INTERNATIONAL LABOUR OFFICE STUDIES AND REPORTS Series N (Statistics) No. 13 METHODS OF COMPILING HOUSING STATISTICS GENEVA 1928 PREFACE At the present moment, when the housing question, with its manifold economic and social problems, is more and more attracting the attention of administrative bodies and public opinion, it is highly necessary that exact information on the. subject should be available. Housing statistics, which have hardly existed for more than a few decades, have recently undergone considerable development ; but they are, however, still somewhat haphazard. Both local and central administrative authorities take part in the compilation of such statistics, and often without regard to what is being done outside their own sphere of activity. An effort towards co-ordination on a common basis is thus essential, since comparisons between one locality and another, or between one country and another, are becoming daily of more urgent necessity. The question of compiling the various statistics on uniform methods had already been broached before the war. The German and the Swiss unions of municipal statistical offices, as well as various learned societies, had been dealing with the matter for some considerable time, but predominantly from a national point of view. The fourteenth Congress of Hygiene and Demography, held in Berlin in 1907, was one of the first to consider these questions in their international aspect. Later, in 1913, at the fourteenth session of the International Statistical Institute, at Vienna, a report was presented on this subject, but no decision was taken. The outbreak of the war then interrupted these first efforts towards standardisation. After the war, this question was placed on the agenda of the International Housing Congress, which was to have met in Rome in 1922, but this meeting did not take place. — IV — Finally, some years later, the International Union of Municipalities took up the question from its own standpoint, and requested the collaboration of the International Labour Office in preparing a report on the standardisation of the various municipal statistics dealing with housing problems. The International Labour Office which has always paid great attention to the problems of standardisation of statistical data \ and which has closely followed all the questions relating to the housing of workers 2, promptly acceded to this request. In September 1925, it laid before the Executive Committee of the International Union of Municipalities, at its meeting in Paris, a preliminare memorandum, of which the present study is an expansion and development. In order to obtain information as to the different municipal statistics in existence, the Office applied to the authorities in a number of large European cities, the greater number of which forwarded a very complete series of records. Particularly useful replies were received from the municipal authorities of Amsterdam, Basle, Berlin, Brussels, Budapest, Florence, Frankfort, Hamburg, The Hague, London, Lyons, Manchester, Munich, Oslo, Prague, Stockholm, Warsaw, and Zurich, to whom the grateful thanks of the Office are due for the valuable cooperation which they have been so kind as to give. 1 Three international conferences of labour statisticians have indeed already been held under t h e auspices of the International Labour Organisation and t h e Resolutions adopted by these conferences on t h e basis of reports presented by t h e International Labour Office have laid down the general principles for an international standardisation of the principal branches of labour statistics. On this subject see Studies and Reports of the INTERNATIONAL L A B O U R O F F I C E , Series N, Nos. 1-12. 2 In 1922 a Resolution of the International Labour Conference invited the Governing Body of t h e International Labour Office to " examine the opportuneness of t h e Office pursuing studies undertaken on this subject, having recourse if necessary t o t h e collaboration of qualified experts " (INTERNATIONAL LABOUR CONFERENCE, Fourth Session, Geneva, 1922 : Stenographic Record of the Proceedings, p . 600). Since t h a t date various publications on the development of housing legislation h a v e been published : INTERNATIONAL LABOUR OFFICE : European Housing Problems Since the War, Studies a n d Reports, Series G, No. 1, Geneva, 1924 ; The Housing Problem in the United States, Studies and Reports, Series G, No. 2, 1925 ; " Housing as a Post-War Problem in Europe " , International Labour Review, Vol. X , Nos. 2 a n d 3, Aug. and Sept. 1924 ; " The Housing Problem in Soviet Russia " , ibid., Vol. X I I , No. 2, Aug. 1925 ; also various comments on housing under the heading " Social Conditions " in Industrial and Labour Information, Vol. X I I et seq. y The Office has not, however, considered it possible to limit its analysis to municipal statistics only, as was suggested by the Union of Municipalities, since they form only part of existing housing statistics. Statistics compiled on a national basis have indeed considerably developed, and it is not possible to leave them out of account, especially since, in virtue of their national character, they represent a basis to which municipal statistics have or will have to be adapted. Consequently, both kinds of statistics have been taken into consideration. I J CONTENTS Page PREFACE INTRODUCTION in ' CHAPTER I. — General Problems Basic Units with their Characteristics and Units of Measurement Delimitation of the Investigation in Space and in T i m e . . . The Methods of Investigation The Collection of D a t a The Presentation of D a t a CHAPTER II. — Statistics of Dwellings Definition of a Dwelling Classification of Dwellings Classification according to Intended and Actual Use. Classification according to Size Classification according to Other Criteria CHAPTER III. — Statistics of Buildings with Dwellings and of Registered Properties with Dwellings Statistics of Buildings with Dwellings Definition of Buildings with Dwellings Classification of Buildings with Dwellings Statistics of Registered Properties comprising Buildings with Dwellings CHAPTER IV. — Statistics of Housing Conditions Distribution of Occupants in the Different Types of Dwellings and Buildings Density of Occupation Living Together of Several Households and Composition of Households Housing Conditions of Different Social Groups CHAPTER V. — Statistics of Rents Definition of a Rented Dwelling and of R e n t Classification of Dwellings by Rents and Determination of Average Rents CHAPTER VI. — Statistics of the Housing Market Censuses of Vacant Dwellings Definition of Vacant Dwellings Classification of Vacant Dwellings and Comparison with Other D a t a Statistics of Public Housing Offices Indexes Based on Qualitative Estimates 1 9 9 11 14 18 20 23 23 31 31 33 39 42 42 43 45 49 51 53 54 59 61 65 67 69 73 73 74 76 78 81 Vili — Page CHAPTER VII. — Statistics of Fluctuations in Total Housing Supply The Different Forms of Statistics of Fluctuations in the Total Housing Supply and Their Characteristics Classification of Buildings and Dwellings and Comparison with Other D a t a Conclusions The Different Housing Statistics The Statistical Units Censuses of Dwellings Censuses of Buildings with Dwellings Censuses of Housing Conditions Statistics of Rents Statistics of the Housing Market Statistics of Fluctuations in the Total Housing Supply . . A P P E N D I X . —List of Official Statistical Reports relating to Housing Used in the Present Study 83 85 89 95 95 96 99 100 101 103 105 106 109 \ INTRODUCTION By the expression " housing statistics " is understood all statistics concerned directly with housing problems, and capable of throwing a light on housing policy. It will thus be necessary to include in this study a number of fairly different types of statistical records. A study of housing censuses proper will first be made, that is, those of separate dwellings and of buildings containing dwellings (and to some extent of registered properties), as well as of censuses of the persons inhabiting these houses or dwellings, by means of which the prevailing housing conditions can be brought out. Statistics concerning rents will then be treated, as well as those which can be used to ascertain the situation existing in the housing market (supply of and demand for dwellings). Lastly, the statistics regarding the construction, transformation and demolition of buildings will be examined, which enable the fluctuations in the total housing supplies, that is, the changes which gradually take place in the total stock of dwellings and buildings, to be studied. This study is, however, confined to statistics dealing with urban centres, since it is in these that housing policy is particularly active. In rural areas housing problems present themselves in quite a different manner, and require peculiar treatment ; statistics applying to these are also still rather rare ; they must, moreover, be compiled according to a plan which conforms to that of the other statistics regarding agricultural life. It should be noted that the various statistics of urban housing have not as their sole purpose the provision of information on which to base housing policy, but that they are also of interest to the town planner, who is concerned with the laying out and development of the city as a whole, or to the sociologist, who studies the conditions of life of the various classes of society. These persons have special needs, which cannot be neglected in studying the methods of compiling housing statistics ; and — 2 — they are, therefore, taken into consideration also as the necessity arises ; but it is primarily from the standpoint of housing policy that this research has been undertaken. Historically, the development of different branches of housing statistics has been fairly gradual ; it has progressed at a rate corresponding with the extent to which public authorities have undertaken responsibilities in regard to housing. Up to the end of the nineteenth century, the public authorities, under the influence of the economic liberalism of the laissez-faire school, hardly dealt at all with housing. The State systematically excluded from its competence the questions which would have involved action in this sphere ; local authorities restricted their responsibility to the problem of town or street planning, and the building as well as the sale or lease of real estate providing housing accommodation was regarded as coming strictly within the scope of private enterprise. Consequently, during this period statistics bearing on housing problems are somewhat rare. Little by little, however, the conception of social policy in regard to housing began to take form, and certain municipalities began to feel concerned as to the conditions under which their population was housed. About 1860 there appeared, principally in Central Europe 1, the first statistical investigations dealing in any really scientific manner with these questions ; but these were still confined to a few isolated localities. A little later the central Governments began to realise the importance of compiling statistics on a national basis which would provide information as to the housing conditions of the populations of their countries, and in the course of the last decade of the nineteenth century and the first of the twentieth they added to the scope of their general population censuses 1 Particular mention should be made of one of the first censuses of this kind taken in Germany, viz. the Berlin census of 1864 (S. NEUMANN : Die Resultate der Berliner Volkszählung vom 3. Dezember 1864 ; Berlin, 1866). This was followed by one taken in Königsberg in the course of the same year and by those of Hamburg, Leipzig and Chemnitz in 1867, as well as the Swiss enquiry, which has become famous and was undertaken in Basle in 1899 (K. BÜCHER : Die Wohnungs-Enquête in der Stadt Basel vom 1. bis 19. Februar 1889; Basle, 1891). On this subject, for the whole of the subsequent period up to the war, see the two following works : Beiträge zur Arbeiterstatistik, No. 12 : " Gebiete und Methoden der amtlichen Arbeitsstatistik in den wichtigsten Industriestaaten ", prepared by the KAISERL. STATISTISCHE AMT, Berlin, 1903 ; and Bernard FRANKE : Grundstücks- und Wohnungsstatistik : Chapter L, " Die Statistik in Deutschland nach ihrem heutigen Stand ", Vol. II, Festgabe für Georg v. Mayr, Munich and Berlin, 1911. — 3 — returns dealing with dwellings, buildings and housing conditions. In the cases in which the national Governments did not undertake these enquiries, or undertook them on a too restricted scale, the various municipal services joined together to take such censuses on a uniform basis so as to obtain comparable returns. The same progress was not made in the matter of the other housing statistics — those of rents, housing market, and fluctuations in the total housing supply. This is because, apart from the general censuses of rents included in the censuses of inhabited buildings, these various statistics require to be compiled periodically for the purpose of tracing the fluctuations in the phenomena studied, and it was (and still is) only the municipalities, which manifest a deeper interest in housing problems, that undertook the task of regularly collecting such data. Again, since some of these statistics are established as a result of the activities of the public services — e.g. supervision of building and organisation of the sale and lease of dwellings — they cannot be obtained unless such departments of the public services are functioning as they should. After the war, the general housing shortage which prevailed in the majority of countries compelled the Governments to take direct action in regard to housing problems. Then it was that the gaps in the existing statistics became evident. Quite elementary data, needed to give direction to their policy of encouraging building, were not at the disposal of the Governments. Estimates of the housing shortage were made in various quarters, but the results obtained did not really agree. In some countries it was decided to undertake systematic enquiries on a large scale. In almost all cases an important place has been allotted to the question of housing in the censuses that have been taken since 1920. It is becoming a much more general practice in the various localities to collect statistics concerning rents, the housing market, and fluctuations in the total housing, and earnest attempts have been made towards standardisation on a national basis. It is certain that this development of the various kinds of housing statistics will continue, for it is evident that in these days social policy in the matter of housing is becoming more and more important. Governments which have been compelled by the post-war housing shortage to adopt a policy of intervention will probably change their policy once the critical period has passed ; but they will certainly not be able to leave — 4 — the matter out of consideration. Sooner or later this problem will have to be looked at on an international plane, and even now it is absolutely necessary to make comparisons between one country and another in order to obtain information as to the situation existing elsewhere and the experiments made by others. So the problem of the unification of housing statistics has to be solved at the present time on both a national and an international basis. It deserves very serious attention, for it is not lacking in difficulties. It is not only that the investigator is confronted with divergencies in statistical methods, but that he is also faced with differences in the actual nature of the things that have to be compared, the kind of housing accommodation varying with the stage of civilisation, the manners and customs and the prosperity of the population to be housed, and with the climatic conditions of the country. So it is that certain structural types have come into use in certain countries, the equivalent of which cannot be found in others. The houses built in tropical lands, for instance, have nothing in common with those of cold or temperate regions. Even without going so far, the hovels which are found in certain quarters of our big modern cities are hardly to be compared with even the modest types of accommodation provided in other more recently built and better planned quarters. Again, the small detached or semi-detached cottage or villa standing in its own little garden, which is found so generally in the AngloSaxon countries, for example, represents quite another class of housing than the cramped flats of the great barrack blocks which have become such a prominent feature of certain parts of Central Europe. The internal arrangements of dwellings also differ very considerably. The methods of heating, which are of primary importance in northern countries, are of hardly any in warmer climes. The " alcove " 1, which is considered an almost indispensable part of dwellings in certain countries, is nonexistent in others. The kitchen, which in some cases is devoted exclusively to the preparation of meals, in other cases plays the part of dining room or living room, or serves as a bedroom for 1 Usually a space at the back or side of a room, shut off by folding or other doors, and serving as sleeping accommodation. — 5 — servants or even for members of the family. The water-closets fulfilling all the requirements of hygiene and modern comfort which are found in certain districts or in the dwellings provided for the well-to-do classes can hardly be compared with the primitive arrangements to be found elsewhere. It will be admitted that under these conditions the possibilities of comparison between one locality or country and another are relatively limited, and that no hopes can be entertained in this respect of obtaining an absolutely strict standardisation of statistical data. Nevertheless, it appears to be possible to eliminate certain divergencies still existing in the fundamental definitions and in methods of compilation, and also to be possible to set up uniform standards to which the various national services could adapt their statistics without sacrificing the value of these in answering local needs and circumstances. That is the task to which this research is devoted. In the pages which follow attention will be paid first of all, in Chapter I, to the general problems presented by all the various branches of housing statistics. The special problems proper to each of the various branches will be treated in subsequent chapters. Three chapters will be devoted to the question of housing censuses, considered under three separate heads, according to the special object in view : Statistics of dwellings, which aim at determining the total supply of dwellings at a given moment and their characteristics ; Statistics of buildings with dwellings and statistics of registered properties 1 comprising buildings, with dwellings, which bring out the structural character of the housing accommodation available ; Statistics of housing conditions, which show the manner in which the dwellings and inhabited buildings are utilised, and consequently the conditions under which the population is actually housed. 1 Registered properties {parcelles cadastrales) : in many European countries registers are kept of all properties with the buildings erected on them, each unit of property (buildings always belonging to the same owner as the land) being entitled a parcelle cadastrale, whether consisting of an unoccupied site, a single house and grounds, a block of houses with grounds, or a whole street. — 6 — The following chapter will be reserved for statistics of rents, under which term will be included, on the one hand, the general censuses of rents made on a similar basis to those of dwellings (of which they often form a supplementary section), and on the other hand, enquiries made on a more restricted scale but at more frequent intervals. In the chapter devoted to housing market statistics, consideration will first be given to the census returns of unoccupied dwellings taken at regular intervals, and to the estimates of the housing requirements of the population which often accompany these investigations, and then to the data which should be used in the production of approximate indexes of the position of the housing market, and particularly the information regarding supply and demand contained in the registers of the public housing offices. The final chapter will be devoted to the subject of statistics of fluctuations in the total housing supply1. There several kinds of statistics will be considered : on the one hand, those which relate to the total amount of building activities, whether they be operations actually carried out or simply plans submitted to or approved by the authorities ; and on the other hand, those relating to construction carried out, or projected, by aid of grants from public funds or organised by big public bodies or private building societies. In conclusion the essential points that have come out in the course of the analysis will be summed up. The whole study will be based on actual facts, and reference will be made repeatedly to concrete examples. Nevertheless, these notes cannot claim to give all the instances in which a particular experiment has been carried out, since the information on which this study has been based, although fairly extensive, could not be considered absolutely complete. In the first place the study undertaken is restricted to comparatively recent documents, dating from the years since and immediately before the war (1910-1914) ; and in the second place, while it includes the principal national publications during that period, in the matter of municipal publications it does not go 1 In French, le mouvement des habitations. — 7 — beyond those of a limited number of selected cities in various countries. Such as it is, however, it may be claimed that the information available has been sufficient for bringing out the principal problems which arise and for examining the various solutions which have been adopted. At the end of the book there will be found a detailed list of the publications which have formed the basis of this study. CHAPTER I GENERAL PROBLEMS BASIC UNITS WITH THEIR CHARACTERISTICS AND UNITS MEASUREMENT OF Basic units are, in the sense in which this term is understood here, the units which are enumerated, measured, and classified according to their various characteristics. In housing statistics, which are intended to provide a basis for housing policy, the principal basic unit is undoubtedly that of the dwelling considered as the home of a household. Since, however, a number of dwellings are often contained under a single roof, it will be necessary also to study the characteristics of the buildings which contain them ; this suggests as another basic unit the building with a dwelling or dwellings. On the other hand, since a dwelling is usually subdivided into several rooms, these have in certain cases also been enumerated, without reference to their distribution between the various dwellings, and in this way a new basic unit, that of the room, has been introduced. Further, in order to approach certain legal problems connected with housing policy — the ownership of dwellings, the forms of acquiring them — use has been made in certain countries of a basic unit of a legal character which lent itself better to this purpose, i.e. the registered property which is the site of the buildings with dwellings 1. Housing policy is not concerned solely with the condition of existing housing accommodation ; it is also concerned with the manner in which it is utilised, and the housing conditions which result therefrom. It will thus be necessary to bring in a basic unit of an entirely different order, the persons inhabiting 1 See footnote to p. 5. — 10 — the dwellings. Further, since these persons do not live in their dwellings as isolated individuals, consideration will also have to be paid to the various groups they may form. Among these there is one of particular importance, the household (giving to this term a wide interpretation) ; for one of the fundamental principles of housing policy is the desirability of reserving one dwelling to each household. This must consequently also be taken as a basic unit., Some of these units are common to several of the different branches into which housing statistics have been divided, and in some of these branches several distinct basic units will be found. Thus in the statistics of dwellings, in those of buildings with dwellings, and in those of registered properties comprising buildings with dwellings, the basic units are respectively — as is indicated by their names — the dwelling, the building, and the registered property on which buildings with dwellings are erected. Similarly, statistics of rents and those of the housing market both have as their basic unit the dwelling, but their scope is different. The first deals only with rented dwellings, the rents of which it analyses, and the second with dwellings offered (that is, vacant dwellings) and those demanded. In statistics of housing conditions there may, on the contrary, be three main basic units — the individual, the dwelling, and the household — each requiring separate analysis. The statistics of fluctuations in total housing supply, again, are built up on two basic units : the building and the dwelling. One of the first processes in statistical analysis consists in classifying the basic units according to their different characteristics. In housing statistics this classification may be made, in the first instance, according to purely qualitative criteria : the absence or presence of certain structural features in buildings containing dwellings, or of certain internal arrangements of the dwellings or rooms. It can equally well be made according to quantitative criteria, which means that each basic unit is first subjected to a process of measurement, and the classification is carried out on the basis of the measurements so obtained. This naturally presupposes the introduction of a unit of measurement, which may be either a standard commonly adopted for measuring size, value, duration (metres or feet, currency unit, month, week, etc.), or even one of the elements of the basic unit : the room, as giving the size of the dwelling or building ; the story for the height of the building ; or the — 11 — window as a measure of the degree of ventilation of the room, etc. The fact must also be emphasised that these units of measurement can be actually the same as those elsewhere adopted as basic units. Thus the dwelling is used as a unit of measurement of the size of the building in statistics of buildings. Similarly, the individual serves as a unit of measurement for calculating density of occupation, and then appears simultaneously as a basic unit and a unit of measurement. The statistics in fact give the number of individuals — basic units — classified according to the number of individuals — units of measurement — who share the same dwelling. The chapters devoted to the special problems of the different branches of statistics will deal with the question of the exact definition of the various basic units, the characteristics of these units which require consideration, and the method of measuring them. DELIMITATION OF THE INVESTIGATION IN SPACE AND IN TIME The geographical scope of a statistical enquiry naturally depends in the first place on the authorities under whose auspices it has been undertaken, since these authorities cannot trespass beyond the bounds of their administrative areas. It has already been mentioned that there has been a tendency in the last score or so of years for housing statistics to pass from the hands of the municipalities into those of the State, and that where their compilation is still in municipal hands an attempt has been made to co-ordinate the various local statistics on a national basis. It should be noted that many so-called " national " censuses are really restricted to urban centres of a certain size x. For 1 Thus, the census taken in Finland in 1919 and that taken in Sweden in 1920, while covering all urban centres dealt only with one-sixth of the total population, by reason of the agricultural character of these countries. In Austria, the census of 1910 was taken in 132 urban communes, and covered nearly one-fifth of the total population ; further, the thoroughness of the enquiry varied with the size of the locality ; in sixty-two communes belonging to the ten most important cities, certain supplementary questions were added to the census papers. The German census of 1918 applied to communes having a population of over 5,000, also to those with a smaller population but situated in an industrial area, or adjacent to a more important commune, the — 12 — these investigations, the problem of delimitation in space consists in the choice of the centres to be studied and in the delimitation of these individual centres. Housing statistics really covering the whole of the national territory are to be found only in comparatively few countries, and here again the investigations have not always been earned out with very great thoroughness 1. Besides, if it is desired to use these statistics for exploring housing problems at all thoroughly, a very clear distinction must be drawn between urban and rural areas, since the problems in these two types of area are very different, and must thus be approached separately. Here again then, in order to make a clear distinction between town and country districts, it will be necessary to lay down exactly what is to be understood by an urban centre. This process is carried out by fairly similar methods in the various statistics, but it must be said that the manner in which it is performed is very generally defective. Almost universally, indeed, the basic area is that of the existing political and or administrative divisions, and the central area of a town may be grouped with the outer district or districts, whether of an urban or suburban character. This procedure, following the methods of the general population censuses, for which it is adequate in many respects, offers marked disadvantages for housing statistics ; for these political or administrative boundaries, fixed often years ago on historical and other grounds, are frequently out of date from a practical standpoint, and it often happens that by adhering to them certain areas of a suburban character are left out and that, on the other hand, a town area may be made to include regions which are rather rural in character. population of which might eventually overflow into the smaller commune;, it is estimated that in this manner reports were obtained covering. about 55 per cent, of the total population of the country. The Swiss census of 1920 was restricted to communes of more than 5,000 inhabitants, to those forming the outer circle of large towns and to other communes specially chosen by the cantonal authorities. As the latter made large use of this discretionary power, the census actually covered 59 per cent, of the total population. 