacy, arising from the fact that in these circumstances run-off is accelerated, causing streams to be more frequently dried up in summer. An important factor in relation to the flow of streams is the accident of whether the precipitation takes the form of snow. In upland districts, particularly in the north, a large proportion of the winter fall remains on the ground in this form until warm weather returns in the spring or even summer, when it rapidly flows off in heavy floods. This introduces a complication for which it is extremely difficult to allow. Another complication arises from the occurrence of prolonged frost. The flow from any area is greatly diminished during such conditions, and there is a corresponding excess of flow on their termination. On the other hand, should rain fall upon frozen ground its effect on the run-off is greater than otherwise. Information connecting the fall of rain with the level of underground water is, if anything, even more meagre and is equally unsuitable for general application. The observations available are those 284 RAINFALL OF THE BRITISH ISLES of the depth of water in wells, and in order that these may be accurate, no appreciable draught must be made upon the well. Even if a correct account is kept of the abstracted water, the records do not admit of correction by this means, since it is not possible to tell with certainty the area from which this water is being drawn, and in any case the zone of depression caused by its withdrawal is steadily neutralized by natural processes. The rapidity with which a well responds to rain- fall depends upon its depth, upon the season of year, and upon the nature of the geological strata. Dr. C. P. Hooker l has shown that a well sunk in the oolite near Cirencester responded extremely readily to local rainfall showing a wide fluctuation. Mr. R. Cooke and Mr. S. C. Russell, 2 dealing with a well in the chalk near Maidstone, found a sluggishresponse ,and a restricted range of water-level. It is probable that sensitiveness is an indication of a small degree of permeability, and that in soil of a highly porous nature the rise and fall of underground water is general rather than local. An important contribu- tion to the literature of the subject, by Mr. D. Halton Thomson, 3 deals with the records of the 84 years 1836 to 1919 at Chilgrove, lying in the high chalk downs of West Sussex. These records, in common with the majority of others, exhibit an almost invariable maximum in the late winter or early spring, a diminution in level till the following autumn, and a rapid recovery during the winter. In years of exceptional rainfall, either in respect of amount or of seasonal distribution, considerable 1 See Q.J.R. Met. Soc., vol. xxix, p. 263 ; vol. xxxiii, p. 287. * See Q.J.R. Met. Soc., vol. xxxvii, p. 125. 3 See British Rainfall, 1919, p. 247. ECONOMIC APPLICATION 285 variations from this simple regime occur, amounting on occasion to almost complete inversion. The accompanying diagram, Fig. 126, gives the the average results for the whole period at Chilgrove. The relation of the well-depth to the actual rainfall is not extremely clear, and a second curve is added showing the estimated effective rainfall computed by deducting from the actual fall an allowance for Rainfall and Wei I -Depth at Chilgrove, Sussex. AVERAGE. 1836-1919. Month JAN. FEB. MAR. APR. MAY. JUNE JULY AUG. SEPT OCT. NOV. DEC. Month. WOL- OCPTH. /*. 200 190 180 170 160 150 140 130 120 \ RAIN- FALL in. 4 3 Z / \ p^lt ^&> / ^+ ^\ 35, 38, 47 ; diameter of, 31, 35, 53 J early, 30, 46 ; errors of, 68, 88, 91 ; experimental, 33, 45, 56 ; exposure of, 54, 65, 273 ; fixing of, 38 ; funnels of, 33, 34, 47 ; height above ground of, 55 ; life of, 38 ; material for, 35 ; mechanical, 67, 86 ; obsolete patterns of, 46 ; principle of, 31 ; shields for, 61, 64; thermal insulation of, 36, 47 Rectangular rain gauges, 47 Residual mass curve, 219, 226 Roof exposures, 56 Rotating rain gauge, 59 Rotherham rain gauge experi- ments at, 45, 56 Russia, rain gauge experiments in, 63 Salt, in rain, 13 Sea-level, average rainfall at, 247 Seasonal variations of rainfall, 188, 192, 211 year, the, 212 Seathwaite rain gauge, 43 Secondary cyclones, 23, 116 Self-recording rain gauges, 67, 86 ; lubrication of, 88 ; sensitiveness of, 72, 91 September, average rainfall of, 190 12, 1911, rainfall of, 155 25-26, 1915, rainfall of, 173 Shelter, effect of, on rainfall, 255, 259 ; of rain gauges, 66 Shields for rain gauges, 61 Shower, definition of, 130 ; iso- chronous maps of, 131 Showers, spring, 22 Side-tube rain gauge, 47 Sleet, 12 Snow, 12 ; measurement of, 38, 61, 1 60 gauge, 38 Snow-storm of December 25-26, 1906, 131 ; of April 24-25, 1908, 160 Snowdon rain gauge, 33, 34, 37 Snowdonia, rainfall in, 257 South Downs, rainfall on, 253 Spring, rainfall of, 196, 202 Squall-showers, 25 Standard rain gauge, 31, 36, 37, 38, Stratfield Turgiss, rain gauge experi- ments at, 45, 56 Streams and rainfall, 270, 278, 282 Summer, rainfall of, 196, 202, 210 Super-saturation, 9 Survey, rainfall, 247, 286 Symons rain gauge, 47 Tap rain gauge, 47 Temperature variations, atmo- spheric, 14, 15 Thermo-dynamic cooling, 16, 18 Three year period in rainfall, 227 Thunderstorm rain, 19 ; distribu- tion of, 141, 144, 148, 149, 151 ; time relations of, 136, 141 Thunderstorms, frequency of, 115 ; winter, 116 Trees and rain gauge exposure, 65, 66, Types of rainfall distribution, 130, J 53, J 63 ; identification of, 100, 103, 271 Ulley Reservoir, rain gauge experi- ments at, 46, 56 Undercutting winds, 23 Underground water, 270, 283 Units for rainfall measurement, 39, 118, 119 Upper air, temperature of the, 17 Valleys, rainfall in, 265, 267 Vapour-pressure, 7, 8 Vegetation and stream flow, 283 Volumetric computation of rainfall, 102, 270, 275 INDEX 295 Walton-on-Thames, rain gauge experiments at, 46, 58 Water power, rainfall and, 272 Resources Committee, 286 supply, rainfall and, 272 vapour, 7 Wells, rainfall and, 284 ' Wet day," the, 121 Wild's rain gauge fence, 64 Wind and rain gauges, 29, 33, 55, 60, 65, 273 Wind-eddies, 60 Winds in cyclone, 23, 179 Windward slopes, rainfall on, 253 Winter, rainfall of, 193, 196, 201, 210 Zones, rainfall, 102, 275 PRINTED IN GREAT BRITAIN FOR THE UNIVERSITY OF LONDON PRESS, LTD. BY HAZELL, WATSON AND VINEY, LIMITED, LONDON AND AYLESBURY; 14 DAY USE RETURN TO DESK FROM WHICH BORROWED LOAN DEPT. This book is due on the last date stamped below, or on the date to which renewed. Renewed books are subject to immediate recalL JM*^ LD 21A-50m-12,'60 (B6221&10)476B General Library University of California fieri 09577 498246 UNIVERSITY OF CALIFORNIA LIBRARY