1 Examples of censuses covering a whole country are to be found particularly in Belgium, France, Great Britain, Italy, Norway, Poland, Australia, and Canada. As a matter of fact, in Belgium the only results published are those from fairly important localities. In Italy the national census is very cursory ; an additional enquiry in greater detail was undertaken by the Union of the Statistical Offices of the Italian Municipalities on the occasion of the 1921 census, but it dealt onljr with the principal urban centres. — 13 — It further sometimes happens that two sets of statistics compiled by different departments do not apply to exactly the same geographical area, in cases where there is some overlapping between the administrative areas. It would, therefore, seem to be preferable to draw new boundaries expressly for the purposes of the statistics, and in such a way as to include in them the whole of the town under consideration. That is clearly a somewhat delicate operation, but it does not appear to be impracticable, given a fair knowledge of local conditions. These new limits would have to include the whole of the area within which there is absence of rural undertakings, the inhabitants, whether by their occupations or as consumers, being in direct and constant touch with the centre. It is true that such criteria could not help being to a certain extent arbitrary. It seems, however, that they would be sufficiently straightforward and self-evident to allow of making the subdivision in a reasonable manner, any doubtful cases being left to the local services to decide according to the special circumstances of each locality. It is clear that this operation would necessarily have to be performed in an identical manner for all the various branches of housing statistics, and that a preliminary arrangement would have to be come to between the various responsible services. The scope of the investigation in time is bound up primarily with the purpose of the statistics and with the nature of the phenomena recorded. By restricting the enquiry to a given date, a static view of the phenomena is obtained. This is only of real interest if the phenomenon is relatively stable, and requires to be studied in all its complexity ; the enquiry being undertaken once only, or at relatively long intervals, it can be carried out very thoroughly and in great detail. On the other hand, by extending the investigation over a period of some length, it is possible to trace the evolution of a phenomenon and to obtain a more kinetic view of it. This is essential when there are fairly rapid changes ; but the phenomenon must be one to which a comparatively simple process of measurement can be applied, for, on account of the large amount of work entailed, the record is necessarily limited to certain elementary characteristics. In housing statistics, the censuses of dwellings, buildings, registered properties and housing conditions provide primarily a static view. In statistics of rents, where the phenomena — 14 — studied are too complicated to be illustrated thoroughly by the brief summary provided by a periodical return, and too unstable for enquiries made at long intervals to be sufficient, both kinds of investigations are used. In the statistics of the housing market it is particularly the changes which take place in this market which are studied, and in the statistics of fluctuations in total housing supply the object pursued is, by recording the alterations which take place in the total supply of dwellings and of buildings, to obtain a kinetic view which will supplement the static record provided by the census of dwellings and buildings. T H E METHODS OF INVESTIGATION In order to obtain a view of the facts — static and dynamic — recourse must be made to two kinds of methods of investigation : either a systematic record must be made of all the basic units in the field covered by the enquiry at the date or during the period fixed for it, or only a fraction of them must be taken, but in such a way that this fraction shall be representative of the whole. It is the former of these two methods which, in the form of a census, is the more frequently employed in housing statistics. The statistics of dwellings, buildings and registered properties, and housing conditions result from general censuses. Similarly, in the statistics of rents, a large part of the data is obtained by censuses, often carried out in conjunction with the census of dwellings. For statistics of the housing market a census of vacant dwellings is sometimes taken. Again, the principle of the complete survey is applied to the statistics of the fluctuations in total housing supply, where what is aimed at is a continuous record of buildings and dwellings constructed, demolished, or transformed. This method has certainly the advantage of providing a complete view of the phenomenon studied, but it has the serious disadvantage of requiring a large amount of work and being consequently too burdensome to undertake at other than relatively infrequent intervals. The other method appears to be more advantageous when the chief emphasis is laid on determining the characteristics of the basic units and the relations between them, or the changes — 15 — undergone. But difficulty lies in the selection of the units to be studied ; for it is essential that the selection should be representative at least approximately of the whole. For this purpose records may be made according to any of three distinct methods. The simplest is that of the purely arbitrary choice of a certain number of typical units which are considered representative, basing this choice simply on a general acquaintance with the prevailing conditions ; it does not, however, appear to provide adequate guarantees of trustworthiness, the more so that no means are available for estimating the extent to which the partial data are representative of the whole. This method, which is sometimes used for the selection of families for the purpose of studying housing conditions in certain social classes, or for picking out a given number of localities to be covered by a census, has hardly been worked out at all in housing statistics, nor is it likely that it will be in the future. A much better method appears to be that of sampling, which consists in picking out a given number of basic units " at random ", but with what might be called systematised chance, the incidence of factors which might bias the results being excluded, thus ensuring to each unit an equal chance of being included in the investigation. In point of fact, in accordance with the law of great numbers, a sample is thus obtained which presents all the characteristics of the whole, and such small divergencies as may occur can be estimated by applying the theory of probability. This method has so far been applied somewhat sparingly; certain instances which appear to have given excellent results can however be quoted. These are, the enquiry into housing conditions in the small dwellings of the city of Christiania (Oslo) carried out in 1913-1914 by the Municipal Statistical Office of that city 1 ; the enquiry into rents made in January 1924 in the principal towns in Switzerland, according to the lines laid down by the Federal Labour Office 2 ; and, finally, two private enquiries carried out before and after the war in six English towns, under the direction of Professor Bowley, for 1 Beboelsesjorholdene i smaaleiligheter i Kristiania, munens Statistiske Kontor, Specialunderskkelser II. 2 1913-1914 ; KomChristiania, 1915. O F F I C E FÉDÉRAL D U TRAVAIL : " Résultats de l'enquête sur les loyers effectuée au mois de janvier 1924 " . Informations sociale, 1924. Vol. 2. de statistique — 16 — the purpose of discovering the proportion of their inhabitants living in poverty, in which housing conditions are treated in detail x . In using this method it is obviously of the first importance to eliminate with care in the selection of the units any element capable of causing a deviation from pure chance. These conditions, however, appear to be quite realisable in the field of housing statistics, where complete registers of the units to be sampled are usually available. As to the proportion of basic units which ought to be examined, this is to be determined by taking into consideration the total number of units comprised in the field to be surveyed, the degree of accuracy required in the results, and the details of the analysis to which they are to be subjected 2. The generally recognised advantage of this method is that it greatly diminishes the work required and consequently the expense entailed by the investigation, while giving fairly accurate results. For this reason, although it has not yet been greatly used, it appears to have a great future, particularly for the thorough examination of the kinds of dwellings and of housing conditions. Such enquiries might be undertaken as a supplement to the censuses, which would then be restricted solely to certain fundamentally important factors. A quite different kind of selection is also to be met with in some statistics. This is in no wise the result of a deliberate experiment by the statistician, but is forced upon him by the circumstances under which he has to collect his data. A selection of this kind is constituted by the data regarding supply of and demand for dwellings furnished by the municipal housing 1 A. L. BOWLEY and A. R. BURNETT-HURST : Livelihood and Poverty. London 1915. Also A. L. BOWLEY and M. H. HOGG : Has Poverty Diminished ? London, 1925. 2 The manner in which the sampling of the dwellings in the enquiries mentioned was carried out varied in a marked degree. In the Norwegian investigation one house in three in each street was selected, and an examination was made of all the small dwellings (one or two rooms) contained in these buildings. In the Swiss enquiry a survey was made of 10 per cent, of the middle-sized dwellings (two to four rooms, with or without attics) occupied by tenants, by limiting the investigation to houses bearing a certain street number and sometimes to certain stories in a house, or to certain streets distributed between various quarters. In the English enquiries the addresses of householders were ticked at regular intervals in local directories, rating lists, or burgess rolls ; the proportion chosen varied, according to the size of the town, between one in eight and one in thirty-six. — 17 — offices in the statistics of the housing market. These statistics are compiled solely on the basis of the returns made to such offices, and unless these are compulsory the figures never include the whole of the current supply of and demand for dwellings. How far a choice made in this way is representative depends on the extent of the activities of the housing offices and the manner in which they function. Similarly, in the statistics of fluctuations in total housing supply, the data published by some large building societies and those regarding Statesubsidised building are the result of an analogous selection. It may be added that these statistics are often published only to show the activities of the services in question, and that it is only by an after-thought that use is made of them as more or less representative indexes of a more general phenomenon. It should be noted that the principle of representative selection is applicable not only from the point of view of space, but also from that of time. In actual fact, where it is a case of following the fluctuations of a phenomenon in time, all that is done generally is to take a survey at regular and more or less frequent intervals of the conditions obtaining at a given moment. This simply means choosing in the "ever-rolling stream " of time certain dates at which it is assumed to be possible to obtain a representative view of the phenomenon. The frequency of the dates chosen will vary directly with the rapidity of the fluctuations studied and the degree of exactitude sought. This is the method used for the statistics of rents and of the housing market, in which regular returns are taken of rents and of vacant dwellings. It is also used, but less frequently, in the statistics of the fluctuations in total housing supply, when the number of houses or dwellings in course of construction or demolition is ascertained at fixed dates. The principle of continuous registration, that is to say, a record complete in time, is applied only when it is desired to record the volume of the changes taking place. This occurs, for instance, in statistics of construction, where a record is made of the number of building permits granted, or the number of buildings completed in the course of given consecutive periods, or, in statistics of the housing market, a record of the number of dwellings offered or applied for as registered at the public housing offices. — 18 — T H E COLLECTION OF DATA As a rule, in statistics of housing, the general censuses of dwellings, housing conditions, buildings, registered properties, and rents are all taken as part of a single process, since many of the data collected for one set of these statistics are also required for another and the sources from which they are derived are the same. These sources are in general two in number, the owners of house property and the occupiers of dwellings. In principle the questions that have to be asked are divided between them in such a way that both parties have to reply on the points regarding which they are best informed. Thus it is generally assumed that the owner is in a better position to supply information on the structural conditions of his property, the materials of which the houses are built, the date of construction, the existence of accessory buildings, yards, etc., and the number of separate dwellings comprised in each property. The latter point is, as will be seen further on, of particular importance, and it certainly appears to be the owner of the building who is the person best able to judge of the merits of border-line cases. Similarly, the owner seems the more suitable person to reply to questions concerning rents, for he is able to distinguish in the sums paid the part which represents rent pure and simple from charges for the upkeep and service of the building and taxes, which distinction the tenant cannot or does not make. In dealing, however, with the questions of the arrangement of the separate lets (number of habitable rooms, amount of additional accommodation, number of windows, existence of closets, etc.), as well as with those of internal installation (means of lighting and heating, running water, drains, lift, etc.), application is made sometimes to the occupier, sometimes to the proprietor. It appears, however, that the occupier actually living on the premises often knows these details better than the proprietor. The occupier is obviously also the source of information on all questions relating to housing conditions, the number of occupants of a dwelling, the relations between them, and the utilisation of rooms. The principle of the double source — owner and occupier — — 19 — is, however, not universally adopted. In certain countries the censuses are based solely on returns by the occupiers of dwellings 1. They are obviously unable to reply to certain questions, which are therefore not put. On other points there is danger of incorrect information being given ; in particular, where the structural bounds of a dwelling are not very distinctly drawn, the tenants may fail to distinguish between the actual dwellings rented by them and other parts of the premises of which they have the use. Further, all dwellings unoccupied at the date of the census will fail to figure in the returns unless special application is made to the owners in these cases. It is true that many of these gaps can be filled to some extent by the direct observations made by the census officer, as is done in certain countries 2. The method of applying to the proprietors alone is only rarely adopted 3, for the proprietors are really not in a position to furnish precise information on many fundamentally important points, particularly those relating to the use made of the premises tenanted. It must also be noted that the various censuses of dwellings are usually taken in conjunction with general population censuses, for it is convenient that both statistical returns should refer to the same date so that they can be compared with ease. There is also the technical advantage of the possibility of using for the collection of data the elaborate administrative machinery set up for the population censuses. On the other hand, the simultaneous compilation of two sorts of data also meets with certain difficulties, since the statistical offices are generally over-burdened with work at such times. 1 For example, Australia, Belgium, France, Italy, and t h e Union of South Africa. 2 In France, for instance, the census officer is expected to note on a separate form the number of blocks of buildings, the number of stories, the workshops, the premises used for both dwelling and business purposes, and vacant dwellings, and to group together for each building all the data obtained from the returns in regard to inhabited dwellings filled in by the tenants. 3 The Stockholm censuses are the best-known example of this form of procedure ; in t h a t city the material on which the enquiry rests consists of lists furnished by the owners of house property for the purpose of making census returns of the population ; these lists contain data regarding existing dwellings, but the responsible services themselves recognise t h a t by this method some inaccuracies pass unnoticed, particularly in the m a t t e r of the occupants of dwellings, a certain number of them not being known to the owners. (" Tenth General Census of Dwellings in Stockholm, 1920-1921. " Statistics of the City of Stockholm, V, New Series, No. 4, p. 21, X. Stockholm, 1922.) — 20 — Censuses of rents are often taken independently of other housing censuses, the enquiry either being conducted on a somewhat different basis or having to be undertaken for special reasons without waiting for the date of the next general census, or having to be undertaken periodically at fairly close intervals. There again, however, the owner of house property seems to be the most reliable source. For statistics of the housing market the method used in taking periodical censuses of vacant dwellings is either to require statements by the proprietors or to make a direct inspection of buildings containing dwellings, this being carried out either by the agents of the statistical services, or, as in many cases, by the staff of the local police. In the case of the statistics of supply of and demand for dwellings as registered by the housing offices, and those of the fluctuations in the total housing supply, the question, of choosing a source does not arise, because the source is here the very public office from which the statistics emanate. This is not always quite satisfactory, for as the compilation of the data is undertaken to fulfil administrative requirements, it often happens that those of a purely statistical nature are not taken into consideration. T H E PRESENTATION OF DATA In housing statistics the presentation of data usually consists in a classification of the basic units according to their different characteristics, for the phenomena under observation are most complex and require, above all things, detailed treatment. The determination of averages is very rarely - undertaken except in the statistics of rents — e.g. average rents for the different types of dwellings — while the calculation of percentage or other rates is almost entirely confined to statistics of housing conditions — e.g. rate of density of occupation. The establishment of the various classes must obviously differ according to the basic unit and characteristics under consideration ; this question is therefore reserved for the following chapters. There is, however, one characteristic which is common to all the basic units, necessitating a preliminary classification in all branches of housing statistics, viz. the geographical unit. — 21 — The term " preliminary classification " is used because it appears necessarily to underlie all others. The whole of the field covered by national statistics, even if limited to urban centres, is too heterogeneous for any satisfactory conclusions to be obtained from collective surveys in which local peculiarities are lost to sight. It is, therefore, desirable to subdivide the field into a series of smaller and consequently more homogeneous entities, within which the statistical analyses can bring out more clearly the real state of affairs. The entity obviously most suited for this purpose, which has been universally adopted, is that of the urban aggregation, a further reason for the choice being the fact that many housing problems require local treatment, which is generally a matter for the municipal authorities. As for the actual demarcation of these urban aggregations, it would be advantageous, as has been pointed out above, to set aside the often out-of-date political and administrative boundaries and to draw new boundaries more in agreement with the existing situation. Extending the principle of establishing geographical units of as homogeneous a character as possible, it would further be useful to subdivide town areas themselves into smaller and more homogeneous units. This system has already been frequently adopted in large towns, where its necessity has been recognised because of the very size of the urban aggregation in question. It has also been worked out in municipal statistics * in which local conditions 1 For instance, in the 1910 census of the city of H a m b u r g , then comprising 930,000 inhabitants, thirty-one subdivisions were made, and in t h a t of the city of Stockholm, which in 1921 had a population of about 420,000, sixteen parishes and thirty-four quarters were marked out. This subdivision has even been carried relatively further in investigations made in certain cities of medium size. In the census of the city of Zurich in 1910, with a population of 225,000, a division was made into not less t h a n twenty-three quarters, grouped in eight administrative areas ; and in t h a t of the city of Berne in 1920 the urban aggregation of 104,000 inhabitants was divided into five administrative areas and twenty-four quarters. One of the censuses in which the principle of subdivision was carried furthest was t h a t taken in the city of Warsaw in 1919, when its population was 846,000. The detailed statistical d a t a for the twenty-six separate administrative areas making up the whole city were published. Further, in the fifteen central areas, the principal data for 435 blocks of houses were published separately. These blocks, the size of which varied with the density of their population, were defined in such a way as to form as homogeneous wholes as possible ; the number of occupants contained in each -block varied between a few score and ten thousand, but was in most instances one or two thousand. — 22 — are usually studied with greater care than in national statistics. It should nevertheless be noted that, with certain exceptions, these subdivisions have been made on the basis of administrative subdivisions (administrative area, district, quarter), which are not always as homogeneous as might be wished. It seems then that here, again, it would be better to have recourse to a method of subdivision worked out for purely statistical purposes. Typical zones might be marked out according to their predominant structural characteristics. Continuous structures (blocks of houses with several stories), semi-detached buildings (houses of small elevation with gardens either front or back), and detached buildings (houses of moderate elevation standing in their own grounds). Possibly the business and factory quarters ought also to be divided off. As regards the actual size of these units, it seems to be impossible to lay down any rule, for this question depends on the more or less homogeneous character of the aggregation. Generally speaking, it would however be better to adopt rather small units so as to render them as homogeneous as possible. CHAPTER II STATISTICS OF DWELLINGS Censuses of dwellings are the fundamental basis of housing statistics. Before the war several countries compiled them, either nationally, as in Australia, Austria, Denmark, Canada, France, Great Britain, Norway, Sweden, the Union of South Africa, and the United States ; or by different municipalities, but on the same principles, as in Germany, Italy, and Switzerland. Since the war the compilation of these statistics has been extended. Bulgaria, Finland, Germany, Hungary, the Netherlands, Poland, and Switzerland now belong to the group in which these data are collected on a national basis. In Italy, the enquiries undertaken by the municipalities have been made more detailed and their standardisation more complete. In Central and Eastern Europe detailed investigations are undertaken in certain large towns such as Lodz, Prague, Riga, Vienna and Warsaw. Statistics of dwellings consist essentially in enumeration of the basic units — the dwellings — and their classification according to various characteristics. The problems to be solved, therefore, are above all those of defining the basic unit, and choosing and defining the characteristics to be studied. DEFINITION OF A DWELLING It is of the first importance to define from the outset the meaning of the term " dwelling ", the home of the household, for it may be interpreted in many different ways. In some cases a very clear distinction is first drawn between dwellings lived in by households of a family nature, consisting of members of one and the same family, to whom may be added more distant relatives, friends, domestic servants, and boarders ; — 24 — and those used to accommodate groups of individuals whose connection with each other may be either social (educational institutions, religious bodies, various societies, barracks, hospitals, prisons, etc.), or economic (hotels, boarding-houses, lodginghouses, etc.). This distinction does not seem expedient here, however, for it refers more to the classification of the data than to the definition of the basic unit. It seems preferable to define the latter as widely as possible, and to include both family dwellings and those not of a family character. The question of the relation between the home and the household, whether of a family character or not, is more difficult. In theory it may be dealt with in two different ways : either by considering the intended use of the premises or by considering only the actual use made of them. If the first conception is adopted, the principal factor is the arrangement of the premises, and a dwelling is considered to be the collection of premises forming a whole, separated from others, and making an independent domestic life possible for the household for which it is intended. Use is therefore made of purely structural criteria : separate entrance from the street or from a common landing or passage, separation from other dwellings by a complete system of party-walls. With the second conception, a dwelling is deemed to be the premises actually occupied by a household at the time of the investigation, no account being taken of the distribution and arrangement of the premises. The criterion is therefore purely social — that of the household occupying the premises. The differences arising out of these two conceptions are clear from the outset. In the first place, certain premises will be counted as dwellings if one conception is used, but not with the other. Taking the idea of intended use, for instance, dwellings will be considered to include any premises intended for living in but still vacant, as also those temporarily used for other purposes although no lasting structural alterations have been made (flats used wholly as offices or shops, or some rooms of an inhabited flat used for a workshop, office, or other business purpose) ; but these will be excluded if the basis is the actual use made of the premises. On the other hand, this second principle would mean the inclusion of temporary constructions and arrangements for spending the day or night in workshops, shops, workplaces, coach-houses, stables, etc., which cannot be considered premises intended for — 25 — living in. Further, if two or more households share premises originally intended for one household only and there has been no change in the arrangement of the premises, the second conception will mean that the number of dwellings is considered to be the same as that of households, whereas the first conception gives only one dwelling. Conversely, if two or more separate flats are occupied by one household, the number of dwellings recorded will depend on the definition adopted. In practice, neither definition has seemed sufficiently complete for general adoption, and as a rule a certain compromise has been reached. In some countries, the intended use of the premises forms the starting point, but at the same time an attempt is made, first, to include premises which, though not intended for living in, are used for this, and, secondly, to exclude those which, although intended for living in, are, used for other purposes 1. But if only some rooms in a dwelling otherwise inhabited are used for other purposes, as in the case of homeworking trades and offices run in the home, the dwelling is counted as such, but the rooms used for other purposes are 1 The English census of 1921 defined " a structurally separate dwelling " as " any room or set of rooms, intended or used for habitation, having separate access either to the street or to a common landing or staircase. Thus each flat in a block of flats is a separate unit ; a private house which has not been structurally subdivided is similarly a single unit whether occupied by one family or by several families. But where a private house has been subdivided into maisonnettes or portions, each having its front door opening on the street or on to a common landing or staircase to which visitors have access, then each such portion is treated as a separate unit " (Census of England and Wales, 1921, General Tables, p. 82). The Italian census of 1911 also used the structural definition of a dwelling. The Ministerial instructions defined a dwelling as " a set of rooms, apartments and alcoves, or a single room, used for the accommodation of one or more families " (Ugo GIUSTI : L'addensamento e l'affolamento nei centri urbani al 10 giugnio, 1911 ; Union of the Statistical Offices of Italian Municipalities, Monographs and Studies, No. 2, Florence, 1913, p. 61), and the census form included a question to show whether several families lived together in the same dwelling. The Union of Statistical Offices also adopted this view ; in its supplementary enquiry undertaken on the occasion of the 1911 census, it gave as the definition of a dwelling : " a set of rooms or a single room with access separate from other rooms " (ibid.), and this definition was maintained when a similar enquiry was undertaken on the occasion of the 1921 census. Similarly, in the Polish census of 1921, a dwelling was considered to be " any part of the building which is definitely isolated from other similar parts and is intended exclusively for habitation ". " The definition of the dwelling must, therefore, not be confused with that of the household. On the contrary, the decisive factor in the definition of the dwelling is the technical one " (The First General Census of the Polish Republic, of 30 September 1921 : " Dwellings, Population, Occu. — 26 — deducted, and in some cases form the subject of special investigation \ In other countries, on the other hand, the idea of actual use is preferred, but the definition of a dwelling is taken to include premises which, although not used, are intended for habitation and could be so used at any moment, in other words, vacant dwellings 2. It is obvious that for the delimitation of these data, the criterion of actual use cannot be applied and recourse must therefore be had to structural criteria. It was probably on the basis of this experience that in 1913 Dr. van der Borght supported a similar view in his report on the question of standardising housing statistics which he submitted to the Vienna Session of the International Institute of Statistics. The Institute took no decision on this point, however, and appointed a committee to continue the study of the question, but its work was interrupted by the war. In a few cases, finally, the compromise has been carried even pations " , p. 5 ; Statistics of Poland, published by the Central Statistical Office of t h e Polish Republic). Finally, Germany has quite recently adopted t h e same view. The forms prepared for most of t h e municipal censuses taken in 1925 a n d 1926 a n d those for t h e national census of 1927 state t h a t t h e whole dwelling must be taken into account including rooms transferred t o a sub-tenant or co-tenant, " and all living rooms must be counted, whether they can be heated or not, and whether actually used as such or not " (official foTm for the national census). Other examples are Austria (1910), Finland (1919), Norway (1910 and 1920), Prague (1921), etc. 1 In the Austrian census of 1910, for instance, dwellings were classified according t o t h e new use made of these rooms. The classification was as follows : textile industry, manufacture and cleaning of clothing, wood, commerce and transport, liberal professions. 2 In France, in the 1901 a n d following censuses, the definition of t h e dwelling coincided with t h a t of t h e household, a n d i t was stipulated t h a t " the number of dwellings occupied must strictly equal the number of households " (MINISTRY O F LABOUR AND SOCIAL W E L F A R E : Résultats statistiques du recensement général de la population effectué le 24 mars 1901, Vol. V ; Paris, 1907, p . 34). In Belgium, t h e instructions for t h e 1920 census defined a dwelling as " the house or p a r t of t h e house serving for the accommodation of t h e household " (MINISTRY OF THE INTERIOR AND H E A L T H : Recensement général du 31 décembre 1920 : Recensement des logements dans les communes de 10,000 habitants et plus ; Brussels, 1925, p . 7). In Germany, most places which undertook housing enquiries on the occasion of the general censuses of 1905 and 1910 also adopted this point of view. Almost t h e only exceptions in t h e 1905 census were a few towns in Bavaria. This view was maintained for t h e national housing census for 1918, b u t for t h e national census of 1927, it was given u p and, as already stated, replaced by t h a t of t h e structural unit. — .27 — further, and both conceptions are adopted side by side, but these cases are on the whole exceptional *. It will no doubt be observed that although these various compromises start from different standpoints, the result arrived a t is much the same. There are differences only in two respects. In the first place, when a dwelling is used partly for living in and partly for other purposes, the uninhabited rooms are completely excluded from the statistics if these are. based on the idea of actual use, but they are merely treated separately if the idea of intended use is considered more important. The divergence is insignificant, however, for in any case the inhabited rooms are always counted, and these, after all, are the most important for the final analysis. The second difference is more important. It arises when two or more households live in a dwelling intended for only one, or two or more flats are occupied by only one household, for it will depend on the compromise adopted whether the households or the dwellings as structural units are counted. It is possible that this difference involved no serious disadvantages before the war, for the housing shortage was not so severe as now, and the cases in which one flat was shared by several families were, on the whole, rare. But since the war the situation has changed, and there are many more cases of several families living together in one dwelling, this being, in fact, one of the symptoms of the crisis which has to be brought out. The change which has taken place in Germany almost universally, from the household conception of the dwelling to the structural conception, is particularly significant in this respect. It may further be noted that in certain statistics a third definition of the dwelling is used, namely, a legal definition, according to which a dwelling consists of the premises covered 1 The best-known examples oí this system are to be found in the Swiss censuses. The few towns which undertook a housing enquiry on the occasion of the general population census of 1910, followed a decision of the Union of Swiss Statistical Offices and adopted as their basic unit the dwelling used by the household, but the form in which the schedules were drawn up made it possible also to determine the number of dwellings as structural units. This number was therefore calculated for certain towns as an accessory item of information, and the same procedure was applied to the cases of several families living together in one dwelling. When the 1920 federal census was taken, the dwelling, as defined by the household occupying it, also served as the basic unit for the statistics, but at the same time, a special heading was given to cover the cases of several families living together in a single flat, that is to say, the structural unit was used. — •28 — by a lease. According to circumstances this unit will obviously tend to coincide sometimes with the structural unit, sometimes with the social unit, without exactly corresponding to either. For statistics of dwellings, this conception has the great disadvantage of restricting the investigation to leased premises and to exclude all those occupied by their owners or left vacant* unless these are dealt with by analogy or on the basis of expired leases. It may be added that this system is seldom used for statistics of dwellings \ It is, on the other hand, more useful» and therefore more used, for rent statistics. The choice between these various definitions of dwellings and the different ways of making a compromise depend obviously on the object of the statistics. This may be conceived in two ways ; that of determining the existing supply of premises equipped for habitation and their characteristics, with a view to ascertaining how far it can meet the demand ; and that of determining the supply of premises actually inhabited and their characteristics, with a view to ascertaining the housing conditions offered by these premises. If the object in view is the first, undoubtedly the idea of intended use must be employed, while, on the contrary, if it is the second, it will have to be principally the idea of actual use. Apart from these theoretical considerations, however, certain technical questions also arise. Thus, the definition of the dwelling as the premises occupied by a household seems often to have been adopted because of its convenience in practice. Censuses of dwellings are in fact generally undertaken on the occasion of general population censuses, the data for which are usually obtained on household returns. The household unit is therefore defined with care for the purposes of the general census, and the demarcation of the dwelling unit follows automatically. The definition of the dwelling as a structural unit, on the contrary, presents certain difficulties in practice which have often 1 Almost the only examples t h a t can be quoted are the housing enquiries undertaken every five years in Stockholm, where a dwelling is defined as " t h e premises or groups of premises for habitation at t h e disposal of any one person " {Statistics of the Municipality of Stockholm ;. V, " Buildings and Dwellings " ; the Ninth General Housing Census of Stockholm of 31 December 1915, New Series 3, Stockholm, 1917, p. 16), and the Warsaw housing census of 1919, in which a dwelling wasdefined as " the whole set of rooms let directly to a principal t e n a n t " (Results of the Census of Buildings and Dwellings in Warsaw in 1919 •„ Vol. II, " Statistics of Buildings ", p. 112). — 29 — led to the rejection of this system. The delimitation of a dwelling solely on the basis of structural arrangements is, in fact, a difficult and even arbitrary operation if the grouping of the rooms is ill-defined. Now, it often happens that old houses, more or less satisfactorily transformed, contain dwellings divided up among the different parts of the building. Even in modern buildings the servants' rooms and box rooms are often in the attics, the closets in the passages, and the laundry accommodation in the cellar or outbuildings. Moreover, these last two types of rooms are often common to several flats. On the other hand, if the definition of a dwelling is based on the premises occupied by a household, all these cases are settled very simply and in accordance with the existing situation, the number of dwellings being taken to be equal to that of separate households. Even this last method has its difficulties, however, and other border-line cases are met with which are not so easy to solve. In the first place, the exact determination of the household, which is easy when it consists solely of a family, even including more distant relatives, friends, or boarders, who share entirely in the life of the family, becomes much more difficult if there are more or less independent sub-tenants. There is, in fact, a whole scale of possible systems, from the complete independence of the sub-tenant to his complete participation in the domestic life of his hosts, and the settlement of the intermediate cases is always open to question. For instance, a sub-tenant who does not board with the family, but whose room is looked after by one of the members of the household with whom he lodges, may be treated either as a partial boarder and therefore included with the household, or as an independent sub-tenant whose room is looked after by a charwoman. In practice, however, it cannot be said that any of these difficulties are really insurmountable. If, therefore, it is proposed to establish standards for dwelling statistics which may be accepted by all countries and yet allow of precise international comparison, it is neccessary to draw up a definition which will cover both conceptions and satisfy both the objects referred to above without sacrificing one to the other. Within the covering definition or framework so obtained, national statistics will then be fitted so far as it is considered necessary to develop them with a view to meeting each of the two ends. Thus, a very general definition must be adopted for the dwelling, based on the ideas of both intended and actual use, after — 30 — which, if necessary, various distinctions may be made according to one or other of these criteria. A dwelling will, therefore, be taken to mean any premises intended for occupation by a household or used for this purpose. But to prevent contradiction between these two conceptions, for instance, in the case of a dwelling shared by several households, it seems essential that one conception should take precedence of the other. For present purposes, that of intended use would appear preferable. The intended use of any given premises reveals their permanent properties, whereas the actual use to which they are put is only a more or less transitional state. Moreover, it should be observed that in another branch of housing statistics, that of building activity, the basic unit of the dwelling can only be defined according to the intended use, for there may not have been any acutal use ; and it is important that, so far as possible, the same point of view should be adopted in determining the basic units for the different branches of housing statistics. On this principle the structural criteria usually adopted to define the basic unit are : the existence of a separate entry to t h e premises, either from the street or from a passage or staircase open to the public, and the complete separation of the premises from other neighbouring premises. These seem satisfactory enough for acceptance as an international standard. Yet these criteria are not completely adequate, for, as has already been pointed out, the dwelling is not always a welldefined structural whole, and the strict application of the criterion of separate entry would mean that rooms separated from the principal building would be considered as so many distinct dwellings, an obviously absurd result. It may also happen that premises originally intended for other than living purposes are sometimes used more or less provisionally for this purpose. As it is characteristic of most of these arrangements that they occupy only part of the structural unit, the rest being used for its original purpose, structural criteria cannot apply here. In both cases, therefore, it will be necessary to employ the idea of actual use and to define the dwelling on this basis, that is to say, by the fact that it is occupied by a household at the time of the enquiry. To conclude, the definition of a dwelling may be given in the following terms : a dwelling is in principle constituted by a separate set of premises which, whatever their use, are — 31 — intended for the occupation of a household and have an independent entrance either from the street or from a passage or staircase open to the public. Premises separated off from these but clearly intended to form part of the dwelling will be included with it. Premises not intended for living in, but used for this, will be considered to have as many dwellings as there are households occupying them. CLASSIFICATION OF D W E L L I N G S Classification according to Intended and Actual Use The definition of a dwelling suggested here is obviously very general, much more so than that accepted in most statistics. A twofold distinction is therefore called for, according to the intended and actual use of the premises, and on this basis certain large categories should be established, so as to subdivide the general international framework. In countries where so wide a definition of dwellings as that proposed here cannot yet be accepted, the statistics will only leave out one or two of these categories, while dealing fully with the rest ; so that for these at least there will be a possibility of exact comparison. Further, this classification will make it possible to compare the actual use made of the premises with their original purpose, and information of this kind will be of great value in determining how far the existing supply of dwellings meets the needs of the population. The following classification may therefore be suggested ; its elements, it may be added, already exist in several countries. A first distinction will be made according to the intended use of the premises, those intended for habitation being separated from those originally intended for other purposes. In the first group account will then be taken of the nature of the household for which the dwelling is intended, and two classes may be distinguished — dwellings intended for family habitation and those intended for other than family habitation. These distinctions, it should be observed, would be made only on the basis of purely structural criteria, the arrangement and equipment of the premises. The operation will no doubt . — 32• — prove difficult at times, but it does not seem impossible. Premises not intended for living in are characterised, in fact, by the absence of the criteria indicating those intended for habitation. The principal features of dwellings intended for other than family groups are first their size, then the presence of relatively large common rooms (refectories, dormitories, recreation rooms, workrooms, etc.), combined, as the case may be, with the existence of a large number of small rooms equipped for one or more persons. These three main classes, divided according to intended use, may then be further subdivided according to actual use, the first distinction made among the dwellings intended for families being between those entirely used for this purpose, those used partly for this purpose and partly for other purposes, those not used (i.e. vacant), those entirely used for other purposes than living in, and those used for habitation by other than family groups. It will no doubt be difficult to determine this last class, for it is sometimes not so easy to make a clear distinction between families taking boarders and certain boarding-houses or the living-in arrangements made by employers for their workers. Certain rules ought, therefore, to be laid down for this purpose, as was done, for instance, in the English 1921 census, in which the dwellings of private families were taken to include all those in which the number of boarders did not exceed the number of members of the employer's or householder's family, including private domestic servants. Among the dwellings not intended for family habitation, a distinction will simply be made between those used as such and those left empty. Classes corresponding to those proposed for family dwellings might no doubt be defined, but the cases would be too few for such a distinction to be of any interest. It should be added that in the group of dwellings for other than family habitation any premises reserved for the household of the manager should be counted separately, as is done in certain countries, and should be included with family dwellings properly so called. This operation is obviously somewhat difficult, but does not seem altogether impracticable. Where premises intended for other purposes than habitation are used for living in, they will be classed simply according as they are used for family habitation or not. — 33 — To sum up, the following classes will be obtained : (1) Dwellings intended for family habitation : (a) used solely for this purpose ; (b) used partly for this and partly for other purposes ; (c) not used — i.e. v a c a n t ; (rf) temporarily used for other purposes t h a n habitation (without structural alteration) ; (e) used for other t h a n family habitation. (2) Dwellings intended for non-family habitation : (/) used accordingly ; (g) used for family habitation ; (ft) not used — i.e. vacant. (3) Premises not intended for habitation : (i) b u t used for family habitation ; (/) b u t used for other purposes than family habitation. It will clearly depend on the object of the statistics how far any further analysis of the data in any of these classes is carried. Thus a study of the conditions of comfort and hygiene of dwellings may well be limited to premises actually used for habitation, for as a rule such an analysis is undertaken only to discover the conditions in which the population actually lives. It will, therefore, leave out of account vacant dwellings, whether intended for family habitation or not, and dwellings used for other purposes than habitation. Moreover, dwellings used partly for living in and partly for other purposes may be treated as smaller dwellings used altogether for living in, the rooms used for working having been subtracted. Premises intended for other purposes than habitation but used for living in will be treated in the same way as other dwellings. Family dwellings used for other than family groups will be counted in the latter class. To sum up, in this case only two large categories will be required : (1) occupied premises intended or used for family habitation (covering categories (a), (b), (g) and (i) in the previous classification), (2) occupied premises intended or used for other than family habitation (covering categories (e), (/) and (/)). Classification according to Size After these first and most important classifications, one of the most fundamental characteristics of dwellings to be the subject of statistical analysis is that of size, for the considerable differences between them from this point of view are of — 34 — essential practical importance. In actual fact this classification of dwellings by size dominates all dwelling statistics. It is to be found also in all other housing statistics in which the basic unit is the dwelling, but the manner in which it is carried out still differs considerably, and it is desirable to remove these differences as much as possible. It will be necessary, however, to distinguish here between family dwellings and those occupied by other than family groups, for the question of size is not exactly the same in the two cases. The simplest measure of the size of a family dwelling is the number of rooms. All the rooms in a dwelling, however, cannot be regarded as equivalent units for measuring its size, as they differ considerably among themselves. A selection is therefore necessary. It has been carried out in most countries, but the criteria vary considerably. Thus, following the example given by one of the first important housing enquiries, that made in Berlin in 1864, several municipal statistics have for long used as a unit for measuring the size of dwellings, the number of rooms that can be heated 1. This method has several disadvantages, the chief of which is t h a t the relative number of rooms which cannot be heated, and are therefore not covered by the returns although actually used for living in, varies from one region to another according to climatic conditions and local customs. Moreover, the very idea of a room which can be heated may be interpreted in different ways 2. This definition has gradually been abandoned 3. In other countries the criterion adopted was that of minimum 1 For example, the municipal statistics of Germany and Switzerland. Thus, in Germany, municipalities in northern districts counted as rooms that can be heated those which actually have some means of heating, but in Western and Southern Germany, where the tenant, and not the landlord, provided the heating installation, this criterion could not be allowed, for many tenants did not undertake the expense of installing heating in all the rooms where this was possible. Accordingly, the criterion adopted was that of the number of rooms in which a stove could be installed without structural alterations. 3 In Germany, where the method was most widely used, the Association of German Municipal Statisticians recommended at its twentyfourth Conference, held in 1909, that dwellings should be classified according to the number of living-rooms (Wohnräume), distinguishing between three classes : rooms which can be heated, rooms which cannot be heated, and kitchens. The distinction between rooms which can and cannot be heated was subsequently dropped altogether, and in 1927 the definition of a living-room was made on an entirely different basis (see below). 2 — 35 size, taking as a standard either the space needed to contain a bed for an adult ' o r a certain clearly defined surface 2. The too strict application of this criterion has the disadvantage, however, of including among living-rooms certain fairly cramped premises, such as alcoves, as well as those not used for living in or only in a subsidiary sense, such as laundry accommodation, corridors, box-rooms, warehouses, pantries, etc. This latter disadvantage, it is true, is usually avoided in practice by the express stipulation that certain types of rooms shall not be counted as living-rooms. Yet another criterion which has been adopted in certain cases is that of the existence of a window. This method suffers from the opposite disadvantage to the last, for it definitely excludes alcoves and partitions without windows. In addition, it entails a risk of counting subsidiary rooms under the same head as the principal living-rooms. In other cases, finally, the types of rooms to be counted are merely enumerated with reference to their intended or actual use, but no definite criterion is given, or, if so, it is only to settle doubtful cases 3. From the international point of view, it seems that this last conception should be preferred, for the adoption of very definite criteria is hardly possible, as the significance of each will vary 1 Thus, in the French censuses : " A room will be deemed to be any division of a house, intended for living in, and separated from the rest by partitions reaching to the ceiling, and capable of holding a bed for an adult (including t h e kitchen, hall, and dressing rooms large enough to contain a bed, servants' rooms, even if separated from the dwelling, b u t not closets ; excluding shops, workshops, stables, etc., unless one or more persons usually spend the night there) " (Resultats statistiques du recensement général de la population effectuée le 4 mars 1906, Vol. I, P a r t I, p. 23). The procedure followed for the enquiry of the Union of Statistical Offices of the Italian Municipalities in 1921 was similar. " A livingroom will be deemed to be any space, whatever the use made of it, which can contain a bed (except the rooms specified in detail)" (Bollettino dell' Unione statistica delle città italiane, Eighth Year, Nos. 2-4). 2 In the statistics for the municipality of Amsterdam, any enclosed space of not less t h a n four square metres is considered to be a room. 3 The German census of 1927 offers one of the best examples of this method : " Living-rooms must be taken to include all rooms, whether they can be heated or not, intended for lasting occupation by persons (whether for living or sleeping in), regardless of whether they were used for living or sleeping in on the date of the census or not. The following must not be included as living-rooms : the kitchen, the bath-room (unless it is used for living in), wardrobes, pantries, alcoves, passages, box-rooms, and every place in which at least one bed for an adult cannot be p u t " (official census form). — 36 — from one country to another owing to differences in local customs and climatic conditions. It therefore seems preferable to consider primarily the intended use of the premises, and to count as units for measuring the size of the dwelling only living-rooms properly so called — that is to say, those intended actually for living in, such as bedrooms, dining-rooms and living-rooms in the restricted sense. This method seems clear, but certain rooms raise difficulties of interpretation. The kitchen is a particularly difficult case in point, for the character of this room differs widely according to the region, the social class for which the dwelling is intended, and even the date at which it was constructed. Sometimes it contains only the equipment needed for preparing food and for household work, sometimes it serves as a general room, a diningroom, a bedroom for the servants or members of the family. The practice of the different countries differs accordingly as well. In some cases the kitchen is included with the livingrooms 1. In others it is counted separately under a special heading 2 (although, in point of fact, there does not seem to be any particular connection between the method adopted and the degree of specialisation of this room). In some cases, finally, the rather important distinction is made between dwellings with a kitchen of their own and those with only a kitchen common to several. The best method would no doubt be to distinguish systematically between kitchens used exclusively as such and those used partly for living in, but it seems impossible to carry out this demarcation adequately in practice. It is, therefore, probably better always to take the kitchen into account as a unit of size, but always to place it under a separate heading, owing to its special importance and nature. This gives the further advantage of allowing a distinction to be made between dwellings with a kitchen and those without. Among other rooms whose inclusion with living-rooms is open to question, reference should also be made to servants' rooms and attics, which in practice are sometimes counted as 1 For example, the national censuses of France, 1911, Great Britain, 1921, Belgium, 1920 ; the enquiry of the Union of Statistical Offices of the Italian Municipalities, 1921 ; and the municipal census of Warsaw, 1919. 2 National censuses of Austria, 1910, Norway, 1920, and Switzerland, 1920 ; municipal censuses of Stockholm, 1910-1920, Prague, 1920, and most German and Swiss towns. — 37 — living-rooms, sometimes as secondary rooms, and sometimes left out of the statistics altogether"1. It seems, however, that from the international point of view it would be better to include them whenever they are really habitable. Even though they are not of the same value as other living-rooms, the error in this case seems less serious than if they are left out altogether. On the other hand, alcoves, lobbies, and rooms intended for other particular purposes (bath-rooms, closets, laundries, etc.) should not be counted as living-rooms. Even with an exact and uniform definition of a living-room, such as that suggested above, the problem of measuring the size of a dwelling is not solved altogether, for a living-room is in itself no more than an imperfect unit of measurement. On the one, hand, this system leaves entirely out of account all subsidiary rooms, whose existence may yet sensibly affect the capacity of the dwelling, and, on the other hand, it includes under the head of living-rooms, rooms which may differ widely between themselves as regards size, conditions of ventilation, and equipment. An attempt has therefore been made in several countries to meet this difficulty by carrying the statistical analysis of dwellings somewhat further than a simple classification by the number of living-rooms. In the first place an attempt is made to extend the investigation to the more important subsidiary rooms : closets, bathrooms, laundries, pantries, lobbies, as also servants' rooms and attics (if the latter are not included with living-rooms). It is true that this involves complicating the enquiry forms by. a large number of supplementary questions, and that if it is subsequently desired to classify these dwellings with reference to the existence or absence of each of these various subsidiary rooms, it will be necessary to multiply the classes of dwelling to an extent which means a considerable increase in the work involved, besides making the tables difficult to read. This distinction between the different subsidiary rooms is not absolutely indispensable, however, if they are only considered as a means of measuring the size of the dwelling. In 1 Reference may be made, for instance, to the French census of 1911, and that of Prague, 1921, in which servants' rooms were counted as living-rooms, and the Belgian census, 1920, and to some extent the Swiss census of 1920, in which attics were also counted as living-rooms. It may be added that in several cases no special mention is made of these rooms, so that it is not certain what procedure is actually adopted. — 38 — countries where it is not desired to study in greater detail the comfort provided by dwellings, it will therefore be sufficient to select a certain number of subsidiary rooms of particular importance from the point of view of the size of the dwelling, and to count them without special reference to their particular characteristics. Such a selection would refer in the first place to rooms intended for special purposes (closets, bath-rooms, laundries), then to those which, owing to the absence of means of ventilation (alcoves, lobbies, box-rooms), cannot be counted as living-rooms, with perhaps a limitation based on size (here the criterion of a bed for an adult might be used). The dwellings would then be classified not only by the number of livingrooms, but also by that of subsidiary rooms. Perhaps it would also be advisable to reduce these two definitions to a common measure. Thus, a subsidiary room might be counted as a fraction of a living-room, either on a standard basis—for instance, one-half •— fixed arbitrarily for all subsidiary rooms, or on a scale based on definite criteria. In other cases the study of the size of dwellings has been carried much further, and various measurements have even been taken on the spot with a view to determining their exact area, or rather the volume of the whole dwelling, or of the living-rooms only, or even of the rooms used for sleeping in. Enquiries of this kind were undertaken, particularly before the war, in certain Austrian, German, Scandinavian, and Swiss towns, but since then they have been continued on a much smaller scale, owing to their very burdensome nature 1. They would be worth resuming, however, for they alone allow of a really exact view of the size of a dwelling. Here the method of sampling might be used with advantage. There are, in fact, in different regions, according to the date at which the building was constructed and to the social class of the persons for whom it is intended, certain current types of size, which might be determined at no great expense on the sampling system. Once these types were determined, the interpretation of the data on the number of rooms in dwellings could be carried out much more certainly. 1 The work of measuring the different rooms in dwellings is in fact «onsiderable, and can be entrusted only to persons specially trained for this purpose. The experience gained at Nuremberg in 1907, for instance, showed that, to obtain a detailed measurement and description of the rooms, it had to be reckoned that two investigators (one technical expert and one secretary) could examine only 25 dwellings a day. — 39 — The whole question of determining the size of dwellings by the number of their rooms is somewhat different in the case of those used for other than family groups. Here, the room is not so satisfactory a unit of measurement as for family d wellings, for, in addition to living-rooms of ordinary size, there are much larger halls (dormitories, refectories, common workrooms) which can hardly be treated as ordinary rooms. In fact this is one of the characteristics of this type of building. A systematic distinction between the two types of premises and the adoption for this purpose of criteria of demarcation are therefore needed. Hitherto, however, this question has not been carried very far, for these collective dwellings have not attracted much attention among statisticians, and have seldom been made the subject of detailed analysis. Classification according to Other Criteria A statistical analysis of dwellings may also refer to other characteristics than that of size. Such an analysis will be carried out to fulfil the second object of statistics, that of studying the conditions in premises actually occupied, and will therefore be divided into the two principal categories established for this purpose : dwellings intended or actually used for family habitation, and those intended or actually used for habitation by other than family groups. The analysis is, as a rule, of a summary kind for dwellings in the second category (unless these are treated at the same time as family dwellings, in which case they will not appear under a special heading). At the most it is sought to classify them according to their nature, which varies fundamentally. Moreover, this classification follows directly from that adopted for the statistics of buildings, and may relate, for instance, to the following classes : hotels and boarding-houses, almshouses and the like, hospitals and nursing homes, educational institutions, religious institutions, barracks, prisons. Dwellings intended or actually used for family habitation, on the contrary, are frequently analysed from the point of view of hygiene, comfort, and convenience, and several criteria have been chosen for this purpose. One of the most significant is the existence of certain subsidiary rooms. For that purpose it is no longer sufficient to count up indifferently the various — 40 — subsidiary rooms in each dwelling, as was suggested for the study of the size of the dwellings, but the existence of particular rooms will have to be specified exactly. This has, however, the disadvantage already mentioned of encumbering the census forms and complicating the classification, for the number of classes may become excessive. The method has, therefore, been adopted only with caution, and is limited as a rule to certain more characteristic rooms, i.e. closets and bath-rooms ; in some cases it is stated whether the use of these rooms is confined to one dwelling or is common to several. As regards closets, sanitary experts propose a distinction between those flushed with water and those connected with a system of sewer mains. On these lines dwellings are sometimes also classified according to their domestic equipment ; means of heating and lighting, installation of running water, means of ventilation (number of windows per room). A question which is given much more attention is that of the position of the dwelling in the building or with respect to the street. Obviously the manner of dealing with this problem will differ with the type of construction most usual in the region. If small, low houses predominate, dwellings will be classified according as they are to be found in houses with one or several households, and sometimes account will also be taken of whether they adjoin a garden. If buildings with several flats are more usual, dwellings will be classified according to the story. If such buildings are constructed in large blocks, a distinction will also be made between the dwellings facing the street, those facing side-streets and those facing the courtyard at the back. Another feature sometimes taken into account is that of the judicial relation between the dwellings and their occupants ; a distinction is generally made between rented dwellings, the owner's own dwelling, service dwellings, and those occupied free of charge. In the group of rented dwellings, it would also be useful to separate those let furnished, as well as the parts of the dwellings or isolated rooms sublet by the principal tenant. All these various data are certainly of great interest, for they give some measure of the standard of hygiene, comfort and convenience of the dwellings. But their collection on the occasion of a census is a very heavy burden, which the authorities may hesitate to undertake. Here, too, it seems that they — 41 — might be obtained with sufficient precision by the method of sampling, which means a great economy in work and costs. The regular collection of such information would thus be possible without burdening the budgets of statistical services which are always comparatively small. It would thus appear that for statistics of dwellings recourse should not be had only to the census method, as is very generally the case, but that it would be preferable to limit the collection of the data by this method to questions on which it is important to have altogether exact information, and to use the method of sampling for the study of all other more or less secondary questions. On this system the census would therefore be limited to enumerating dwellings and classifying them, first according to their intended and actual use, and then according to the size (measured by the number of living-rooms, and, if need be, of certain subsidiary rooms). On the occasion of the census a partial enquiry on the sampling method would be undertaken, which would, in the first place, supplement the information on the size of the dwellings by determining the types of size of different living-rooms, and, secondly, determine the standards of health and comfort of the dwellings by examining the types of the various subsidiary rooms, the equipment of the premises, the position of the dwelling in the building and with respect to the street, the neighbourhood of gardens, etc. CHAPTER III STATISTICS OF BUILDINGS WITH DWELLINGS AND OF REGISTERED PROPERTIES WITH DWELLINGS STATISTICS OF B U I L D I N G S W I T H DWELLINGS As a dwelling is conditioned structurally by the building in which it is placed, the characteristics of that building necessarily help to determine those of the dwelling. These points may be studied by means of the statistics of buildings with dwellings. Moreover, if in a district the most ordinary type of building is that of the one-family house, the building may be assimilated with the dwelling itself, and to some extent the statistics of buildings with dwellings will give the same information as that to be obtained from the statistics of dwellings. For these two reasons housing statistics are usually made to include statistics of buildings with dwellings, which from other points of view belong more to the field of town-planning than that of housing policy. The statistics of buildings with dwellings are, moreover, technically related to those of dwellings ; for the source of the information on buildings, that is to say, the owners of the buildings, is the same as that used for obtaining certain information on the dwellings. It is therefore expedient to combine the two operations, and as a rule a single form is used. Most countries with statistics of dwellings also compile statistics of buildings. In certain cases even, the latter were compiled on a national basis before the former, as a record of buildings with dwellings was made by the officers engaged in taking a general census of the population. This was the case before the war, for instance, in Germany, Hungary, the Netherlands, Switzerland, as also in the United States of America, — 43 — Canada and India. And in the three latter countries the position has remained unchanged since the war as well 1. It should be remembered that the statistics of buildings with dwellings form only a part of the statistics of buildings, which cover all existing buildings. But as this is the part which alone belongs to housing statistics, it is all that has to be considered here. It has been thought necessary to make this distinction, which is in point of fact very generally practised. Many investigations are limited to buildings with dwellings, and those which cover all constructions make a definite separation between the two types of buildings. Definition of Buildings with Dwellings The statistics of buildings with dwellings consist essentially in the enumeration of such buildings and their classification according to certain characteristics. It is generally agreed that a building is to be considered any covered construction, whether isolated or separated from others by party-walls reaching from the foundations to the roof. On this point hardly any difficulties arise, but the distinction between buildings with dwellings and the rest is rather more delicate. As for the statistics of dwellings, use may be had of two different criteria — those of the intended and of the actual use of the buildings. The first of these criteria meets with practical difficulties, for it is not always clear for what a building was intended. There can obviously be no doubt in the case of buildings intended solely for habitation, nor even in that of mixed buildings in which dwellings occupy a comparatively important place. The position is less obvious for business premises (offices, shops, warehouses, factories) in which some rooms have been reserved for the occupation of a caretaker, and may easily be passed over, especially if they are temporarily vacant. Moreover, the criterion of intended use, taken by itself, will completely exclude all buildings which, although intended for other purposes, are actually used more or less 1 As regards India, however, a» attempt is being made gradually to substitute the census of dwellings for that of buildings, the dwelling being defined on the social principle of the premises occupied by a householder. This substitution is being made gradually in the different provinces according to local possibilities, so that at present the existing »data are of a hybrid nature. — 44 — provisionally for living in, without any structural alteration in the building having been made. In consequence of these difficulties, the figures relating to mixed buildings are often accompanied by reservations as to their exactitude. In some, cases an attempt has even been made to determine what the consequent error may be 1 . On the other hand, if only the other criterion, that of actual use, is employed, the contrary disadvantages arise. Buildings intended for habitation, but temporarily vacant or used for other purposes, will be left out of the statistics. Such cases occur in particular when one-family houses are left empty, or when all the flats in a block are used for offices, shops, etc. It therefore seems advisable to adopt the same procedure as that decided on for the definition of dwellings, namely, t o accept both criteria of intended and actual use, making them supplement each other. Buildings with dwellings may thus be taken to be any buildings intended wholly or partly for habitation, or actually used for this purpose. It may be added that there are also certain types of habitations which hardly correspond to the idea of a building. This applies to temporary constructions (huts, tents, etc.), vehicles 1 Thus, in the census for the municipality of Basle of 1920, a brief calculation was made, which seems sufficiently interesting to summarise here. The method was as follows : to the number of dwelling houses enumerated in 1910 was added the number of those constructed in 1910-1920, while the number of those demolished during t h a t period was deducted. The result thus obtained should have agreed with t h e n u m b e r of houses enumerated in 1920. In actual fact a slight difference was recorded, more marked in the rural districts, as will appear from the figures below : N u m b e r of dwelling h o u s e s in 1920 Basle Town Kural districts . . . Total Difference Census figure Calculated figure Absolute figure 11,167 il,266 99 0.9 537 548 11 2.7 11,704 U,814 110 1.0 Percentage This calculation shows not the error made in one or other of the censuses, but the combined errors of both. Even so, the figures are quite significant. It should be added t h a t the check is only of value if the data concerning the construction and demolition of buildings are more reliable t h a n the census data, which seems admissible, given t h e average size of the town of Basle and the excellence of its municipal, services. — 45 — and boats (wagons, caravans, ships, barges, etc.). These types of construction are seldom included in statistics of buildings l , and even when they are they are usually enumerated separately. As they are few in number, however, except in certain oversea countries, their absence involves no serious omission. Classification of Buildings with Dwellings The importance of the building from the point of view of housing problems rests in the fact that it brings under one roof the various premises used for occupation by households, or for economic or social activities. The first object of a statistical analysis will, therefore, be to bring out the manner in which this grouping actually takes effect in the various buildings. For this purpose the degree of independence offered by the building to the different households occupying it will be examined, a distinction being made between family households and those not of a family nature. At the same time the relative importance in the building of the premises used for living in and those used for other purposes will be taken into consideration. These questions may and are approached by means of fairly different systems of classification 2, which yield however much 1 Reference may, however, be made to the German census of 1910, t h e Dutch of 1907, the Australian of 1911 and 1921, and the American censuses of 1910 and 1920, in which habitations of this kind were also enumerated. 2 Thus, in the Austrian census of 1910, a distinction was made between t h e buildings intended exclusively for habitation, those intended principally for habitation and subsidiarily for other purposes, and those intended subsidiarily for habitation and principally for other purposes. In the last two classes a further subdivision into seven categories was made according to t h e intended use of the premises not reserved for habitation : agriculture and forestry ; industry, commerce, or a profession ; administration and similar activity ; and various combinations of these three principal groups. In the English census of 1921, a separation was first made between buildings containing dwellings from those containing none, and the former group was then subdivided into five : undivided private houses ; structurally divided private houses ; blocks of flats, tenements, etc. ; shops ; others, under which head separate mention was made of (1) offices, factories, warehouses, workshops, (2) hotels, (3) institutions, (4) places of worship, and (5) places of amusement. In the Berlin census of 1910, a distinction was first made between inhabited and uninhabited buildings, and then each of these two large groups was subdivided into forty-four classes, the first of which consisted of buildings intended solely or principally for family habitation, the following seven of buildings intended for various forms of habitation b y other t h a n family groups, and the other thirty-six of all the various forms of buildings intended for the economic and social activities of the population. — 46 — the same results. From the international point of view, therefore, a very strict classification does not seem called for, and it will be sufficient to determine which data are the most important to obtain. It must be possible to distinguish first of all between the buildings intended solely for habitation, those of which only a part, although a relatively important part, is used for this purpose, and finally those in which only a small part has been equipped, with a view to housing a caretaker for instance. The separation of this third group seems useful enough, although it is not always made, for in this latter case the characteristics of the building are much less important from the point of view of determining those of the dwelling itself than when the building is mainly used for habitation. For mixed buildings, i.e. those of which only part is used for living in, whether this is the principal part or not, it is also interesting to determine for what purpose the part not lived in is intended. The classification to be used must obviously be based on that used for buildings entirely devoted to the social and economic activities of the population. Further, among buildings used entirely or principally for habitation, it is necessary to distinguish whether they are intended for family households or other than family groups, for the analysis to which they are subjected will differ in the two cases. Buildings for other than family habitation may be classified according to the nature of the community for which they are intended, and therefore the same large groups as those already suggested for the classification of non-family dwellings should be used 1. Buildings intended for family occupation should be classified first according to the number of dwellings they contain. In many cases, it is true, it is considered sufficient to distinguish between houses for one or two families and those with several flats. It would be worth while, however, making a more detailed classification, especially as this presents no difficulties, the building census usually being connected with that of dwellings. Moreover, as dwellings differ largely in size, there would be every advantage in supplementing this classification by combining it with another based on the number of rooms in the buildings. In this way it would be possible to estimate exactly the capacity of the building as well as the 1 See Chapter II, p. 39. — 47 — number of dwellings into which it is divided. This method should be developed more than it actually is 1, because of the interest attaching to it. Most building censuses include a statistical analysis of certain structural characteristics, though as yet it is far from being conducted on more or less uniform lines. The question of the materials of which the buildings are constructed has been dealt with in certain countries, though in fairly different ways. In some cases account is taken of all building materials, and a distinction is made between buildings of wood, stone, brick or several materials 2. In other cases the only feature taken into account is the material of the outside walls 3, or of the roof 4. Data concerning the number of stories in the building are more usual and also more comparable, but certain borderline cases may be interpreted in different ways. At what point is a story to be regarded as the basement rather than the ground floor ? What is the exact limit separating off the attics ? These points must be defined very carefully if the figures published are to be properly interpreted. Some countries supply certain other information on the structural arrangements of buildings. Thus they may be divided into the principal block and outhouses 5 , or the principal building may be distinguished from the adjacent secondary 1 Data concerning the number of rooms in buildings are to be found, for instance, in the Austrian census of 1910, the Bulgarian census of 1920, and the Stockholm censuses of 1910-1920. 2 Thus, in the Norwegian census of 1920, a distinction was made between wooden buildings and buildings of mixed materials ; in the Warsaw census of 1919, between wooden buildings and those of stone and brick ; in the Christiania (Oslo) census of 1920, three classes of buildings were defined, those of wood, brick and other materials (together with one heading " particulars not stated " ) . 3 This classification is to be found in particular in the Australian censuses of 1911 and 1921, and those of the Union of South Africa of 1911 and 1921, in which a distinction was made between fourteen and ten different classes respectively ; stone, brick, concrete, wood, iron, sun-dried brick, and other materials commonly used in these parts were considered separately and in various combinations. It should be added t h a t these classifications were applied to the dwellings themselves. As, however, in these countries the dwelling and the building tend very much to coincide, it seems reasonable to quote these examples here. 4 This classification is to be found, for instance, in the Austrian census of 1910, which distinguished between seven classes, with reference also to the existence or absence of a fire insurance policy. 5 The Norwegian censuses of 1910 and 1920. — 48 — buildings \ or else the position of the building may be defined with respect to the street (buildings on the street, at the side, or at the back) 2 . In other cases information is given on questions of internal arrangement — means of lighting (gas, electricity) and heating (central heating), provision of a lift, drinking-water, drains, closets ; neighbourhood of a garden or court ; the existence of dwellings in the cellar or attics 3. In actual fact these various data do not seem of the first importance from the point of view of housing policy. Several of the structural features in question have only a very indirect influence on the actual dwellings, and their study belongs rather to the field of town planning. Moreover, an analysis of the internal arrangements of buildings seems more expedient in connection with the statistics of dwellings, for these details are of value particularly in so far as they illustrate the conditions of comfort of the dwellings. Besides, several of these classifications are suitable only in certain countries, being the outcome of specifically local conditions. It therefore seems unnecessary to propose standards with respect to them for international comparison. The question of excluding them entirely from census forms might even be examined if, as has already been proposed in connection with the statistics of dwellings, it were decided to make supplementary enquiries by the method of sampling for dealing with these points of detail. It may be mentioned finally that in some cases buildings are analysed also from a legal standpoint. They are classified according as they are in individual ownership and non-individual ownership and according to the manner in which the property was acquired (purchase, inheritance, construction, etc.) *. 1 Austrian census of 1920. The principal censuses of German municipalities. 3 The German census of 1911 ; the Austrian census of 1910 ; the Swedish censuses of 1910.and 1920 ; and that of Warsaw of 1919. 4 The Austrian census of 1910 is one of the best-known examples of this type of analysis. The classification according to method of acquisition contains the following classes : construction, purchase, inheritance, •other methods, methods unknown ; a distinction also being made between houses built before 1901 and those built after that date, so as to bring out the changes that have taken place in the form of ownership during the last ten years. In the classification according to legal personality of the owner, a distinction is first made between those who are individuals and those who are not (e.g. societies, companies, institutions, etc.). In the first 2 — 49 — STATISTICS OF REGISTERED PROPERTIES COMPRISING BUILDINGS WITH DWELLINGS In some countries statistics of registered properties are to be found side by side with the statistics of buildings, being in some measure connected with the latter. This applies, for instance, to the censuses taken in various German, Swiss, Swedish, Norwegian, and Polish towns. The fact is that in these countries the property registers offer the best means of reaching all owners of real estate, and it is therefore obvious that they should be used for building censuses. A form is therefore provided for each registered property and not for each building, and the enumeration of the buildings on them is obtained by examining these returns, thus making it possible to compile statistics of buildings properly so called. In certain cases, however, the registered properties themselves are analysed, being considered to form the basic units. As a rule, such statistics are limited to sites already built on, but sometimes they also cover those which have not yet been built on. There is no need to consider in detail the definition of the basic unit, the registered property, as this depends directly on the legal definition adopted in the country in question. Among the different registered properties, however, it would be necessary to distinguish those comprising buildings with dwellings, the only ones to be taken into account here. This distinction will be made as a matter of course if the criterion taken is the existence on the site of a building with dwellings, and if the latter is defined as suggested. In some cases, however, only the principle of actual use is adopted, the properties being classified according as they are actually inhabited or not. Any consequent differences in the nature of the data are of comparatively little importance from the point of view of housing problems however, considering the type of information usually expected from these statistics. group those who are sole owners are separated from those who are coproprietors, the cases in which the property is shared between husband and wife being distinguished separately in the latter group. Owners are also classified according to occupation, a distinction being made between four large groups (agriculture and forestry ; industry, commerce and transport : public administration, liberal professions ; no occupation), each of them being separated up into independent and paid workers. The group of owners who are not individual persons is divided into seven categories : companies run for profit ; religious bodies ; foundations ; municipalities ; provinces ; the State ; miscellaneous. — 50 — The point is that the registered property is a legal unit, and that consequently statistics of these properties are useful mainly for studying the legal aspects of housing problems. They therefore serve principally to study the forms of ownership of, and methods of acquiring, the sites on which buildings with dwellings are placed, and therefore indirectly of the dwellings themselves. These problems, it is true, are also approached in the statistics of buildings, as already indicated. The systems of classification used are therefore similar to those already discussed, although often conceived on somewhat different lines 1In the statistics of built-on registered properties, a classification of the number of buildings and even of the dwellings on the site is fairly general. Some enquiries even go so far as to analyse certain structural characteristics of the building, but this study seems more appropriate to that of the statistics of buildings. 1 Thus, the Berlin census oí 1910 subdivided the methods of acquisition into eight classes : purchase, exchange, sale by auction, endowment, inheritance, combinations of different methods, miscellaneous, method unknown. As regards classification according to the nature of the proprietor, those who were physical persons were divided into two groups — persons of whom at least one lived in the building, and those who did not live there ; the latter were again subdivided according to their place of residence, whether • inside or outside the town. The analysis also covered the occupation of the owners, no fewer than twentythree industrial classes being distinguished. It was combined, moreover, with information on the period of their ownership rights. CHAPTER IV STATISTICS OF HOUSING CONDITIONS The statistics of dwellings and of buildings with dwellings examined in the two preceding chapters are based solely on technical units. The object was to bring out with as much detail as possible the different characteristics of dwellings and buildings, without making any attempt to determine the manner in which they are used by their occupants and the consequent housing conditions of the latter. The latter problems belong to the field of statistics of housing conditions. For this purpose use is made of the data obtained from the statistics of dwellings and to a less extent of those obtained from the statistics of buildings, while at the same time new factors are introduced concerning the persons and households occupying such dwellings or buildings. Systems of classification are. prepared according to the particular features of these new factors, and the relations between them all are established. The statistics of housing conditions are therefore technically connected with those of dwellings and buildings, and as a rule they are compiled at the same time. They are to be found in most countries where the first two sets of statistics have alreads been stated to exist. Another source of data on housing lies in the special enquiries undertaken in several countries for studying the conditions of life of the workers, in which housing conditions are the subject of special investigation. Examples of such enquiries are to be — 52 — found particularly in Austria \ the United States 2, France 3, Italy *, and Sweden 5. Most of them date from the last few years of the nineteenth century or from the first ten of the twentieth, a period during which the study of working-class conditions in this particular form was much to the fore. Comparatively recent data are to be found for Far-Eastern countries, particularly Japan 6 , where they still form one of the only sources of information on housing questions. The results obtained from these investigations are usually limited to certain quarters of the town or a particular group of workers. Moreover, they are not based on all the dwellings affected but only on a certain number of families chosen as typical. The extent to which the study of housing conditions in general censuses has developed varies somewhat. As a rule the statistics are limited to family dwellings, and it is only when the statistics of dwellings make no distinction between 1 Die Wohnungs- und Gesundheitsverhältnisse der Heimarbeiter in der Kleider- und Wäschekonfektion, K. K. ARBEITSSTATISTISCHES A M T IM HANDELSMINISTERIUM, Vienna, 1901. Die Wohnungs- und Gesundheitsverhältnisse der Schuhmacher, K. K. ARBEITSSTATISTISCHES A M T IM HANDELSMINISTERIUM, Vienna, 1906. Die Arbeitszeit in Eisenhütten und Walzwerken, Bericht über die in der Zeit v o m 14. J u n i bis 14. August 1909 durchgeführte Erhebung, K. K. ARBEITSSTATISTISCHES AMT IM HANDELSMINISTERIUM, Vienna, 1911. Die Lage der Wächter der K. K. Staatsbahnen, auf Grund einer Erhebung des K. K. Eisen- bahnministeriums, K. K. ARBEITSSTATISTISCHES A M T IM H A N D E L S - MINISTERIUM, Vienna, 1903. Die gemeinnützigen Kleinwohnungsanlagen in den im Reichsrat vertretenen Königreichen und Ländern, nach den Ergebnissen der Erhebung vom 17. April 1909, auf Veranlassung und m i t Unterstützung des K. K. Ministeriums für öffentliche Arbeiten bearbeitet u n d zusammengestellt von dem BUREAU DER K. K. STATISTISCHEN ZENTRALKOMMISSION, Brunn, 1910. Statistik der hauszinssteuerpflichtigen Wohnungen nach dem Stande vom Jahre 1908, Beilage I zu Heft I der Mitteilungen des K. K. Finanzministeriums, Vol. X V I , 1910, Vienna, 1909. 2 The Slums of Baltimore, Chicago, New York, and Philadelphia, 1894 ; Seventh Special Report of t h e Commissioner of Labour. Report on Conditions of Women and Child Wage-Earners in the United States ; Washington, 1911. 3 O F F I C E DU TRAVAIL : Enquête sur le travail à domicile dans l'industrie de la lingerie, Vols. I-V. Paris, 1907. 4 UFFICIO D E L LAVORO : Inchiesta sulle abitazioni degli impiegati d'ordine e subalterne in Roma e del personale ferroviaro in Roma e in altre città d'Italia. Series B , No. 20. Rome, 1908. 5 SVERIGES OFFICIELI A STATISTIK : Undersókningar ròrande de minare bemedlades bostadsfórhallanden i vissa svenska städer, 1912. Stockholm, 1916-1918. 6 KYOCHO K A I (OF TOKYO) : Investigation on the Conditions of Living of Salaried and Wage-Earning Workers in November 1923. Jiji Yearbook, 1925, " Housing Conditions of t h e Workers of Yokosuka Arsenal " . Rodo Tokei, Vol. I. — .53 — these and the dwellings of other than family groups that the analysis of housing conditions also covers the latter. DISTRIBUTION OF OCCUPANTS IN THE DIFFERENT TYPES OF DWELLINGS AND BUILDINGS The simplest operation in statistics of housing conditions is to enumerate the occupants covered by the census and to classify them according to the characteristics of the dwellings or buildings they occupy. This involves, however, the adoption of a new basic unit — the occupant — - which must be defined. The definition may be based on one of two different conceptions. It may either keep to the actual situation on the day of the census, and regard as an occupant of a dwelling any person spending the night there on that date ; or else it may consider the normal situation, and count as occupants only those who ordinarily live in the dwelling. From the theoretical standpoint, this latter conception has certain advantages, for it leads to a more representative picture of the actual facts ; but from the practical point of view it is more complicated, for it means that when recording the occupants of a dwelling, any person absent on the date of the census but ordinarily living there should be included, while those temporarily there have to be excluded 1. The other method, which may have the advantage of simplicity, does not give as satisfactory results. It should be stated, however, that the differences due to the fact that these various persons are recorded away from their ordinary homes are hardly noticed in the final totals, for the errors tend to cancel each other out. The classification of occupants according to the characteristics of the dwellings they occupy makes it possible to determine which section of the population enjoys the advantages of each type of dwelling. It forms a direct complement to the data of the statistics of dwellings, which bring out the proportion of dwellings of different types in the total. The same criteria for 1 Thus the form for the German census of 1927 stipulated that persons would have to be counted as occupants if " they were temporarily absent, but not those temporarily present ". The household form used for the French census states that mention must be made of the number of persons composing the household, " those present or absent, but not those temporarily there ". — 54 — classification will therefore be used. The chief of these is the size of the dwelling measured by the number of living-rooms, including the kitchen and, as the case may be, by the number of certain subsidiary rooms. Various other characteristics of the dwellings may also be taken into account : use of subsidiary rooms (bath-rooms, closets, laundry accommodation, etc.) and domestic equipment (means of heating and lighting, installation of running water, drains), proximity of a court or garden. It seems, however, that for studying these various details the method of sampling may with advantage be used rather than that of the census, as already suggested for the examination of these characteristics in the statistics of dwellings. Similarly, the occupants may be classified according to the characteristics of the buildings they occupy, but this operation is rarely undertaken. In. some cases, however, the occupants are classified according to the size of the building, in others a distinction is made between the occupants of houses for one or two families and those of houses with several flats. D E N S I T Y OF OCCUPATION Any further classification used in statistics of housing conditions will be based on the characteristics of the occupants themselves and the groups they form. That most widely adopted, because it is at once simple and significant, is based on the number of persons living in one and the same dwelling. In fact, the part it plays in the statistics of housing conditions is similar to that of the classification based on the number of rooms in the statistics of dwellings. In other words, any further statistical analysis is carried out within this fundamental classification. The significance of the number of persons occupying a dwelling is not fully realised, until it is related to the size of the dwelling occupied, in other words, the number of rooms. For this purpose, a twofold classification is used, based on both these criteria at once, which may be applied either to the persons or to the dwellings. Two sets of double entry tables are accordingly prepared, one giving the number of families of 1,2,3, etc., persons living in dwellings of 1, 2, 3, etc., rooms, and the other — 55 — the number of dwellings of 1, 2, 3, etc., rooms occupied by 1, 2, 3, etc., persons. The composition of these different classes varies somewhat from one country to another. In some cases a very detailed classification of all dwellings is given ; in others, it is limited to small dwellings only (not over four or five rooms, and with not more than five or six occupants), and all others are combined in an additional class. In yet other cases, finally, only a few classes are defined more or less arbitrarily, a distinction for instance being made between small, average size, and large dwellings. This rough grouping has no doubt the advantage of bringing out general tendencies more clearly, but it has also the disadvantage of concealing certain facts, and unless it is carried out on uniform lines it makes international comparison very difficult. From this point of view, therefore, it will be preferable to maintain altogether detailed classifications. With a view to bringing out more clearly the density of occupation of the dwellings, it is often customary to calculate also the ratio of the number of rooms to the number of occupants, thus obtaining a coefficient of density. The most usual is that of the number of persons to a room. In some cases, however ] , the inverse ratio has been preferred, i.e. the number of rooms per occupant. This coefficient, in whichever way it is calculated, serves for a double analysis. In the first place, the dwellings or their occupants may be classified according to their coefficient of density of occupation, and, secondly, the average coefficients of density are calculated for dwellings of different sizes or occupied by different numbers of persons. Several countries have even gone further and introduced a standard with a view to determining how many dwellings may be considered to be overcrowded. For this purpose, Bertillon's criterion has usually been adopted, according to which any dwelling is considered overcrowded if it has more than two persons per room. In some countries rather stricter or wider criteria are preferred 2 ; in others, finally, a complete scale of 1 Census of England and Wales of 1921 ; French censuses of 19011911. 2 In Finland, for instance, for the census of 1919, only dwellings with three or more persons per room were considered to be overcrowded, and in Prague, for the 1921 census only those with over two persons per room which could be heated. — 56 — density of occupation has been compiled, the various classes being defined in a more or less arbitrary manner 1. But these differences are not the most important from the point of view of international comparison. Those arising out of the definition of the terms " dwelling ", " room " and " occupant " — and it has already been shown how many there are — are much more serious. For it is clear that the significance of the density of occupation depends fundamentally on the meaning given to the units compared. Without returning to what has already been said on the subject of the standardisation of these various units 2, it is necessary, however, to stress the importance here of the distinction already made between living-rooms, kitchens, and subsidiary rooms, for in most countries only the principal rooms are taken into account in the study of the density of occupation. For the present purpose, it is important also to include servants' rooms and habitable attics, for the actual density of occupation of a dwelling would certainly be exaggerated if the person or persons acting as domestic servants were to be counted among the occupants, but the room or rooms they occupied were to be excluded. The question may even arise whether in the calculation of density of occupation, subsidiary rooms should not also be counted, such as bath-rooms, alcoves, or lobbies, as has been done in some countries 3, for the density of occupation 1 In the French censuses of 1901-1911, dwellings with one or two rooms per person were considered spacious and those with over two persons per room overcrowded. In the Copenhagen and Frederiksborg census of 1921, a distinction was made between overcrowded dwellings, or those with over two persons per room, and very overcrowded dwellings, with more than four persons per room. In Norway, the 1920 census fixed five classes : very spacious dwellings, more than two rooms per occupant ; spacious dwellings, less than one person per room ; crowded dwellings, more than one and less than two persons per room ; overcrowded dwellings, more than two and less than three persons per room ; very overcrowded dwellings, more than three persons per room. The "Warsaw census of 1919 also distinguished between five classes, but the graduation was somewhat different : very large dwellings, more than two rooms per occupant ; fairly large dwellings, more than one room per occupant ; normal dwellings, one room per occupant ; inadequate dwellings, more than half a room and less than one room per occupant ; overcrowded dwellings, less than half a room per occupant. 2 For the definition of a dwelling, see Chapter II, p. 31 ; for that of a room, Chapter 11, pp. 34-35; and forthat of an occupant, Chapter IV, p. 53. 3 This method was adopted, for instance, in the Austrian census of 1910 and the Zurich census of the same year. — 57 — may be decidedly different if various subsidiary rooms are available in the dwelling in addition to the living rooms. Perhaps these subsidiary rooms should be counted at a reduced rate, for instance, as equivalent to half a unit, as has already been suggested for the study of the size of the dwelling. With respect to occupants, no distinction is made as a rule between different persons, who are all considered as units of the same value. It is clear, however, that from the point of view of density of occupation, the occupants must differ in importance according to age, sex, and the relations between them. Thus, the degree of overcrowding of a dwelling of two rooms occupied by four persons is much less if these persons are a married couple and two children of the same sex or of tender years, than if the children living with the parents are an adult son and daughter. For a detailed study of the degree of overcrowding, these differences ought to be taken into account, which would obviously involve establishing a scale similar to that used for the consumption of individuals in the statistics of family budgets, although it would be based on different criteria. Even though this would be a difficult operation, it ought to be examined. A first step of not too much difficulty might be to distinguish between three groups of persons : adults, children of between four and fourteen to fifteen years of age, and those under four years, counting the children as only a fraction of an adult person, one-half or one-third, for instance. Once an agreement has been reached on the definition of the terms " dwelling ", " room ", and " occupant ", the manner of establishing the different classes would have to be decided. It would probably be best to give up any idea of fixing a standard, and to provide a sufficiently detailed classification to allow of any form of analysis. Separate classes should therefore be allotted to the different numbers of rooms of dwellings, except perhaps for relatively high numbers, 8 rooms and 6 occupants, for instance, as here the number of cases is small, so that they may without inconvenience be grouped under a single heading. In studying density of occupation, it is not enough to obtain information on the degree of occupation of the different dwellings. Some account must also be taken of the extent to which the groups of persons living in the dwellings are independent of each other. This need not involve considering the size of — 58 — the building, i.e. the number of rooms it contains, as this would to some extent overlap with the analysis of the density of occupation of the dwellings. It is enough to determine the total number of occupants in the building so as to ascertain the number of persons living together in the same house. The buildings will then be classified according to the number of their occupants 1 , and the occupants by the number of occupants in the building. This operation presents no difficulties if the census of buildings is combined with that of dwellings. In some countries the average number of occupants per building is also determined 2. These figures, it should be observed, may be obtained without counting the number of occupants in each type of building, but simply by dividing the total number of occupants in the locality by the number of buildings with dwellings. If this figure of average density is to be truly significant, it is preferable, however, to calculate it separately for the different types of dwelling-houses (houses used exclusively, principally or subsidiarily for habitation on the one hand, one- and two-family houses and houses with several flats on the other). Otherwise the averages obtained may be altogether fictitious. To obtain this result it is obvious that the number of occupants per building have to be counted. But almost equivalent data may be obtained by subdividing the area into zones according to the principal type of building, and comparing the number of occupants and buildings in each zone. In certain cases, finally, the number of occupants per square metre of habitable area in the whole of the urban aggregation is also determined. These figures are obtained, not by measuring the premises on the spot, but by determining the superficial area of the houses from the registered plans and multiplying this figure by the number of stories. Such data can therefore be obtained only where the registered plans are prepared with great accuracy, and, even so, they are always rather approximate. Data of the general density of occupation of the city, obtained by dividing the total population by the total area, are also of some interest. The significance of these figures varies some1 This classification is to be found, for instance, in the censuses for the municipalities of Prague in 1921 and Warsaw in 1919. 8 National censuses of Germany, 1911, and Sweden, 1920 ; various municipal censuses, particularly of Germany and Switzerland. — 59 — what, however, as they depend above all on the manner in which the urban aggregation is delimited and the extent to which the suburban or semi-rural surroundings are taken into account. This type of data belongs more to the subject of town planning than to housing statistics properly so called. LIVING TOGETHER OF SEVERAL HOUSEHOLDS AND COMPOSITION OF HOUSEHOLDS Although the relations between the different persons living together in the same dwelling have been given little consideration in the study of the density of occupation, they have not altogether escaped statistical analysis. Thus, in several countries they are the subject of special classifications applied either to the personal or to the dwelling units. In Austria, for instance, a distinction is made between households with and without children, with and without other relatives, with and without domestic servants, with and without sub-tenants and boarders. In Finland, the occupants are classified into five groups : members of the family properly so called, relatives, domestic servants, boarders, and co-tenants, and these groups are combined in various ways to bring out the general tendencies. This form of classification has perhaps been carried furthest in Berlin. The occupants of dwellings are subdivided into six classes : members of the family, relatives, employees, domestic servants, lodgers, and " night boarders ". The class of members of the family is then combined in all possible ways with the five other classes, thus giving no less than thirty-two different groups \ These different systems of classification bring out very different kinds of relations : family relationship, relation of master 1 The combinations are a r e distinguished : (1) Members of the family (2) „ „ „ (3) (4) (5) (6) (7) (8) as follows : First, the following eight cases alone. and employees. relatives. and relatives and employees. and domestic servants. domestic servants and relatives. domestic servants and employees. domestic servants, relatives and employees. Each of these eight groups is subdivided into the following four : (a) (b) (c) ,(d) without sub-tenants or boarders ; with sub-tenants, but without boarders ; without sub-tenants, but with boarders ; with sub-tenants and with boarders. — 60 — to servant, of host to boarder, or the simple living together of two independent households. From the point of view of housing policy, it seems that a definite distinction should be drawn between this latter group and the rest, for they reveal an exceptional, and in some measure abnormal, situation, as the dwelling is originally intended for a single household. It therefore appears preferable to consider the question of several households living together in one and the same dwelling separately from that of the composition of the households. At this stage of housing statistics, it is necessary to introduce a new unit : the household. This has already been defined as a group of persons with a common and independent domestic organisation. Now it is necessary to determine where this independence begins and ends. The demarcation is comparatively obvious if the households consist exclusively of members of the family, for then they are defined by this very family relationship. The task becomes much more difficult when there are boarders and lodgers, in view of the whole range of systems possible. Two fundamental types may, however, be distinguished. The first is that of the person who shares in all or part of the domestic life of the family, either he takes all or some of his meals with it, or he has only one room or part of a room, which, however modest, is looked after by the household with whom he lives. This group will include " night boarders " (Schlafleute) and " hirers of a bed " (Bettgeher), who will be considered to be members of the household. The second type is that of persons who have organised in the room or rooms they occupy a domestic life as independent as the arrangement of the premises will allow, and are therefore counted as an independent household. Often this group is taken to cover only sub-tenants and co-tenants, that is to say, persons who have secured the right to occupy part of the dwelling by a sub-tenant's lease concluded with the principal tenant or by a separate lease concluded with the owner, This distinction is no doubt of interest for the study of the legal forms of leases of dwellings, but it does not seem absolutely fundamental here. It would be more useful to distinguish so far as possible here between persons not connected with the family, but forming part of the household, and those who remain completely independent. The question of several households living together in the — 61 — same dwelling has fairly often been the subject of statistical research. One of the most complete was probably that undertaken on the occasion of the Norwegian census of 1920, which involved two distinct kinds of analysis. The first consisted in determining the housing conditions of the various households living together in one dwelling but considered them as independent units. For this purpose the number of living-rooms occupied by each household were examined, as also the available subsidiary rooms, the density of occupation, etc. The standard of comfort of this particular group of households, which is compelled to share their dwelling with other households, was thus brought out and could be compared with that of the population as a whole. The second analysis consisted in comparing the housing conditions of the households living in the same dwelling, a distinction being made between the principal and the subsidiary households. In this way it was found whether one of the two households living together was worse housed than the other. As regards the composition of households, it seems that from the point of view of housing problems it would be suitable to distinguish between five classes, which is in fact the fairly general practice. These are : members of the family, other relatives and friends, domestic servants, employees, boarders. Each of these chief groups might be further subdivided so far as considered necessary for the study of demographical problems connected with the composition of households ; but this question does not arise here. HOUSING CONDITIONS OF DIFFERENT SOCIAL GROUPS So far only the question of analysing the housing conditions of the population as a whole has been considered, and the systems of classification examined have aimed only at determining the different standards of housing and the number of persons to whom these standards apply. It is also of value to relate housing conditions to the different social classes of which the population is composed, so as to determine the standards prevailing in each class. It is all the more important to bring out this relation as housing is one of the fields in which the influence of social class is most strongly felt. — 62 — The difficulty of this operation lies in the delimitation of the different social classes. It is based as a rule on the occupation of the head of the household, or the industry in which he works, and his occupational status. But the different systems of classification adopted vary markedly from one country to another. The Netherlands census of 1911 established twenty-nine industrial classes, distinguishing for each industry between employers, salaried employees and wage-earners. The same method was adopted for the Prague census of 1921, although the number of industrial groups was only twenty. In Switzerland an occupational classification has been generally adopted. The Zurich enquiry of 1910 comprised about forty classes, which were then combined into three large groups, corresponding approximately to the well-to-do, the middle class, and the workers K The census for the municipality of Berne has a rather more detailed distribution by social classes 2 . In Italy for the enquiry undertaken in 1921 by the Union of Municipal Statistical Offices, provision was made for a classification into eleven social classes, based on the declarations of the inhabitants as to their occupation and their occupational status 3 . In Norway the social classification in the 1920 census was also 1 More exactly, the first class comprised independent workers in different occupations, the owners of factories and their partners, directors, high officials, members of the liberal professions, and persons without occupation ; the second class, employees with technical or commercial training, the staff of administrative departments and the official transport services ; the third, wage-earners, shop assistants, hotel and restaurant employees, domestic servants in public or private service, and unskilled labourers. (" Die Wohnungen in der Stadt Zürich am 1. Dezember 1910 ", Statistik der Stadt, Zürich, No. 21, p. 14.) 2 I. Dependent workers : wage-earners, hotel and restaurant employees, staff of banks and insurance institutions, various private employees, officials. II. Independent workers : employers (large undertakings), hotel and restaurant keepers, lawyers, doctors, etc., contractors and architects, artists and journalists, small traders, miscellaneous occupations, pension holders, no occupation, occupation not stated. (" Die Wohnverhältnisse in der Stadt Bern " —• Beiträge zur Statistik der Stadt Bern, No. 6, p. 42.) 3 The social classes were as follows : persons living on their income ; liberal professions ; directors and managers of industrial, commercial and banking undertakings ; independent small artisans ; small traders ; officials and salaried employees, including pension holders ; industrial wáge-earners ; subordinate staff in public services, the armed forces, domestic service, retail trade, hotels and restaurants, etc. ; agricultural workers ; occupation not stated. (Bollettino dell' Unione Statistica delle Città italiane, Eighth Year, No. 2, 4 Dec. 1921, p. 66.) — 63 — fairly detailed K In Sweden, on the contrary, there are only five large groups 2. The differences, it will be seen, are considerable. It should also be added that in these various systems of classification the relative importance of the different classes is very unequal. Thus, in most cases the working class forms only one group, although the workers everywhere represent a very large proportion of the population covered by the censuses. Other groups, on the contrary, are distinguished in the same way as the working class, although their number is much smaller and they have no very distinctive characteristics. An entirely different method of classifying the population into social classes is that based on income 3 . Such a system has the disadvantage, however, of not allowing of a distinction being made between certain social classes whose income is much the same but whose standard of living differs. The problem of standardising the system of classification into social classes is a very delicate one. The very idea of social class is vague and may be interpreted differently according to country. It seems, however, that if a threefold criterion is taken, the industry to which the person belongs, his status in the industry, and the amount of his earnings or income, a fairly reliable social classification might be obtained. A distinction might for instance first be made in the working class between certain large groups, each of which would comprise a relatively large number of persons and would correspond approximately to a different standard of life from the rest, such as skilled workers with possibly a separate heading for certain industries like the textile and mining industries, unskilled workers, homeworkers, and artisans. For other social groups a middle class might be defined as consisting of traders, manufacturers, members of the liberal professions, officials, and 1 It was as follows : manufacturers and wholesale traders ; artisans ; retail traders ; higher officials, liberal professions ; clerks and commercial employees ; foremen and technical staff ; seamen and fishermen ; skilled workers ; other workers ; pension holders ; occupations not stated. (" Folketellingen i Norge i desember 1920, VII, Boligstatistik", in Norges offisielle Statistik, VII, 98, p. 135.) 2 As follows : workers ; officials ; small artisans ; persons retired or pensioned and persons living on their income ; and well-to-do persons, 3 This method was adopted for instance in the Swedish housing censuses of 1910 and 1920, where this information was obtained from the income-tax returns. — 64 — responsible employees in private undertakings, and a well-to-do class as consisting of persons living on a comfortable income and highly-paid persons in all occupations. Fairly satisfactory, although more rudimentary, results may also be obtained for certain urban aggregations by an altogether different and indirect method. It consists in compiling all the data by relatively small districts or geographical units, and in grouping these units according to social character into workers' districts, middle-class districts, and upper-class districts. This method has the advantage of simplicity, but it can only apply when the social character of each quarter is well defined. Even in such conditions the decision as to the class into which a particular district is to be placed will always be somewhat arbitrary. The relation between housing conditions and social class is not the only one that might be examined. For a detailed study of housing problems, it would for instance be very instructive to bring out by a suitable system of classification the relations between housing conditions and nationality, race, and religion. Very little has been done so far to explore these fields, apart perhaps from the question of race, which has been examined in certain oversea countries where a distinction between the dwellings of white and coloured persons is found necessary. CHAPTER V STATISTICS OF RENTS Among the various sets of statistics dealing with rents, a distinction should be made between the censuses or enquiries, whether periodical or not, from which the absolute level of the rents of the different types of dwellings may be determined in more or less detail, and the periodical investigations intended to determine fluctuations in rents without reference to their absolute level and based either on enquiries limited to a certain number of typical dwellings, or simply on the estimates of certain persons particularly acquainted with fluctuations in rents, or finally on the deductions that may be drawn from the legislation in force. This second type of investigation is undertaken as a rule with a view to establishing index numbers of the cost of living. The treatment of these questions is therefore dealt with in that connection and need not be discussed here 1 . The question of rents has, moreover, been the subject of thorough investigation in certain countries, for the purpose of determining the actual cost of living in certain social classes. There has even been an international enquiry, undertaken by the British Board of Trade, into the cost of living of the working class in Great Britain, Germany, France, Belgium, and the United States from 1905 to 1909 2, in which an attempt was made, among other things, to establish comparable data on rents in these countries. These studies were undertaken, 1 Cf. INTERNATIONAL LABOUR O F F I C E : Methods of Compiling Cost- of-Living Index Numbers. Studies and Reports, Series N , No. 6. Geneva, 1925. 2 Report of an Enquiry by the Board of Trade into Working-Class Rents, Housing, and Retail Prices, Together with the Rates of Wages in Certain Occupations in the Principal Industrial Towns of the United Kingdom, ... in Germany, ... in France, ... in Belgium, ... in the United States. London, 1908-1911. — 66 — however, from a special point of view and need not be considered further here 1. Reference may also be made for completeness to certain rent statistics prepared, more or less as an exception, on the basis of the returns made in certain countries in connection with rent taxation 2. The statistics considered here as a rule take one of two forms. The first is that of a general rent census undertaken at comparatively distant intervals and intended, above all, to give as detailed a view as possible of the conditions prevailing at a given moment. These statistics are usually compiled on the occasion of housing censuses, from which some of the data required are derived. They are considered, however, to form a distinct section of the enquiry and are not compiled in all countries. Reference may be made to the national censuses of Austria in 1910, Germany in 1918, Norway and Sweden in 1910 and 1920, and Switzerland in 1920, as also to the municipal censuses of several German, Scandinavian, and Swiss towns before and after the war, and those of Prague in 1921 and Warsaw in 1919. In some cases special enquiries were undertaken on the subject, as in Sweden and Switzerland in 1924. The second form of census is that relating as a rule only to urban aggregations and covering only the principal characteristics of dwellings, but undertaken regularly at fairly close intervals (half a year or a year) for the purpose of following changes in the general level of rents. This system is used particularly in German, Scandinavian, and Swiss towns. It should be added that in some cases these data are also used for the construction of index numbers which are incorporated in the cost-of-living index numbers. Rut however this may be, they are first compiled in the absolute form. It is clear that the purpose of these two types of statistics is not exactly the same. The first aims above all at studying the existing situation in all its details and at a given date, and if need be at determining the most important changes that have happened by a comparison with similar enquiries conducted 1 Cf. INTERNATIONAL LABOUR O F F I C E : Methods of Conducting Family Budget Enquiries. Studies and Reports, Series N , No. 9. Geneva, 1926. 2 One of t h e best-known examples of this type of statistics is t h a t published before t h e war by t h e AUSTRIAN Statistik der hauszinssteuerpflichtigen 1908 ; K. K. Finanzministerium ; MINISTRY OF F I N A N C E : Wohnungen nach dem Stande für Vienna, 1909. — 67 — some five or ten years before. The second aims principally at following as closely as possible the constant changes in rents, while limiting itself to a more approximate examination of the situation at the time of each return. Apart from these differences in the scope, thoroughness, and frequency of the investigation, the problems which arise are very much the same. DEFINITION OF A RENTED DWELLING AND OF R E N T The unit of " the dwelling " on which rent statistics are based differs somewhat from that described for the censuses of dwellings and housing conditions, in the sense that it is delimited not on the basis of the technical and social criteria of intended and actual use of the premises but on a legal criterion, the lease relating to the premises. Reference was already made to this conception of the dwelling when dwelling statistics were examined \ and it was objected that it was too narrow as it excluded all dwellings not rented (i.e. vacant dwellings, those occupied by their owner, or ceded free of charge or in return for services). This objection obviously has no force here, for it is clear that rent statistics can apply only to dwellings actually rented. It may be added that in certain cases rent statistics are taken to include vacant dwellings which are to let. In this case the rents demanded by the owner are recorded, but naturally such data have not the same value as those relating to dwellings actually leased. In any case they should be placed under a seperate heading. In this connection it may be observed that in certain towns special statistics are compiled concerning the rents of vacant dwellings registered with the public housing offices 2. These figures can hardly be considered representative of the general level of rents, unless the percentage of vacant dwellings is relatively high and the conditions of the housing market have been very stable for several years. They serve much more to illustrate the prices prevailing in the housing market, and even so it is necessary to ascertain that they do not comprise a large number of dwellings with excessive rents, which remain vacant for that very reason. 1 a See Chapter II, p. 27. The statistics of the municipalities of Basle and Hamburg. — 68 — In other cases *• rent statistics are taken to include dwellings occupied by their owner and those let free of charge or in return for services. The estimated rental value of the premises is then taken into consideration, being based either on previous leases or on analogous premises. These data are usually presented separately, not only because they are more approximate, but also to furnish interesting comparisons. Actually, however, inclusion of these groups modifies the statistics, which then become statistics of rental value and not of rents. Although the legal conception of the dwelling as defined by the lease relating to it has the advantage of clearness, it suffers on the contrary from the disadvantage of not corresponding exactly with the conception of the dwelling advocated here for other forms of housing statistics. Admittedly, the differences are not very considerable. They arise only if one and the same tenant acquires the right under one lease to two or more separate dwellings, or if two or more independent households each directly rent from the owner part of one and the same dwelling. These two possibilities, and particularly the latter, are seldom realised, for in most cases where households live together in the same dwelling, one of them acts as principal tenant and sub-lets to the others certain parts of the dwelling. If in spite of this it is desired to base rent statistics on a definition of the dwelling identical with that adopted for the other branches of housing statistics — which in theory at least would be preferable — the owner would have to be asked to determine the exact rent applying to each of the dwellings, either by adding the rents charged to independent co-tenants, or by dividing among the different dwellings the total rent he receives for several dwellings under a single lease. Although not impossible, such an operation seems somewhat difficult. Whatever system is adopted, it is also necessary to separate altogether dwellings let furnished and isolated rooms or parts of flats sub-let by the principal tenant under special contracts, the characteristics of these being so distinct that they must be studied separately. These questions are not so important as that of determining the significance of the word " rent ". In some districts it is not the custom to include in the rent more than the strict cost 1 Stockholm censuses of 1910-1920 and those of Berlin and Basle of 1910. — 69 — of hiring the premises, the tenant being entirely responsible for the cost of heating, lighting, and the water supply. In other districts, however, the owner takes the responsibility for the latter charges and therefore makes certain additions to the rent, but sometimes also adds a sum for certain services such as the clearing away of rubbish, services of caretaker (concierge), the maintenance of parts of the building shared by all the occupants, or for reimbursing his payment of certain taxes ; whereas in other localities many of these services are included in the rent, it being impossible to distinguish what part is actually paid for the actual hiring of the premises. These differences of custom must obviously lead to fairly marked divergencies in the significance of the rent statistics. If greater uniformity is to be obtained on this point, an understanding should be reached on the standard to be followed so far as possible. For instance, everything not directly connected with the hiring of the premises might be excluded, i.e. all charges for heating, lighting, water, and the removal of rubbish ; while, on the contrary, those connected merely with the caretaking of the building and its maintenance would be included, as the payment of the rent should give the tenant the right to use the entrances and parts of the building open to the public which the owner normally manages and maintains. As regards taxes, only those should be included which under the law are expressly payable by the tenants. If it is impossible to keep to this standard, differences should be clearly stated so that they may at least be taken into account in interpreting the data. CLASSIFICATION OF DWELLINGS BY RENTS AND DETERMINATION OF AVERAGE RENTS Rent statistics usually supply two sets of fundamental data : the number of dwellings classified by rent groups, and the average rents of different types of dwellings. The method of classifying dwellings into rent groups differs somewhat according to country. In some countries there are less than ten groups in the classification, in others over twenty ; but no international standard can be imposed here, for in the first place the currency units in terms of which the classes are defined differ fundamentally from one country to another, and — 70 — secondly the field to be covered varies with the degree of dispersion in rents as a whole. All that can be recommended is that the subdivision should be detailed enough to bring out the nature of the distribution, and that the subdivisions should not be too many as to mask the general tendencies. The computation of the average rents of different types of dwellings is usually made by calculating the simple arithmetic mean. It should be observed, however, that if it is proposed to compare the general level of rents of different localities, it is necessary when determining it to eliminate any cause of difference lying in the proportion of large to small dwellings in the district. For the average rent will clearly be higher in a locality where there is a big proportion of large dwellings, even if the rent is the same for dwellings of the same size. To eliminate this factor, it is necessary to calculate first the average rent of the dwellings in each size-group, and then to obtain a weighted mean of these average rents, choosing the same weights for all the localities. The choice of the standard weights must obviously be arbitrary. That most generally used is based on the relative importance of the number of dwellings of different sizes in all the localities covered by the census at a given date. For international comparisons, another choice of one or more standard weights would be necessary. A similar principle applies if it is proposed to compare one date with another. In this connection it may be added that if it is proposed to determine the rise in rents by means of a retrospective enquiry, the information given by tenants who occupied the same dwelling on the dates covered by the investigation must not be considered enough, for then there would be a risk of under-estimating the general rise that has actually taken place. Rents are increased usually on the occasion of the conclusion of a new lease, and if all persons who have moved are excluded from the investigation, a large proportion of those whose rents have been increased will be eliminated. For a retrospective enquiry, therefore, it is necessary to apply to the owners, who are alone in a position to give the rents paid for the different dwellings, irrespective of changes among their tenants. In some cases the average rent has been calculated not only for the dwelling, but also per room \ so as to bring out the 1 In the Swedish and Swiss censuses of 1920, for instance. — 71 — average prices paid for each room according to the different types of dwelling and the different localities. Similarly, the average rent per occupant is sometimes determined, with a view to bringing out differences in the conditions of large and small families. It is clear that such data cannot be of real interest from the international standpoint unless the terms " room " and " occupant " have been defined on uniform lines, a subject which has already been discussed 1. Mention should also be made of certain distinctions drawn in rent statistics, based on the fundamental characteristics of the dwellings. It has already been pointed out that the size of the dwelling must be taken into account in calculating the average rents for a number of localities. This fundamental criterion must obviously also be used in classifying dwellings by their rents. In point of fact, many periodical rent censuses go no further in analysing the characteristics of the dwellings. This form of study has been developed much further in special enquiries and the censuses undertaken on the occasion of general housing censuses. Systems of classification according to certain structural characteristics of the dwellings to which the rent figures relate are for instance used. Thus an attempt is made to determine the relation between rents and the story on which the dwelling is situated, with particular reference to those in attics or the basement 2 , the relation between rents and the provision of certain subsidiary rooms (bath-rooms, servants' rooms, lobbies, closets, etc.) 3, and that between rents and the existence of means of heating, particularly central heating 4. Other investigations have aimed at determining the relation between rents and certain characteristics of the occupants of the dwellings. Rents have for instance been classified by the occupation and social status of the occupants, or by their income 5 . The systems of classification adopted for these 1 See Chapter II, pp. 34-35, and Chapter IV, pp. 53. Examples of this kind of enquiry are to be found, for instance, in the Norwegian census of 1920, that of Basle in 1910, of Prague in 1921, and of Warsaw in 1919. 3 An analysis of this kind was to be found in the Austrian census of 1910, that of Berlin in 1920, and of Warsaw in 1919. * This question was studied with particular care in the Swedish censuses of 1910 and 1920. 5 An example of this analysis is to be found in the Scandinavian statistics. 2 — 12 — various ends are the same as those already discussed in previous chapters, so that there is no reason to dwell on them here. Some countries make enquiries into yet other questions. In some places, for instance, consideration is given to the question of the rent paid by sub-tenants to the principal tenant 1 . Elsewhere, an attempt is made to throw light on the influence of rent legislation by classifying dwellings according to whether they are subject to regulation or not 2. These are very interesting questions, but they seem of too specialised a nature, or not sufficiently widely taken up, for the adoption of at all definite international standards in their respect. In conclusion, it may be mentioned that in addition to all these various data, the total sum spent on rent in all dwellings of a locality or country is sometimes calculated. This operation presents no problem of method, as it consists simply of an addition ; it is chiefly of interest for the study of fiscal questions and the calculation of the national income. It may be added that this type of enquiry is not very widespread 3. 1 Reference may be made here to the Munich census of 1904-1909, and the Swedish enquiry into small dwellings of 1920. 2 This definition was, for instance, drawn in the Swedish census of 1920, and the Copenhagen census of 1921. 3 Examples are the Swiss census, 1920, those of Stockholm, 1910 and 1921, Copenhagen, 1921, and Prague, 1921. CHAPTER VI STATISTICS OF THE HOUSING MARKET In statistics of the housing market, the object of which is to determine the supply of and demand for vacant dwellings and to follow the fluctuations in these, a distinction must be made between two fundamentally different sets of data : those which aim at determining as exactly as possible the actual supply and demand in the housing market, and those which may be described as " indexes " and give only an indication of fluctuations of supply and demand without defining the amplitude of such fluctuations. The first set of data are obtained from censuses of vacant dwellings and systematic estimates of the housing needs of the population (which sometimes accompany the censuses); those of the second type consist principally of the statistics published by the public housing offices concerning their activities. For the sake of completeness, certain statistical series may be mentioned, based on the simple qualitative declarations of competent persons. CENSUSES OF VACANT DWELLINGS Hitherto, the census method has mostly been applied to only one of the parties in the housing market, that of supply ; for the other, that of demand, lends itself only with difficulty to an operation of this kind. The census properly so called is therefore used as a rule only for determining the number of vacant dwellings. The number of households in need of a dwelling is usually obtained by way of estimates based on various demographical statistics. — 74 — A census of vacant dwellings may be undertaken either as a separate operation or as part of a general census of dwellings. Most countries where general censuses are undertaken adopt the latter method. The information given concerning vacant dwellings in the enquiries varies from one place to another, but the distinction between inhabited and vacant dwellings is nearly always drawn. It will be remembered that it was recommended above for the classification of dwellings by intended and actual use 1. The very fact that general censuses of dwellings are on such a large scale means that they can be undertaken only at relatively distant intervals, as a rule ten years. It is obvious that this cannot be enough for following fluctuations in the housing market. The consequence has been that in several localities a special census of vacant dwellings is undertaken regularly once a year. Investigations of this kind are made particularly in Austria, Denmark, Germany, the Netherlands, Norway, Sweden, and Switzerland, but nearly all of them are still purely local, and it is perhaps only in Germany that a first effort has been made to standardise the various municipal censuses on a national basis. It should be observed that the season of the year in which the census is taken is of some importance. In many countries it is the custom to concentrate the expiry of leases in a certain quarter of the year. The number of vacant dwellings will therefore vary considerably according as the census is taken before, during, or after that quarter. In actual fact, most of the local investigations referred to above take the census in the last months of the year (or the first days of January), i.e. at a date following the removal quarter, which is usually in autumn. Definition of Vacant Dwellings Whether vacant dwellings merely form one group in the classification used for the general census of dwellings or represent the basic unit for a census limited to vacant dwellings, it is essential that the term " vacant dwelling " should be carefully defined. There is no need to reconsider the definition of the dwelling itself, for that already adopted in the discussion of statistics of dwellings can very well be applied here. 1 Chapter II, pp. 31-32. — 75 — The state of vacancy of a dwelling, on the contrary, may be interpreted in fairly different ways. If these are to be standardised for purposes of international comparison, the best would probably be, here as elsewhere, to adopt the widest conception and to consider any habitable dwelling vacant if, at the date of the census, it is unoccupied and immediately available for occupation. This would exclude dwellings in buildings still under construction, or undergoing important repairs, or, on the other hand, about to be demolished. It would similarly eliminate dwellings the occupation of which has been prohibited, and those for which the necessary formalities for obtaining a permit of occupation have not yet been concluded, where such formalities are compulsory. The definition would also exclude those vacant dwellings which the owner can use a t any time, although leaving them temporarily unoccupied — such as the houses used only in summer or winter, as the case may be — and those reserved for other purposes than habitation. In all these cases the logical application of the definition adopted raises no difficulties and is generally observed. On some other points, on the contrary, the interpretation is more open to question. There is, for instance, the case of a dwelling actually empty at the time of the enquiry, but already the subject of a lease which comes into force only at a later date. In theory, the dwelling may be considered vacant, but in practice this vacancy is of no interest as it will cease comparatively soon, and it is more or less impossible to use the dwelling for any other purpose in the interval. It seems that the latter point of view should be preferred. On the other hand, certain statisticians include among vacant dwellings those for which notice to terminate the lease has been given, or the lease has not been renewed, so that they will become vacant during the quarter when the enquiry was made. The method is somewhat disadvantageous, however, for there is nothing to prove, on the date of the census, that the dwelling will not be rented again on the expiry of the lease by the same or another occupant. The idea of anticipating the future is, moreover, entirely contrary to the very purpose of the census, which aims at giving a view of the actual position on the date of the return. The terms of the definition accepted here involve including among vacant dwellings all flats to be let furnished and onefamily houses for sale with immediate possession. In most »existing statistics these two groups are dealt with separately — 76 — or even excluded. The latter alternative cannot be accepted, however, if the statistics claim to be complete, nor does it seem expedient to include these two sets of vacant dwellings with the rest, for they satisfy different needs, and in some measure belong to special markets which should preferably not be confused with the general market of rented dwellings. The best would therefore be to count these special groups separately. In several countries, finally, a special heading is reserved for separate rooms to let. In fact these generally do not correspond to the definition of the dwelling adopted, as they form only a fraction of the dwelling which the occupants are prepared to cede, whether furnished or not, to an outsider. Yet any complete statistics of the housing market must take them into account ; they may therefore be enumerated in a special class, being counted as basic units of a separate type. Classification of Vacant Dwellings and Comparison with Other Data Vacant dwellings are usually classified in different ways. The most fundamental system of classification, as in other branches of housing statistics, is that according to the number of rooms. It is very generally used, but if it is to have any real value from the international standpoint, the term " room " must be defined and standardised in the manner advocated at the beginning of this study. Among other forms of classification, reference may be made to those already indicated in connection with general censuses of dwellings : the story on which the dwelling is situated or its situation with respect to the street, the rent-group to which it belongs, the connection between the dwelling properly so called and premises intended for business activity, such as an office, shop, or workshop. Other forms of classification belong more definitely to the statistics of the housing market. These are, on the one hand, that based on the length of time the dwelling has been vacant at the date of the census, and, on the other hand, that distinguishing between dwellings in newlybuilt houses and those which have already been inhabited. The number of vacant dwellings as an absolute figure is not of great interest in itself. If it is to be properly interpreted it has to be brought into relation with other data. — 77 — The most usual comparison is that with the total number of dwellings in existence, which eliminates the very variable factor of the size of the urban aggregation. It should be observed, however, that the significance of this percentage is far from the same everywhere, for it depends on several factors whose importance differs with the urban aggregation. This fact became very clear when an attempt was made to establish a normal percentage as a means of describing a satisfactory state of the housing market. Hasse's rule, so called, which fixes the normal ratio at 3 per cent., was soon found inapplicable in many localities. The fact is that the' stock of vacant dwellings has to play a very complex part. It forms a reserve which has to meet, in the first place, the needs of the newly formed households seeking a home and of persons coming into the town for the first time, and, secondly, the current needs of households living in the locality who have to move either because the dwelling they occupy no longer suits them or because it is to be demolished, transformed, or used for other purposes. The more intense these various needs, the larger the stock must be. Thus a town in the full swing of development with a high marriage rate or a large volume of immigration must have a larger reserve of vacant dwellings than one of the same size but more stagnant, in which the population has been stationary for some time. Similarly, a very old town consisting largely of buildings which have to be replaced needs a much larger reserve than another which, other things being equal, consists mainly of quite new buildings. As a new dwelling cannot be built in a day and as the various needs do not appear with constant regularity, it is also necessary to have a certain margin, so as to satisfy them as and when they occur. This margin should, moreover, be all the wider as a whole when there is more variety in the size, situation, and comfort of the dwellings, so that each type of need may receive equal satisfaction. The fact that the kind of housing needed has to be taken into account in interpreting the percentage of vacant dwellings shows the interest there would be in comparing the number of vacant dwellings, i.e. those offered with that of the dwellings demanded. It need hardly be added that an examination of the ratio of supply to demand is fundamental in the study of any market. The determination of the demand for dwellings has not often — 78 — been undertaken, for it is a very difficult operation. To count up the actual number of persons in search of a dwelling would mean a very large-scale enquiry, for every person would have to be questioned on his intentions in this respect. It can therefore be undertaken only in connection with general censuses 1 . When it has been wished to determine the housing demand periodically in order to compare it with the number of vacant dwellings recorded by the annual censuses, all that has been done as a rule has been to make a more or less approximate estimate. Various calculations have been made for this purpose in certain localities, chiefly in Germany, the Netherlands, and Switzerland. They are based as a rule on the statistics of marriages, divorces, deaths, immigration and emigration, and data concerning removals in general. The value of the results obtained varies with the availability of the data, the strictness of the methods employed, and the nature of the needs to be estimated. Owing to this diversity in the problems raised and in the possibilities of solving them, it seems inexpedient to propose international standards in this connection. It would be of great interest, however, to continue the researches in this field 2. STATISTICS OF PUBLIC HOUSING OFFICES It was stated at the beginning of this chapter that these statistics could give only an indication of the fluctuations in the housing market, without offering a precise view of the actual volume of supply and demand, or even of the amplitude of the fluctuations. Sometimes the public housing offices have absolute control over the housing market (when the law makes it compulsory to apply to them for any kind of housing trans1 An investigation of this kind was undertaken, for instance, on t h e occasion of t h e German housing census of May 1918, when every married woman or widow who h a d no dwelling of her own on t h e date of t h e census h a d t o state whether she proposed to set up a home of her own after t h e war, and if so in w h a t commune. 2 The following investigations m a y be cited as worthy of interest : STATISTISCHES A M T D E R STADT ZÜRICH : " Die Erhebungen über den Zürcher Wohnungsmarkt " , in Statistik der Stadt Zürich, No. 15. " Wohnungsvorrat u n d Wohnungsbedarf in der S t a d t Bern, Anfang 1926 " , Beiträge zur Statistik der Stadt Bern, No. 7. Dr. STOWESAND : " Der notwendige Mindestvorrat an Leerwohnungen " , Verhandlungsbericht über die Konferenz der Vorstände Statistischer Aemter Deutscher Städte, 1913. — 79 — action, for instance), and then the statistics may be considered to give an appropriate measure of the volume of supply and demand. But hitherto these conditions have been realised only in very exceptional cases, during periods of acute crisis, and they have not been maintained when the situation on the market has returned to the normal. There therefore seems little justification for stressing these possibilities, which appear to be altogether temporary, and in addition arise very seldom. The essential purpose of the statistics of public housing is not of so much interest here. In general, these statistics are compiled not to serve as an index number of fluctuations in the housing market, but simply to show the growth of the activities of the institutions concerned. But as this is of some interest from the point of view of housing policy, it may therefore be given consideration. The statistics of public housing offices have not yet been developed very far. They are to be met only in localities where institutions of this kind are systematically organised and it is considered necessary to publish reports on their work. They are, therefore, strictly local, and Germany is practically the only country in which a first effort at standardising the data of the different municipal offices has been made. There are several respects in which progress might be made at little cost, both in improving the statistics themselves and in making them more comparable internationally. In the first place, it would be necessary to define exactly what is meant by vacant dwellings asked for and supplied, as this varies with local customs and the administrative regulations in force in the different offices. For the term " vacant dwelling " itself, it seems that the definition already suggested for censuses of vacant dwellings might be adopted as it stands for the statistics of housing offices. With respect to demand, it may be observed that in several cases a distinction is made between more and less urgent demands ; that is to say, between the demand of persons who have no dwelling of their own or are about to lose that which they occupy, and the demand of persons who simply wish to change their home because that which they are occupying no longer suits them entirely 1 . This is, no 1 This distinction is made in the forms prepared for organising the statistics of the German housing offices on a national basis. In Vienna the present statistics on demand are limited to the more urgent needs. — 80 — doubt, a distinction which ought to be made, but the criteria used must be clearly defined. Similarly, with respect to supply, a distinction should be made between the actual offer of a dwelling immediately available and the anticipated offer made some time in advance of a dwelling which it is expected will be vacant in the near future. It is also essential that very strict measures should be taken to ensure that all the applications recorded in the registers are still valid. It seems as if the only satisfactory means of arriving at this end is to make it compulsory, on pain of being struck off the list, to renew unsatisfied applications periodically. This was, in fact, one of the most formal of the resolutions adopted at the thirty-second Conference of German Municipal Statisticians in 1924, when the organisation of the statistics of public housing offices was discussed *. Even so, the period for renewal ought to be fixed as uniformly as possible ; the most suitable period would seem to be one month. The systems of classifying the data of the supply of, and demand for, dwellings vary somewhat from one country to another Some of them are probably necessary if it is wished to illustrate completely and in detail the growth and activity of the housing offices. For instance, a classification must first be made by the nature of the application. A distinction will thus be made between : Applications not satisfied at the beginning of a given period ; Applications received during t h e period ; Applications which could n o t be satisfied through t h e office ; Applications withdrawn by t h e applicants, either because they were able t o satisfy their own requirements Or because t h e y changed their minds during the period ; Applications struck off the list because not renewed within the period prescribed by t h e regulations. By striking the balance between the applications thus added to the register and those withdrawn, the number of those not satisfied at the end of the given period is obtained, a figure which forms the starting point for the data for the following period. The length of the period covered each time should also be standardised, as it varies somewhat from one case to another. 1 VERBAND DEUTSCHER STAEDTESTATISTIKER : Verhandlungsbericht über die am 12. und 13. September 1924 abgehaltene XXXIl. Konferenz der Vorstände Statistischer Aemter Deutscher Städte ; Kommunale Wohnungsmarktstatislik, p . 11. — 81 — The most usual seems to be the month, which is suitable for general adoption. This classification should also be combined with one based on certain characteristics of the dwellings themselves. The most important appear to be the size of the dwelling and the question whether the premises are used exclusively for habitation or partly for other purposes as well. Among these various data, that of the applications not satisfied at the end of each period, and of those recorded during the period, lend themselves best to the construction of an index number of fluctuations in the housing market. It is preferable, however, not to use these figures in their absolute form, for the number of applications may vary considerably, owing to growth in the activity of the offices, without any relation to real changes in the market. So as to remove at least part of this source of error, the ratio of demand to supply is usually taken, and its fluctuations are followed from one period to another. But this index is not altogether exact, for, if so, it would have to be admitted that the two sides of the work of the offices grows to the same extent, which is obviously not always the case. It is true that the changes are as a rule very gradual, so that the effect is scarcely noticed on short-period fluctuations. It follows that the indexes constructed by means of the statistics of housing offices are most reliable for the study of the latter fluctuations. For those of longer duration, greater caution is needed in the interpretation. INDEXES BASED ON QUALITATIVE ESTIMATES Reference may finally be made to the method employed, for instance, in Sweden, for estimating the^situation in the housing market — a method based exclusively^on the qualitative estimates made by persons acquainted with the local situation. Questionnaires are sent periodically to the municipal councils of localities of some importance, and to special correspondents in small localities, asking them to state whether, judging from the information at their disposal, they consider the number of dwellings offered to be " sufficient ",rt" hardly sufficient ", or " definitely insufficient " to satisfy the demand. The localities are then classified and counted in three groups according to the replies received. The three sets of figures obtained by the — 82 — regular repetition of this operation thus give an index expressing in some measure the general situation of the housing market 1 . Such a method is clearly only of real interest if the persons responsible for making the periodical estimates have a fairly reliable practical knowledge of the situation, and if it can be assumed that the estimates made for different localities are of much the same value. If these conditions can be realised, it is certain that a method of the kind may yield interesting results at small cost. It may be added that it has been applied, apparently satisfactorily, in another field, that of the labour market 2 . Perhaps one day it may also be found useful with respect to the housing market. 1 These operations are carried out separately for all localities of some importance, and for all small localities on the one hand, and for small dwellings (two rooms and a kitchen), average-sized dwellings (three to five rooms and kitchen), and large dwellings (six rooms and kitchen) on the other. 2 In this particular case, the heads of a certain number of undertakings belonging to different industries are requested to report periodically on their views of the state of activity in their firms or the amount of employment they can give. The replies are then compiled in the manner indicated above, so as to give some kind of index of the conditions in the labour market. The chief examples of this type of data are to be found in Germany (Reichsarbeitsblatt) and Sweden (Sodala Meddelanderi). CHAPTER VII STATISTICS OF FLUCTUATIONS IN TOTAL HOUSING SUPPLY The term " statistics of fluctuations in total housing supply" is taken to cover all the statistics used for determining any quantitative or qualitative change in the total supply or stock of dwellings and buildings. These statistics, therefore, form a direct complement to the housing censuses, which give only a view of this stock at a given moment. Fluctuations in the total housing supply are caused primarily by the construction of new dwelling houses, and the transformation or demolition of those already in existence. It follows that the statistics relating to these points are the fundamental element of the statistics of fluctuations in the total housing supply. They are derived, as a rule, from the public services responsible for supervising building operations. This fact sometimes makes it difficult to give them the exact form or scope desired, for theoretical considerations have to give way to administrative necessities. Thus it may happen that the data are established in a form bringing out more clearly the activity in the building industry than the consequent fluctuations in the total housing supply. As a rule, however, this is merely a formal defect. It should be made clear that it is not proposed to examine here statistics relating to the construction, transformation, and demolition of buildings other than those used for living in, as these do not belong to the field of housing policy. The distinction between these two branches of the statistics of building activity is in fact very generally observed. Statistics of fluctuations in the total housing supply are to be found in one form or another in most of the countries where housing censuses are taken, but the degree to which they are — 84 — developed still varies considerably. In some cases they are compiled by national services, and cover, if not all localities, at least the more important centres, as in Denmark, France, Germany, Great Britain, Netherlands, Norway, Sweden, and Switzerland, as well as in Canada, New Zealand, the Union of South Africa, and the United States. In other countries, such as Austria, Czechoslovakia, Hungary, Italy, and Poland, this work is still in the hands of the municipalities and is undertaken only in the principal centres. To obtain a complete view of the fluctuations, it is not enough to consider only all the dwellings added to the existing stock, or withdrawn from it owing to structural activities. Account must also be taken of those withdrawn from the available stock owing to accident (fire, flood, earthquake, war, etc.), or to a simple decision of the authorities prohibiting their use. On these latter points, however, there are very few data available \ for as a rule such information does not come within the competence of the services supervising buildings which furnish the statistics of fluctuations. They must therefore be obtained from entirely different sources of information — insurance companies and inspectorates — and these are not always available. In actual fact, the gaps caused by the omission of these data seem inconsiderable. It often happens that dwellings which can no longer be used owing to an accident, or the use of which has been prohibited, are, very soon made habitable or demolished to make room for new buildings. In this case they are included in the statistics, and the only error is a certain delay in registration, which is no doubt regrettable, but has hitherto not seemed serious enough to call for the compilation of more exact statistics. It may be added that to give a special heading to dwellings declared uninhabitable is of no value in the determination of fluctuations in the total housing supply, unless the authorities which have competence to declare them uninhabitable also have the power to have them evacuated and to secure the observance of the prohibition, a condition which, in the present state of the law, is not always fulfilled. 1 As an example, reference may be made however to the German statistics, in which dwellings destroyed by fire are counted with those demolished, and the Dutch statistics, which have a special heading for dwellings declared uninhabitable. — -85 — T H E DIFFERENT FORMS OF STATISTICS OF FLUCTUATIONS IN THE TOTAL HOUSING SUPPLY AND THEIR CHARACTERISTICS Statistics of fluctuations in the total housing supply of dwellings may take several different forms, to be distinguished by the source of information used and the nature of the data obtained. In the first place, there are the two sets of statistics obtained in connection with the work of the authorities responsible for supervising building. The first set is based on the supervision exercised before the building is undertaken — the submission of plans or granting of building permits — and the second, on the supervision carried out after the work is completed — check that the construction complies with the approved plans, or grant of a permit to occupy Among the statistics compiled on a national basis, the first of these sources is the one generally used in non-European countries, and the second in European countries. In some of the latter, however, Germany, Netherlands, and Switzerland, for instance, both sets of statistics are compiled side by side, as also in many of the municipal statistics. There can be no question that the statistics relating to the work actually done give a more exact view of the state of affairs. Statistics based on building permits have one disadvantage and one advantage compared with these : one disadvantage, because they are less exact owing to the very unequal periods that always elapse between the fulfilment of the formalities recorded by these statistics and the actual starting of the work (sometimes even, the work is not started at all, or is not completed) ; but also an advantage, for, owing to these very delays, the data give an anticipated view of future construction and the results to be expected. But these are clearly not the considerations actually dictating the choice between the two methods, which depends above all on how building is supervised in the different countries, and on the consequent facilities for preparing one or other of these sets of statistics. In addition to these statistics, which may give a view of building activity as a whole and the consequent fluctuations in the total housing supply, reference should also be made to data of more limited scope, namely, those relating exclusively to building carried out with the assistance of the authorities, or — 86 — through the medium of public services or large private organisations for constructing cheap dwellings. Statistics of building subsidised by the public authorities are to be found in more or less complete form in most countries where there is a system of official assistance for the building industry 1. Statistics of the building activities of large public or private institutions are to be found chiefly in Belgium, France, Great Britain and Italy, as also in the United States, where this type of institution has been developed very far. Owing to their very nature, these last two sets of statistics can give only partial information on fluctuations in the total housing supply, unless subsidised building, or the work of the bodies in question, represents the whole building activity of the country, as may happen in a time of crisis, when all private building comes to a stop. Such conditions are the exception, however, and as a rule the importance attached to these data from this point of view is due largely to the lack of other and more satisfactory figures. The statistics are altogether appropriate, on the contrary, for following the building activities of the Government or the work of the bodies from which they are obtained, and as such they are a very useful source of information on housing policy. It is therefore of interest to develop them on similar lines to those of other statistics. Owing to their particular object, these statistics give, more often than others, not only figures of the buildings finished or plans approved during specified periods, but of all building in progress at fixed dates. They often give also detailed information on the nature of new buildings, the cost of production, the way in which the necessary capital was raised, etc. As a rule, all these various statistics are based on two units, 1 One of the most remarkable examples of these statistics are those compiled in Great Britain by the Ministry of Health on the basis of the reports of the local authorities. The method of compiling the statistics differs somewhat according to the regulations in force when the work was undertaken (1919 Act, 1923 and 1924 Acts, and 1925 Act) and according as the subsidies are granted by the State and the local authorities together or by the latter alone. Data are given of the work in progress at a given date, a distinction being drawn between the houses for which only the foundations have been completed, those which have been roofed in, and those completed ready for occupation. For the buildings undertaken under the 1919 Act a distinction is also made between houses of which the first-floor joists have been laid, those ready for roofing, and those ready for painting. These statistics are completed by more summary data on non-subsidised building. (Communication from the Ministry of Health.) — 87 — the building and the dwelling, which are analysed side by side. This makes it possible to examine the problems from two different points of view, which are mutually complementary. It may be added that in some cases use is made of a third basic unit, the room. There is no need to return here to the definition of these basic units, for it is clear that, from the point of view of the possibility of comparison, there is every advantage in adopting strictly the same definitions as those suggested for other branches of housing statistics 1. Some difficulties may arise in this respect owing to the fact that the statistics of building activity are derived from the working of administrative services. But it seems that the definitions proposed are sufficiently near to the actual fact to satisfy the practical requirements of an administrative department. In certain cases, finally, an attempt is made to measure the importance of building operations, by referring not to the number of buildings, dwellings, or rooms, but only to the total cost. Information of this kind is found most often in the statistics of building permits, being based on the declarations made to the authorities on the estimated cost of construction 2. In the statistics relating to the work done, this type of information usually expresses the amount for which the property is insured 3 ; in the statistics of subsidised building and those of building by special institutions, these figures are calculated from the credits granted by the Governments and the expenditure of the institutions. Information on the value of buildings is usually expressed in the form of totals for all buildings. If these are compared with the number of buildings, dwellings, or rooms to which they relate, the average cost of building these various units may be calculated. This information is particularly of interest for following the activity of the building industry, for it gives, on the one hand, a means of estimating the value of authorised or completed work, and on the other, a view of the volume of this activity which is often more exact than might be obtained from the other basic units. The work in volved in the erection of a building, 1 See Chapter II, pp. 31, 35, and 36, and Chapter III, pp. 43-45. For example, statistics of Canada, Great Britain, the Netherlands, New Zealand, the Union of South Africa, and the United States. In Canada and the Union of South Africa, these data are the only ones published on a national basis. 3 Statistics of Basle and Zurich. 2 — 88 - or even of a dwelling or room, may vary considerably according to circumstances, whereas the cost of a building operation, or the value of a building, represents fairly closely the total labour and material consumed. It is necessary, however, if information of this kind is to be usable, that the monetary unit should remain stable. When prices are rapidly fluctuating, the figures cannot serve as an index of building activity unless this source of error can be eliminated, which is always a delicate operation. In this respect an interesting experiment made in the United States may be mentioned 1. Reference may finally be made to statistics of an altogether different kind, prepared in certain cases, which may also give useful information on fluctuations in the total housing supply. These are the statistics obtained from financial registers kept in connection with the taxation of house property. The increase and decrease in taxable value may in fact be used to determine the number of cases of building, transformation and demolition of dwelling houses. If the system of taxation in force applies to every building without exception, a complete return may be obtained of newly-built houses as well as of those demolished. The figures relating to the transformation of buildings will be much less satisfactory, for they will relate only 'to operations of sufficient scope to affect the taxable value of the property, and many changes in use or alterations will thus be left out of account. These statistics have moreover the disadvantage that they can be compiled only after some delay. In the absence of more satisfactory data, however, they may be of appreciable service 2. 1 The method consists in the first place in expressing, in the form of an index number (base : 1914), a series of estimates made once a year of the aggregate value of all building, as shown by the permits issued by t h e public authorities during the year. Next an index number of the cost of construction (base : 1914) is calculated by combining an index number of the wholesale prices of building materials with an index number of the-cost of labour (wages) in the proportion determined by a special investigation into the cost of construction of different types of building. By relating t h e index number of the estimated aggregate value of all buildings constructed to t h a t of t h e cost of construction, an index number is obtained expressing the approximate amount of building done. (Monthly Labour Review, J u l y 1925, pp. 173-176.) 2 The best known example of these statistics is t h a t of France, where for each department, under the head of " increase in taxable value " , the number of whole buildings and the rental value of buildings and additions to buildings, are given, and under the head of " decrease in taxable value " , the number of cases of total demolition and the rental. value of total or partial demolitions (houses and factories separately). — 89 — All the statistics here reviewed are compiled periodically. As a rule, the smaller the field covered, the shorter the interval ; thus purely local statistics are usually compiled once a month, whereas statistics covering the whole country are compiled only once a quarter or once a year. From the point of view of international comparison, annual statistics appear sufficient. CLASSIFICATION OF BUILDINGS AND DWELLINGS AND COMPARISON WITH OTHER DATA The fundamental classification which predominates in all statistics of building activity is that based on the kind of work involved. As a rule, three large classes are defined — construction, transformation, and demolition. In addition, the cases of transformation are divided into those which result in increasing the stock of dwellings and those which result in reducing it, so that the actual fluctuation in the supply may be determined. It should be observed that the data to be placed under these two heads should not, in theory at least, be collected in the same way if the object is to determine the activity of the building industry, as when it is to determine the fluctuations in the total housing supply. If it is proposed to obtain a view of the total activity in the building industry, the figure to be sought is the total number of buildings and dwellings actually in hand at a given date. In point of fact, this sort of information is rarely given 1 . In the absence of such data, use may be made of the number of dwellings in respect of which work has been begun or completed during a given period, remembering, however, that the record must always be taken at the same stage of the work, whether the beginning or the end. If the fluctuations in the total housing supply are to be determined, on the contrary, the return ought to relate to the moment when the stock is actually affected by the building operations. Cases of construction must, therefore, be recorded when the work is completed, and those of demolition when it is begun. In cases of transformation, the work should, properly speaking, be recorded both when it is started and when it is completed, so that, first, the dwellings made temporarily 1 There are a few examples : for instance, in the statistics of the municipalities of Amsterdam, Oslo, and Warsaw. — 90 — uninhabitable may be deducted from the stock, and then those again made available be added. But as this work of transformation usually does not last long, nor does it always prevent the premises from being inhabited, the method is seldom applied, and it is considered sufficient to give the net result of the work when completed. It should be observed that in certain cases this classification is extended by a few supplementary subdivisions ; thus a distinction is made between new buildings on vacant land, and rebuilding on land already built on \ and in some municipalities alterations are classified according to their special characteristics 2. For purposes of international comparison, however, it seems unnecessary to enter into these details. The different basic units may also be classified according to the persons or institutions undertaking the work. This classification is only used in some countries 3, being applied sometimes to the building unit, sometimes to the dwelling unit, and sometimes to both. The systems of classification vary somewhat in the different cases, but in general there are three main classes. The first two relate to work undertaken in the interests of the community, either by the authorities or by public utility societies, and the third to work undertaken in private interests by building societies or private persons. The subdivisions of these classes are to be considered rather as differences in detail, often the result of local conditions 4 . For purposes of international comparison, it seems that these can 1 E.g. statistics of the Netherlands, Stockholm, and Warsaw. Thus in the statistics for the municipality of Vienna, a distinction is made between work of enlargement, increasing the height of the building, fitting u p of attics, internal alterations. In those of the municipality of Basle, special headings are given to work relating to kitchens, laundry accommodation and bath-rooms ; roofs, terraces, and balconies ; and, finally, changes in use classified according to the new use of the premises. 3 E.g. statistics of certain municipalities of Germany, Netherlands, Switzerland, and those of Vienna, Warsaw, Prague, and Oslo. 4 A distinction is for instance often made between the authorities, i.e. between the State and the municipality ; or in some cases, as in Copenhagen, building actually undertaken by the authorities themselves is distinguished from t h a t merely subsidised. In Prague, a distinction is made between building societies of officials, of workers, and others. In Zurich, public utility societies are separated from others. In The Hague, private building undertaken for speculative purposes is distinguished from t h a t undertaken by private persons to satisfy their own needs. 2 — 91 — be neglected and that the three large categories already mentioned may be considered sufficient. The other systems of classification to bé found in statistics of building activity relate to the actual characteristics of the buildings and dwellings. They involve most of the distinctions already mentioned in the discussion of the statistics of buildings and dwellings. Buildings are classified principally according to their use, a distinction being made as a rule between those intended exclusively for living in, those intended only partly for living in, and those intended solely for other purposes. (The latter group remains outside the field of housing statistics.) In some cases other factors are taken into account : the nature of the building (one- or two-family houses, buildings with several flats, buildings intended for occupation by other than family groups, mixed buildings containing shops, workshops or offices) x or its size, which may be indicated by a classification according to the number of stories, dwellings, or rooms 2, or by a statement of the superficial area 3. Some statistics, finally, give details of the materials of which the building is made 4, of the domestic equipment (means of lighting and heating, running water), the available subsidiary rooms, adjoining gardens, etc. 5 For dwellings, the obvious system of classification, which is to be found in the majority of more detailed statistics, is that based on the number of rooms 6. Sometimes the analysis is 1 Thus the statistics of the chief Swiss municipalities classify dwelling houses according as they are intended for one family, two families, or several families ; and mixed buildings according as the premises intended for other purposes than habitation are shops, inns or restaurants, or workshops. In the United States statistics of building permits, a distinction is drawn between one-family dwellings, two-family dwellings, one-family and two-family dwellings with stores combined, multifamily dwellings, multi-family dwellings with stores combined, hotels, and lodging-houses. 2 This threefold classification is to be found, for instance, in the Swedish statistics, as in those of the municipalities of Basle and Oslo. Buildings are classified according to the number of dwellings in the German statistics, and in those of Prague ; according to the number of rooms in the Stockholm and Zurich statistics. 3 Statistics of Berlin and Oslo. 4 The Swedish statistics, for instance, distinguish between buildings of stone and wood, the Berlin statistics between permanent and temporary buildings. 5 Statistics of Basle and Oslo. 6 National statistics of Denmark, Germany, the Netherlands, Norway, and Sweden and the statistics of Prague, Vienna, Warsaw, and certain municipalities of Switzerland and the Scandinavian countries. — 92 — carried so far as to give the habitable area *, but this is too complicated an operation to be of general application, so that in spite of its interest it cannot be recommended for international purposes. The classification of dwellings according to the number of rooms, on the contrary, seems fundamental, and it is regrettable that in certain cases it is not adopted in the final tables of fluctuations in the supply of dwellings. Every precaution must of course be taken to ensure that the definition of the term " room " (serving here as a unit for measuring size) has absolutely the same value as that adopted for this purpose in the censuses of dwellings. In some cases there are certain other systems of classification of less general interest, based, for instance, on the degree of comfort of the dwelling (specification of certain subsidiary rooms and domestic equipment 2 ), or on the rent-class to which the dwelling belongs 3. Finally, in certain statistics, the room is taken as the basic unit, and is also classified according to its characteristics. It is fairly usual to distinguish between the three classes already suggested : living-rooms, kitchens, subsidiary rooms, although the delimitation of these classes does not always conform to that proposed here. In some cases, a separate heading is given to attics 4 or workrooms 5. Once a regular return of the exact fluctuations in the total housing supply has been organised, it may further be useful, if its real significance is to be understood, to compare it with other data. Thus the number of new dwellings of different sizes, or of new buildings, may be compared with the total supply at a given date (determined by a census), and the percentage increase may be calculated. An even more significant comparison is that between the newly-created dwellings and the demand to be satisfied. For this purpose, the number of new dwellings available may simply be expressed as a function of the population of the locality, or the changes in the total housing supply may be compared with the changes in the population for whom these dwellings are intended. If bias in the results is to be 1 2 3 4 5 Statistics of the municipalities of Basle and Frankfurt. Statistics of Sweden and the municipality of Basle. Statistics of the municipality of The Hague. Swiss municipal statistics. Statistics of the municipality of Vienna (post war). — 93 — avoided, however, it is preferable to calculate the ratios between certain units rather than others. Thus, for accuracy, the number of new dwellings should be compared with that of the households for which these dwellings are intended, and if it is wished to make a comparison with the number of persons, it seems more just to relate this number to the number of new living rooms. Moreover, a distinction should be made between persons of different age groups, for these differ in importance from the point of view of density of occupation. The work of calculating these ratios is obviously part of the supplementary research to be undertaken by the national statistical services. They cannot be too strongly recommended to undertake this work as often as possible, for they are in the best position to know the data which can be compared with each other and the restrictions to be observed in interpreting the rates and indexes obtained. CONCLUSIONS T H E DIFFERENT HOUSING STATISTICS The various sets of statistics forming what have been called here " housing statistics " still constitute only an extremely heterogeneous whole, and the problem of standardising them arises both in the national and in the international field. These statistics relate to very different subjects, often aim at different ends, are undertaken by different services, depending sometimes on the central authorities, sometimes on the municipal authorities, and they therefore demand in the first place that they should be co-ordinated among themselves in such a way as to form a purely homogeneous system for each country. But while this work of standardisation is being carried out, it is essential to take into account also the needs of international comparison, which are steadily growing in importance. Housing statistics are unquestionably built up on general censuses, from which a complete and detailed view of the total supply of existing dwellings and their characteristics can be obtained, and of the prevailing housing conditions. These censuses may also give information on rents and the state of the housing market at the date of the investigation. But owing to their wide scope, they are carried out only at fairly distant intervals, as a rule every ten years, and it is therefore essential to complete them by periodical returns, which can be used to follow changes in rents (summary rent returns) and the constant changes in the housing market (special censuses of vacant dwellings, statistics of housing offices, etc.). At the same time it is necessary to be able to follow changes in the total housing supply which occur during the interval between censuses owing to new construction, the transformation of existing premises, and demolition or accidental destruction. This is the purpose of the statistics of fluctuations in total housing supply. — 96 — These various branches of statistics supplement each other. It therefore seems of particular importance that they should be undertaken on a general systematic plan, but it is not essential that they should be compiled by one and the same service ; all that is needed is a definite agreement between the competent institutions. If, however, one branch is not yet within the competence of the national authorities, as often happens in the case of periodical statistics, the necessary measures ought to be taken to ensure not only uniformity in the data compiled by the municipal authorities, but also their compilation in a sufficient number of localities with the appropriate characteristics for them to be considered representative of the country as a whole. The various branches of housing statistics serve above all to throw light on and direct housing policy, but they do so in varying degree. The censuses afford a complete and sound basis on which to judge of the general situation and base longperiod measures. The periodical returns make it possible to follow immediate results and take the necessary measures to deal with current problems. All the statistics form a valuable source of information for any town-planning action and any kind of social study. T H E STATISTICAL UNITS If it is proposed to make a coherent whole of these different statistics, it is necessary in the first place to standardise the fundamental conceptions on which they are based. The first task undertaken here has therefore been to define the different units to be found in housing statistics, and this in as precise a manner as possible, but yet so widely that they can be used for the different branches of the statistics and in all countries. Six principal units have been recognised ; the dwelling, the building with dwellings, the registered property (comprising buildings with dwellings), the room, the household, and the person. According to the branch of statistics in which these are used, they may serve either as basic units, i.e. the unit which is counted, measured, and classified according to its different characteristics, or as the unit of measurement, i.e. the unit which serves precisely as a standard for measuring the basic unit. Among these various units there are two which — 97 — seem particularly important, those of the dwelling and the household ; for it is one of the great principles of housing policy to secure its own dwelling for each household. The dwelling, the fundamental unit in housing statistics, may be defined either by the intended use of the premises or by the use actually made of them. The idea of intended use seems more interesting, however, for it brings out better the true — and, so to speak, permanent — characteristics of the premises. But in some cases it is difficult to apply, or even impossible or absurd, and then it must be supplemented by the idea of actual use. Adopting this compromise, the definition proposed as an international standard will be as follows : the dwelling is, in principle, constituted by a separate set of premises which, whatever their use, are intended for the occupation of a household, and have an independent entrance either from the street or from a public passage or staircase. Premises separated off from these, but clearly intended to form part of the dwelling, will be included with it. Premises not intended for living in, but used for this, will be considered to have as many dwellings as there are households occupying them. The same ideas of intended use and, subsidiarily, of actual use serve to distinguish buildings with dwellings from other buildings. A building with dwellings is considered to be any covered construction, whether isolated or separated from others by party-walls reaching from the foundations to the roof, intended altogether or in part for living in, or actually used for this purpose. (Dwellings consisting of huts, tents, wagons, caravans, ships, barges, etc., which are not covered by this definition, should be placed, if need be, in a separate class.) Registered properties are clearly delimited by the land registration offices, and housing statistics may, if need be, analyse the properties on which buildings with dwellings are situated. With respect to the room — a completely enclosed space — it is necessary to distinguish between three main types of rooms with fundamentally different characteristics, the basis being the intended use of the premises. These are (a) living-rooms (bedrooms, dining-rooms, parlours, etc.), which are the fundamental factor in dwellings ; (b) kitchens, the characteristics of which vary according to district and social class and therefore necessitate a special heading, and (c) subsidiary rooms, which, although of secondary importance, are not without influence — 98 — on the comfort and convenience of a dwelling ; servants' rooms and habitable attics are counted with living-rooms ; alcoves, on the contrary, are considered to be subsidiary rooms. The household is defined as a group of persons who together organise their domestic life separately from other persons. A distinction should, however, be made between family households and non-family households, owing to the fundamental difference in the housing conditions involved. It seems that the best line between the two is where the number of outside boarders exceeds the number of members of the family and domestic servants. In defining the household itself, it is more important to take into account the fact that a person shares in the life of the household than the form of the contract concluded between the occupants of the dwelling concerning their right to the premises. As the individual is considered to be a member of the household, it seems preferable to count those who ordinarily live in the dwelling, rather than those who happen to be there on the night of the census, for this gives a more representative view of the actual situation. The occupants of a dwelling will therefore be taken to include persons who, although absent on the date of the census, usually live there ; but persons who are there only temporarily will be excluded. In addition to these various basic units and units of measurement, account must also be taken of the manner in which the geographical units are defined, within which the various housing problems are analysed. The urban aggregation, as usually conceived on the basis of political and administrative boundaries, does not seem altogether satisfactory. For this almost inevitably leaves out certain districts of an altogether urban nature, while, on the other hand, it makes the geographical unit include certain rural districts. As housing problems differ fundamentally in urban and rural districts, it would be useful to adopt an entirely new definition specifically for the purposes of housing statistics. The urban aggregation would then include the whole of the area within which there is no kind of rural undertaking, most of the inhabitants, whether by their occupations or as consumers, being in direct and constant touch with the centre. It might perhaps be well to subdivide these geographical units into a number of smaller units or quarters, according to the prevailing type of construction : continuous structures (blocks of houses with several stories), semi-detached — 99 — buildings (houses of lesser elevation with gardens on one or more sides), detached buildings (houses of moderate elevation standing in their own grounds), with perhaps a separate heading for business and factory quarters. CENSUSES OF DWELLINGS It seems at first sight that the work of taking housing censuses might be considerably reduced by limiting it to enumerating the basic unit and recording certain essential characteristics, the other points of detail being studied by means of enquiries on the sampling system undertaken at the same time as the census itself. The latter information would be obtained with quite sufficient exactitude, and the economy in work so realised would be considerable. On this principle the census of dwellings would entail only two classifications, according to the intended and actual use of the premises and according to their size. The first of these classifications is used above all to define a certain number of fundamental large categories, offering definite headings under which to record the data collected by those national services which cannot accept as wide a definition of the dwelling as that suggested here. This arrangement will make international comparison possible, at least in the fields covered by the different statistics. The classification might take the following form : (1) Dwellings intended for family habitations : (a) used solely for this purpose ; (ft) used partly for this and partly for other purposes ; (c) not used — i.e. vacant ; (rf) temporarily used for other purposes t h a n habitation (without structural alteration) ; (e) used for other t h a n family habitation (hotels, etc.). (2) Dwellings intended for non-family habitation : (a) used accordingly ; (ft) used for family habitation ; (c) not used — i.e. vacant. (3) Premises not intended for habitation : (a) b u t used for family habitation ; (6) b u t used for other purposes t h a n family habitation. The classification of dwellings according to size is altogether fundamental, for they differ markedly among themselves from this point of view, and the differences are of great practical importance. The simplest and most usual way of expressing — 100 — the size of a dwelling is to give the number of its rooms, but then the nature of the rooms taken into account must be clearly indicated. Those referred to above as living-rooms, as also the kitchen (mentioned separately), seem most suitable for this purpose. The process of measurement might be further improved by taking into account certain subsidiary rooms, such as bath-rooms, closets, alcoves, and lobbies, but enumerating them all without distinction under an additional heading, or counting them as only a fraction of all the rooms in a dwelling. For other than family dwellings, a distinction should also be made between the large common-rooms and living-rooms of the ordinary size. The number of rooms in a dwelling is, however, only a somewhat rudimentary form of measurement, the significance of which may vary markedly with the size of the rooms themselves. Detailed information on this point would thus much increase the value of the above data. It could be obtained without too much work by undertaking systematic measurements in a certain number of dwellings selected on the sampling method. All the other characteristics of dwellings which it is of interest to know may be studied by the sampling method : the existence of the subsidiary rooms of most importance from the point of view of comfort (closets, bath-rooms, laundry accommodation) ; domestic equipment (means of lighting and heating, running water, drains); the neighbourhood of a garden ; the story on which the dwelling is situated ; the position of the dwelling with respect to the street (on the front, at the side, or on the back) ; the nature of the building containing the dwelling (one- or two-family houses or blocks of flats). Similarly, the nature of the right to occupy the dwelling may be studied : rented dwellings (and among these, dwellings or parts of dwellings sub-let and those let furnished) ; dwellings occupied by their owners ; service dwellings ; and those occupied free of charge. CENSUSES OF BUILDINGS WITH DWELLINGS Building censuses are included in housing statistics only so far as they relate to buildings with dwellings. They too might be somewhat simplified, on the principle already advocated for the censuses of dwellings. Among the various systems of — 101 — classification at present used in statistics of this kind, it would probably be sufficient to keep only two. The first would be that which distinguishes between dwellings used altogether for habitation, those of which only a part, although a relatively important part, is used for this purpose, and finally those in which only a small part has been equipped, with a view to housing a caretaker, for instance. The second system would be that based on the nature of the building. Buildings used solely or principally for habitation would first be classified according as they are used for family habitation or non-family habitation. The first of these classes would be further subdivided into buildings for one or two families and those with several flats ; or, even better, they would be classified according to the number of dwellings as well as according to the number of rooms. Buildings for other than family habitation would be grouped according to the bodies occupying them (hotels and boarding houses, hostels and almshouses, hospitals and nursing homes, educational institutions, religious institutions, barracks, prisons). The other characteristics of buildings would then be studied by the sampling method. This might, for instance, be used to bring out the number of stories in the buildings, the building materials used, their arrangement with respect to each other and to the street — leaving out, however, questions of internal arrangement, which seem better dealt with under the head of statistics of dwellings. It would also be interesting to examine buildings from the legal standpoint, distinguishing between the different methods of acquisition (purchase, inheritance, construction, other methods) and showing the personality of the owner (whether individual ownership or non-individual, and in the latter case whether a company run for profit, a private society, or a local or national public authority). These last forms of analysis may also be undertaken, it should be observed, with the registered property as a basic unit, as, owing to its legal nature, it lends itself well to this type of enquiry. CENSUSES OF HOUSING CONDITIONS The study of housing conditions may be divided into four parts, each of which involves a new analysis, which may be combined with the others. For the first two a new basic unit is introduced : the person, in addition to those of the dwelling — 102 — and building, so far the only ones taken into consideration. In the last two yet another unit, that of the household, is used. The first analysis consists of a simple classification of persons according to the different characteristics of the dwellings they occupy : the intended and actual use of the dwellings, the size of the dwellings, and the other characteristics of the dwellings examined by the sampling method. This analysis is followed by that of the density of occupation. The degree of occupation of the dwellings is first studied by classifying them both according to the number of living-rooms {including kitchens) and according to the number of persons occupying them. The occupants are also classified in the same way, so as to bring out the conditions in which the population is housed. In order to express in more condensed form the relation between the number of rooms in a dwelling and the number of occupants, a coefficient of density is calculated for each dwelling, generally in the form of the number of persons per room. The dwellings and occupants are then classified according to the coefficients relating to them. The average coefficients are also calculated. These various data at present constitute the most significant statistics of existing housing conditions. It should be observed, however, that the simple ratio between the number of livingrooms in a dwelling and the number of its occupants is still only a somewhat approximate measure of the degree of occupation. It would be interesting to be able to take into account also subsidiary rooms, and to make certain distinctions between persons according to age, sex, and the relationship between them. The study of density also involves examining the degree of occupation of buildings, so as to determine how far persons are crowded in a single building. For this it is sufficient to classify, on the one hand, the occupants according to the number occupying one and the same building, and, on the other hand, the buildings according to the number of persons housed in them. It would also be possible to calculate for the different types of building the average number of persons living in one and the same building. Finally, the density of occupation in the whole of the urban aggregation might be examined by comparing the area of the locality with the number of its occupants, but these calculations fall somewhat outside the scope of housing statistics. — 103 — The two remaining sections of statistics of housing conditions resume the analyses already indicated, but introduce a new idea, that of the household, and bring out its characteristics. The relation between housing conditions and the composition of households is first studied, distinguishing between members of the family, other relatives and friends, domestic servants, employees, and boarders. The special conditions created by the fact that several households may live together in one dwelling are then considered, for which purpose the housing conditions of such households are first examined, considering them as so many independent units, and then the conditions of the principal household are compared with those of the subsidiary household. In the last section an attempt is made to show the housing conditions of different social classes. For this purpose persons are usually classified according to the social status of the head of the household to whom they belong ; but in this respect it is difficult to establish any international standard, for the question goes definitely beyond the scope of housing statistics. It need merely be indicated that in principle this classification should rest on a threefold criterion ; the occupation, the status in the occupation, and the income of the head of the household, and that the following main classes should be distinguished : unskilled workers, skilled workers (with perhaps a separate heading for certain industries, such as the textile and mining industries), artisans, the middle classes, the comfortably-off, and the well-to-do. In the absence of such a classification, a distinction according to the quarter of the town might give some information on these points in localities where the separation of the social classes by quarters is relatively clear. STATISTICS OF RENTS The statistics of rents consist essentially in a classification of dwellings according to their rents, combined, if need be, with a classification according to the different characteristics of the dwellings and a determination of the average rent for the different types of dwellings. Two types of investigation are undertaken for this purpose — general censuses and periodical summary returns. The first are taken at relatively distant intervals, either in connection with housing censuses or as special enquiries covering all rented dwellings and all urban centres, and bring — 104 — out in detail the relations between rents and the different characteristics of the dwellings and their occupants. The second are undertaken at closer intervals, as a rule once a year, but are limited to a certain number of selected dwellings in various towns and take into consideration only the essential characteristics of these dwellings — in particular their size — the chief object being to follow fluctuations in rents. The present study has however left out of account data obtained from estimates of fluctuations in rents made with a view to constructing index numbers of the cost of living, as well as those furnished by enquiries into the conditions of life of certain social classes, both of which belong to other fields than that of housing statistics. The basic unit for statistics of rents is the dwelling. As the information on rents is usually obtained from the leases in force, it is natural to consider that the dwelling unit coincides with all the rooms covered by a lease. This definition obviously differs somewhat from that suggested for the other branches of housing statistics, but the differences do not seem to be very serious. More important differences arise out of the very varying manner in which the term " r e n t " is interpreted according to local custom. The logical standard seems to be that of including with the rent only what is strictly connected with the renting of the premises, as well as payments for the caretaking and internal maintenance of the building, if these are charged for separately. On the other hand, all additional charges for heating, lighting, running water, and the removal of rubbish, which are often made together with the rent, should be systematically excluded, as well as rates and taxes, unless these are expressly payable by the tenant under the law. Separate headings should also be made for the rents paid for dwellings let furnished, and for separate rooms or parts of flats sub-let by the principal tenant, as these two forms of occupation have special features which need to be studied separately. As regards the actual classification of dwellings according to rent, it is not possible to lay down a definite international standard owing to the differences in the monetary units used for defining the classes. It seems reasonable, however, that the number of classes should not be less than about ten, and not more than about twenty. The average rent, on the contrary, will be calculated for the different types of dwellings as defined in the statistics of dwellings and of housing conditions. — 105 — It should be added that if it is proposed to compare the average level of rents in the whole of one locality with that in other localities, or in the same locality at different dates, it is necessary to eliminate any influence of changes in the proportion of large to small dwellings. For this purpose the average rents of dwellings in each size-group must first be calculated, and then a weighted average of these averages, the same weights being chosen for each locality in accordance with a common standard. STATISTICS OF THE HOUSING MARKET Here, too, there are, on the one hand, the general censuses of vacant dwellings which form an integral part of the decennial censuses of dwellings, and, on the other hand, the enquiries undertaken at closer intervals on a more limited basis, and confined to vacant dwellings only. The term " vacant dwelling " should cover any habitable dwelling which is unoccupied at the date of the census and immediately available for occupation. Dwellings which, although vacant at the time of the enquiry, are already let for a later date will, however, be excluded. Dwellings to be let furnished and one-family houses for sale should preferably be classified apart, as they belong to a very special market. Single rooms to let, although not actually covered by the definition of a dwelling, might also be placed under a special heading. The fundamental classification for vacant dwellings is again that based on the number of rooms. A more detailed analysis of the characteristics of the dwellings might be undertaken, if necessary, in connection with the general censuses, on the model of that used for occupied dwellings. As the figures of vacant dwellings are not very significant in themselves, they are generally related to the total number of dwellings in the locality. Care must be used in interpreting this percentage, however, for its significance varies according to circumstances. The relation between the number of vacant dwellings and the housing needs of the population is more significant — in other words, the ratio of supply to demand in the housing market. The determination of the second factor, however, is of some difficulty, although an attempt has been made to estimate it in certain municipalities on the basis of various statistics relating to changes in the population. — 106 — The statistics of housing supply and demand recorded in the public housing offices are used above all to illustrate the growth and work of these institutions. To make these data more comparable, it is necessary to define exactly the terms " supply of " and " demand for " dwellings,. and to distinguish, for instance, between urgent demands (those of persons without a dwelling, or about to leave that which they occupy) and less urgent demands, as also between the actual supply of immediately available dwellings and the anticipated supply of those available in a few weeks. It would also be desirable, if possible, to show separately for dwellings of different sizes or in different rent groups : the number of applications (supply and demand separately) not satisfied at the beginning of a given period, preferably one month ; the number of applications received during that period ; the number satisfied through the public housing office ; the number withdrawn by the applicant ; and the number struck off the list because not renewed within the period prescribed by the regulations. It would also be necessary for the offices to check the register systematically in order to strike off applications no longer valid, and the rules adopted for this purpose should be as nearly the same everywhere as possible. The figures of applications not satisfied at the end of each period, or of those recorded during the period, are often used to construct an index by which the fluctuations in the housing market may be followed with more or less reliability. For this purpose the ratio of supply to demand is generally taken, so as to eliminate in some measure any influence on the absolute figures of the growth in the work of the offices. Owing to the imperfection of this correction, it is as well, however, to interpret these indexes only with the greatest reserve. For the sake of completeness, reference may also be made to certain indexes based on simple qualitative estimates. This procedure, which has been of some service in other fields, might also prove useful in that of the housing market. STATISTICS OF FLUCTUATIONS IN THE TOTAL HOUSING SUPPLY The principal source of these statistics is the data derived from the work of the public authorities responsible for supervising the construction of dwelling houses. The figures are based on the supervision exercised either before the building is under- — 107 — taken — the submission of plans or granting of building permits — or after the work is completed — check that the construction complies with the approved plans or grant of a permit to occupy. The statistics relating to the work actually done are obviously more exact than those based on building permits, for there is always some delay, not entirely free from uncertainity, between the granting of the permit and the completion of the work. In addition to these statistics, reference may also be made to those relating solely to building carried out with the assistance of the authorities or through the medium of public services or private organisations for building cheap dwellings. These data, again, give more information on the work of the bodies in question than on fluctuations in the total housing supply. The various statistics of fluctuations in the total housing supply are based on two units, the building with dwellings and the dwelling, which are analysed from different points of view. In the first place, the nature of the work done, or to be done, is brought out (construction, transformation resulting in an increase in the supply, that resulting in a reduction, and demolition). It should be noted, however, that for the purpose of determining fluctuations in the total housing supply, it is necessary to record the operations at the moment when they actually affect it. Work of construction should therefore normally be recorded when the work is completed, and that of demolition when it is begun. The work of transformation should normally entail a double record, namely, when it is begun and when it is completed. In practice, however, the latter is considered sufficient. The second classification is based on the nature of the parties who undertake the work (public authorities, public utility .societies, companies run for profit, private persons). These analyses are further combined with classifications based on the fundamental characteristics of the dwellings (number of livingrooms, including kitchens) and of the buildings (one-family houses ; blocks of flats ; buildings intended solely, principally, or subsidiarily for habitation). In addition to these systems of classification a distinction is also sometimes made between the dwellings according as they have certain subsidiary rooms and domestic equipment, or according to rent-group, while the buildings may be classified according to the material used for construction, the number of stories, of dwellings, or of rooms. Finally, these various data are sometimes completed by — 108 — information on the total value of the buildings considered, the cost of construction being a useful measure of their importance. In the absence of other statistics, these figures may also be used to estimate fluctuations in the total housing supply, provided that they are suitably corrected to eliminate the influence of fluctuations in prices. * * These are the conclusions to which a study of the different housing statistics at present published in the principal countries has led. They have been formulated on the basis of the most recent tendencies observed both in the particular field of housing statistics and in that of statistics in general. Without claiming to constitute a final programme, it is believed that the principles adopted here may provide a first foundation for standardising (both nationally and internationally) the different sets of statistics. APPENDIX List of Official Statistical Reports relating to Housing Used in the Present Study l I. — GENERAL CENSUSES. AUSTRALIA Census of the Commonwealth of Australia, taken for the night between the 2nd and 3rd April 1911. Vols. XVIII to XXV. Ministry of State for Town and Territories, Bureau of Census and Statistics. Census of the Commonwealth of Australia, taken for the night between the 3rd and 4th April 1921. Vols. XVIII to XXV. Ministry of State for Town and Territories, Bureau of Census and Statistics. AUSTRIA Die Ergebnisse der Volkszählung vom 31. Dezember 1910. Neue Folge, österreichische Statistik, 4. Band, 1. u. 2. Heft. Herausgegeben von der K. K. Statistischen Zentralkommission. Ergebnisse der ausserordentlichen Volkszählung vom 31. Januar 1920. Beiträge zur Statistik der Republik Oesterreich, 6. Heft. Herausgegeben von der Statistischen Zentralkommission. Vorläufige Ergebnisse der Volkszählung vom 7. März 1923. Beiträge zur Statistik der Republik Oesterreich, 12. Heft. Herausgegeben vom Bundesamte für Statistik. Vienna Die Ergebnisse der Wohnungszählung in Wien vom Jahre 1916. Herausgegeben vom Wohnungsamte der Stadt Wien. BELGIUM Recensement général du 31 décembre 1920. Recensement des logements dans les communes de 10.000 habitants et plus. Ministère de l'Intérieur et de l'Hygiène, Service de la statistique générale. Brussels Recensement général de la population ; recensement spécial des logements (Bruxelles). Les recensements de 1910. Administration communale de Bruxelles, Service de la population. BULGARIA Recensement des bâtiments et des logements dans le Royaume de Bulgarie au 31 décembre 1920. Royaume de Bulgarie, Direction générale de la statistique. 1 This list comprises the chief national reports published since 19091910, and the. reports published since that date by selected important municipalities in the different countries. — 110 — CANADA Fifth Census of Canada, 1911. Ministry of Trade and Commerce, Census and Statistics Office. CZECHOSLOVAKIA Prague Soupis b y t u v Praze, podle stavu z 15. unora 1921. Statisticka k o mise hlavniho mesta Prahy. (Statistics of Dwellings in Prague at the Census of 15 F e b r u a r y 1921. Statistical Commission of Prague.) DENMARK Boligkommissionen af 1916, Betsenkning. of 1916, Report.) Boligkommissionen af 1918, Betsenkning. of 1918, Report.) (The Housing Commission (The Housing Commission, Copenhagen Statistisk Aarbog for Kobenhavn, Frederiksberg og Gjentofte Kommune, 1924. Udgivet af Kobenhavns Statistiske Kontor. (Statistical Yearbook for Copenhagen, Frederiksberg and Gjentofte, 1924. Published by the Copenhagen Statistical Office.) FINLAND Asuntolaskenta huhtikuun 25 p : nä 1919. Suomen "Virallinen Tilasto X X X I I . Sosialisia erikoistutkimuksia II, julkaissut Sosialihallitus. (Housing Census of 25 April 1919. Official Statistics of Housing Census of Finland. Special Enquiries into Social Questions. Published by the General Department of the Ministry of Social Affairs.) FRANCE Resultats statistiques du recensement général de la population, effectué le 24 mars 1900. Tome V, Habitation. Ministère du Travail et de la Prévoyance sociale, Service du recensement. Statistique des familles et des habitations en 1911. Ministère du Travail et de la Prévoyance sociale ; Statistique générale de la France. GERMANY Die Volkszählung im Deutschen Reiche am 1. Dezember 1910. Statistik des Deutschen Reichs, Band 240. Reichswohnungszählung im Mai 1918. Statistik des Deutschen Reichs, B a n d 287, I. u. II. Die Kriegs-Wohnungszählung vom 16. Mai 1918 in Bayern. Heft 91 der Beiträge zur Statistik Bayerns. Herausgegeben vom Statistischen Landesahit. Statistisches Jahrbuch deutscher Städte. Herausgegeben von Prof. Dr. M. Neefe, Breslau 1916. Berlin Die Grundstücks-Auf nähme vom 15. Oktober 1910 sowie die W o h nungs- und die Bevölkerungs-Aufnahme vom 1. Dezember 1910 in der Stadt Berlin und 44 Nachbargemeinden. Herausgegeben vom Statistischen A m t der Stadt Berlin. Die Grundstücks- und Wohnungsaufnahme sowie die "Volks-, Berufs-, und Betriebszählung in Berlin im J a h r e 1925. Mitteilungen des Statistischen Amts der Stadt Berlin, Heft 1, Nr. 5. Essen Die Wohnungszählung in Essen vom 8. Oktober 1919. Statistisches A m t der Stadt Essen. Nr. X I , 22-20. — Ill — Frankfortron-Main Statistische Jahresübersichten der Stadt Frankfurt am Main, 1913 bis 1914. Herausgegeben vom Statistischen Amt. Nuremberg Nürnberger Wohnungszählung vom 16. April 1925. Durchgeführt und bearbeitet vom Statistischen Amt der Städte. GREAT BRITAIN Census of England and Wales, 1911. Vol. VI, Buildings of Various Kinds. Cd. 6577. U.K. Ministry of Health, Registrar-General for England and Wales, Census Office. Report on the Twelfth Decennial Census of Scotland, 1911. Vol. II. Cd. 6896. Census of Ireland, 1911. General Report with tables and appendices. Cd. 6663. Census of England and Wales, 1921. General tables. U.K. Ministry of Health, Registrar-General for England and Wales, Census Office. Report on the Thirteenth Decennial Census of Scotland, 1921. Vol. I I . HUNGARY Recensement général de la population. Publications statistiques hongroises, nouvelle série. Vol. 69. Publié par l'Office central de statistique du Royaume de Hongrie. Die Bevölkerung der Städte und Gemeinden mit mehr als 20.000 Einwohnern nach der Volkszählung im J a h r e 1920. Magvar Statisztikai Szemle, 1923, Nos. 9-12. (Hungarian Statistical Review, 1923, Nos. 9-12.) Budapest Népszàmlàlàs, 1920. Budapest Székesfováros Statisztikai H i v a t a lának Kiadványai, No. 52. (Volkszählung, 1920. Arbeiten des Statistischen Amtes der H a u p t und Residenzstadt Budapest, Nr. 52.) INDIA Census of India, 1911. Vols. I and II. Indian Government. Census of India, 1921. Vols. I and II. Indian Government. ITALY Censimento della popolazione del Regno d'Italia, al 10 giugno 1911. Ministero di Agricultura, Industria e Commercio, Direzione generale della Statistica e del Lavoro. Ufficio del Censimento. Annuario Statistico delle Città italiane, anno VI, 1915-1916. Il Comune di Firenze e la sua popolazione al 1 dicembre 1921. Comune di Firenze, monografie e studi dell' Ufficio di Statistica, No. 7. LATVIA Riga Die Wohnungen in Riga. Veröffentlichung des Statistischen Amtes der Stadt Riga, Neue Folge, Nr. 3. NETHERLANDS Uitkomsten der negende tienjaarlijksche Volkstelling in het Koninkrijk der Nederlanden, december 1909 ; I en I I I Deel. Bijdragen t o t de Statistiek van Nederland. Uitgegeven door het Centrasi Bureau voor de Statistiek. (Results of the Ninth Census of the Population of the Kingdom of t h e Netherlands, 31 December 1909. Parts I and I I . Published by the Central Statistical Office.) — 112 — Volkstelling, 31 december 1920. Statistiek van Nederland, No. 342, 352, 378. Uitgegeven door het Centraal Bureau voor de Statistiek. (Census of 31 December 1920. Netherlands Statistics, No. 342, 352 and 378. Published by the Central Statistical Office.) Amsterdam Verslag van de Woningtelling gehouden in m a a r t 1909, Gemeentelijke Woningdienst te Amsterdam. (Report of t h e Housing Census of March 1909. Amsterdam Municipal Housing Department.) Verslag van de Woningtelling gehouden in de maand October 1919. Gemeentelijke Woningdienst te Amsterdam. (Report of t h e Housing Census of October 1919. Amsterdam Municipal Housing Department.) The Hague Verslag van de Resultaten der Woningtelling gehouden tusschen 6 en 25 October 1919. Verzameling 1910, 's Gravenhage. (Report on the Results of the Housing Census of 6-25 October 1919, a t The Hague.) NORWAY Folketellingen i Norge, 1. December 1910 ; tredje hefte, bebodde hus og husholdinger. Norges Officielle Statistik, V 188 ; u t g i t av det Statistiske Centralbyraa. (Census of 1 December 1910 ; Inhabited Houses and Households.) Folketellingen i Norge 1920 ; Boligstatistikk, By er og Bygder. Norges Officielle Statistikk, V I I 98 og 144. (Census of 1 December 1920 ; Statistics of Dwellings, Towns, and Rural Districts.) Statistisk aarbok for Kristiania by. Utgit av Kommunens Statistiske Kontor. (Statistical Yearbook for the Municipality of Christiania. Published by the Municipal Statistical Office.) POLAND Le premier recensement général de la République polonaise du 30 septembre 1921. Logements, population, professions (volumes par départements). Statistique de la Pologne, tomes X I V à X X . Publiée par l'Office central de statistique de la République polonaise. Warsaw Resultaty Spisu nieruchomosci i mieszkan wielkiej Warszawy 1919. Prace Wydzialu Statystycznego M. St. Warszawy. (Results of t h e Census of Buildings and Dwellings in Warsaw in 1919. Publications of the Warsaw Statistical Office.) SWEDEN 1912-1914 árs allmänna bostadsräkningar ; Del I, Allmän redogörelse. Sveriges Officiella Statistik, Socialstatistik ; av K. Socialstyrelsen. (General Housing Censuses, 1912-1914 ; P a r t I, General Report. Swedish Official Statistics, Social Statistics. Published by t h e Social Board.) Allmänna bostadsräkningen ár 1920. Sveriges Officiella Statistik, Socialstatistik ; av K. Socialstyrelsen. (General Housing Census, 1920. Swedish Official Statistics, Social Statistics. Published by the Social Board.) Stockholm Allmänna bostadsräkningen i Stockholm, den 31 december 1905. Stockholms Stads Statistik, V, Byggnader och Bostäder. Utförd av Stockholms Stads Statistiska Kontor. (General Census of Dwellings in Stockholm, 31 December 1905. Statistics of the Municipality of Stockholm ; V, Buildings and Dwellings. Stockholm Municipal Statistical Office.) Allmänna bostadsräkningen i Stockholm, den 31 december 1910. Stockholms Stads Statistik, V, Byggnader och Bostäder. Utförd av — 113 — Stockholms Stads Statistiska Kontor. (General Census of Dwellings in Stockholm, 31 December 1910. Statistics of the Municipality of Stockholm ; V, Buildings and Dwellings. Stockholm Municipal Statistical Office.) Allmänna bostadsräkningen i Stockholm, den 31 december 1915— 1 januari 1916. Stockholms Stads Statistik, V, Byggnader och Bostäder. Utförd av Stockholms Stads Statistiska Kontor. (General Census of Dwellings in Stockholm, 31 December 1915—1 January 1916. Statistics of the Municipality of Stockholm ; V, Buildings and Dwellings. Stockholm Municipal Statistical Office.) Allmänna bostadsräkningen i Stockholm, den 31 december 1920— 1 januari 1921. Stockholms Stads Statistik, V, Byggnader och Bostäder. Utförd av Stockholms Stads Statistiska Kontor. (General Census of Dwellings in Stockholm, 31 December 1920—1 January 1921. Statistics of the Municipality of Stockholm ; V, Buildings and Dwellings. Stockholm Municipal Statistical Office.) SWITZERLAND Die Ergebnisse der Eidgenössischen Volkszählung vom 1. Dezember 1910. Vol. I. Eidg. Finanz- u. Zolldepartement, Statistisches Bureau. Hauptergebnisse der Eidgenössischen Volkszählung vom 1. Dezember 1920 ; Wohnungsbevölkerung und ortsanwesende Bevölkerung nach Gemeinden. Eidg. Finanz- und Zolldepartement, Statistisches Bureau. Aarau Die Ergebnisse der Wohnungszählung vom 1. Dezember 1910, in den Gemeinden Aarau, Baden, Ennetbaden und Brugg. Aargauische Statistische Mitteilungen, Neue Folge, Heft III. Basle Die Wohnungen des Kantons Basel-Stadt am 1. Dezember 1910. Mitteilungen des Statistischen Amtes des Kantons Basel-Stadt. Die vorläufigen Ergebnisse der Wohnungszählung vom 1. Dezember 1920 im Kanton Basel-Stadt. Mitteilungen des Statistischen Amtes des Kantons Basel-Stadt. Berne Tabellarische Uebersichten betreffend die Wohnungszählung in Bern im Monat Mai 1916. Beiträge zur Statistik der Stadt Bern. Herausgegeben vom Städtischen Statistiker. Die Wohnverhältnisse in der Stadt Bern, nach den Ergebnissen der Wohnungszählung vom 1. Dezember 1920. Beiträge zur Statistik der Stadt Bern, Heft 6 ; herausgegeben vom Statistischen Amt im Auftrage der Direktion der Industriellen Betriebe. Saint Gall Die Ergebnisse der Wohnungszählung vom 1. Dezember 1910, in den Gemeinden St. Gallen, Straubenzell und Tablât. Statistik des Kantons St. Gallen ; Heft XXXIX. Zurich Die Wohnungen in der Stadt Zürich am 1. Dezember 1910. Statistik der Stadt Zürich. Herausgegeben vom Statistischen Amte der Stadt Zürich. UNITED STATES Thirteenth Census of the United States, taken in the Year 1910. Vol. I. U.S.A. Department of Commerce, Bureau of the Census. Fourteenth Census of the United States, taken in the Year 1920. Vols. I and II. U.S.A. Department of Commerce, Bureau of the Census. Abstract of the Fourteenth Census of the United States, 1920. U.S.A. Department of Commerce, Bureau of the Census. — 114 — UNION OF SOUTH AFRICA Census of t h e Union of South Africa, 1911. Annex to General Report ; P a r t I, Population and Dwellings. D e p a r t m e n t of Interior, Office of Census. Census of the European or White Races of t h e Union of South Africa, 1918 ; P a r t V I I I , Dwellings. Department of Interior, Office of Census. Third Census of t h e Population of t h e Union of South Africa, enumerated 3 May 1921 ; P a r t I X , Dwellings. D e p a r t m e n t of Interior, Office of Census. II. — SPECIAL ENQUIRIES DENMARK Husleje og Boligforhold, November 1922. Danmarks Statistik, Statistiske Meddeleser, 4 Raekke, 68 Bind, 1 Hefte. (Rents and Housing in November 1922.) FINLAND Tehdastyöväen asunto-olot. Sosialinen aikakauskirja, 1926, No. 3, 4, 5, 8. Sosialimimsteriön julkaisema. (Housing Conditions of Industrial Workers ; Social Review, 1926, Nos. 3, 4, 5 and 8. Published by t h e Ministry of Social Affairs.) GERMANY Dresden Die Lage des Dresdner Wohnungsmarktes im J a h r e 1907. Mitteilungen des Statistischen Amtes der Stadt Dresden ; Heft 17. Frankfort-on-Main Die Besiedlung des Frankfurter Stadtgebiets und die Befriedigung des Wohnungsbedürfnisses der Bevölkerung. Beiträge zur Statistik der S t a d t Frankfurt am Main, 11. Heft. Bearbeitet vom Direktor des Statistischen Amtes, Dr. A. Basch. Munich Zählung der leerstehenden Wohnungen in München am 7. November 1909. Mitteilungen des Statistischen Amtes der Stadt München ; Band X X I I , Heft I. Wiesbaden Wiesbadens Wohnungs- und Grundstücksmarkt. Wiesbadener Statistik, Heft 6. Herausgegeben vom Städtischen Statistischen Amte. GREAT BRITAIN London London Statistics. London County Council. Housing Statistics for the year 1923-1924. London County Council. Estates, ITALY Florence Le Abitazioni e i Redditi ; il mercato delle abitazioni nel 1914 ; l'Attività edilizia. Monografie e studi dell' Ufficio di Statistica, No. 5, Comune di Firenze. — 115 — NETHEBLANDS Amsterdam H e t technisch Woningonderzoek en de systematische perceelsbeschrijving van Amsterdam. Uitgave van den gemeentelijken woningdienst Amsterdam. (Technical Housing Enquiry and Systematic Description of Sites of Amsterdam. Published by the Municipal Housing Department.) Verslag van h e t onderzoek n a a r de woningtoestanden in het z. g. Amstellaankwartier. Gemeentelijke woningdienst, Amsterdam. (Report on the Enquiry into Housing Conditions in the so-called " Amstellaan " Quarter. Amsterdam Municipal Housing Department.) NORWAY Beboelsesforholdene i smaaleiligheter i Kristiania 1913-1914. Specialundersökelser II, utgit av Kommunens statistiske kontor. (Housing Conditions in Small Dwellings in Christiania, 1913-1914. Special E n quiries II. Published by the Municipal Statistical Office). SWEDEN Undersökningar rörande de mindre bemedlades bostadsförhallanden i vissa svenska städer, S venges Officiella Statistisk, Socialstatitik ; a v K. Socialstyrelsen. (Enquiries into Housing Conditions of Persons with Small Means in Certain Swedish Towns. Swedish Official Statistics, Social Statistics. Published by the Social Board.) Hyresräkningen âr 1924 ; Del I, Hyresforhàllandena m.m. ; Del I I , Boendeforhâllandena. Sveriges Officiella Statistik, Socialstatistik ; av. K. Socialstyrelsen. (Rent Census, 1924; P a r t I, R e n t Conditions, etc. ; P a r t II, Housing Conditions. Swedish Officiai Statistics, Social Statistics. Published by the Social Board.) Stockholm Bostadsförhallanden for stadens funktionärer àr 1919. Stockholms Stads Statistik, X, Specialundersökningar ; utförd av Stockholms Stads Statistiska Kontor. (Housing Conditions of Municipal Officials 1919. Statistics of the Municipality of Stockholm; X , Special Enquiries. Municipal Statistical Office.) SWITZERLAND Die Ergebnisse der Mietpreiserhebung von J a n u a r 1924. Sozialstatistische Mitteilungen, II. Jahrgang, 1924, Heft 2. Herausgegeben vom Eidg. Arbeitsamt. Basle Die Mietpreise in der Stadt Basel 1910-1920. Mitteilungen des Statistischen Amtes des Kantons Basel-Stadt. Berne Die Erhebungen über den Berner Wohnungsmarkt, im J a h r e 1917. Beiträge zur Statistik der Stadt Bern, Heft 2. Herausgegeben vom Städtischen Statistiker. Wohnungsvorrat und Wohnungsbedarf in der Stadt Bern, Anfang 1926. Beiträge zur Statistik der "Stadt Bern, Heft 7. Herausgegeben vom Statistischen Amt. Die Lage des Wohnungsmarktes in der Stadt Bern im J a h r e 1918. Beiträge zur Statistik der Stadt Bern, Heft 3. Herausgegeben vom Statistischen A m t der Stadt Bern. — 116 — Zurich Die Erhebungen über den Zürcher Wohnungsmarkt. Statistik der Stadt Zürich, Nr. 15. Herausgegeben vom Statistischen Amt der Stadt Zürich. Der Wohnungsmarkt in Zürich und Umgebung am 1. Dezember 1913. Statistik der Stadt Zürich, Nr. 17. Herausgegeben vom Statistischen Amt der Stadt Zürich. Der kommunale und subventionierte Wohnungsbau in der Stadt Zürich bis zum Jahre 1920. Statistik der Stadt Zürich, Heft 27. Herausgegeben vom Statistischen Amte der Stadt Zürich. III. — PERIODICAL RETURNS. AUSTRIA Statistisches Jahrbuch der Stadt Wien. Bearbeitet von der MagistratsAbteilung XXI für Statistik. Wochenbericht der Statistischen Abteilung. Magistrat der Stadt Wien. Statistische Mitteilung der Stadt Wien. Herausgegeben von der Magistrats-Abteilung für Statistik. CANADA The Labour Gazette. Published by the Department of Labour, Canada. CZECHOSLOVAKIA Prague Statistická zpráva hlavniho mesta Prahy. Statisticka komise hlavniho mesta Prahy. (Statistical Yearbook for Statistical Commission, Prague.) Mesícní zpravy statisticke kancelare hlavniho mesta Prahy. (Monthly Bulletin, Prague Municipal Statistical Office.) DENMARK Statistisk Aarbog. Udgivet af det Statistiske Departement. (Statistical Yearbook. Published by the Statistical Department.) Huslejen og Byerne. Danmarks Statistik, Statistiske Meddelelser. Udgivet af det Statistiske Department. (Rents in Towns. Statistical Review. Published by the Statistical Department.) Copenhagen Statistisk Aarbog for Kobenhavn, Frederisberg og Gjentofte Kommune. Udgivet af Kobenhavns Statistiske Kontor. (Statistical Yearbook for Copenhagen, Frederiksberg and Gjentofte. Published by the Copenhagen Statistical Office.) FRANCE Annuaire statistique. Statistique générale de la France ; Ministère du Travail, de l'Hygiène, de l'Assistance et de la Prévoyance sociale. Lyou Bulletin municipal officiel. Municipalité de Lyon. 1 — 117 — GERMANY Bautätigkeit und leerstehende Wohnungen in deutschen Städten ; Sonderbeilage zum Reichsarbeitsblatt. Verarbeitet im Statistischen Amte. Viertel] ahrshefte zur Statistik des Deutschen Reichs. Herausgegeben vom Statistischen Reichsamt. Berlin Statistisches Jahrbuch der Stadt Berlin. Herausgegeben von dem Statistischen Amt der Stadt Berlin. Bautätigkeit und Wohnungsmarkt in Berlin ; Mitteilungen des Statistischen Amts der Stadt Berlin. Vierteljahrsberichte des Statistischen Amts der Stadt Berlin. Hamburg Statistisches Handbuch für den Hamburgischen Staat. Herausgegeben vom Statistischen Landesamt. Hamburger Statistische Monatsberichte. Herausgegeben vom Statistischen Landesamt. Der Wohnungsmarkt in der Stadt Hamburg... Sonderabdruck aus dem Amtlichen Anzeiger. Die Miete der Wohnungen und Geschäftsräume in der Stadt Hamburg... Sonderabdruck aus dem Amtlichen Anzeiger. Endgültige Ergebnisse der Personenstands- und Wohnungsaufnahme vom... im hamburgischen Staate. Sonderabdruck aus dem Amtlichen Anzeiger. Munich Münchener Jahresübersichten. Statistisches Amt der Stadt München. GREAT BRITAIN The Ministry of Labour Gazette. HUNGARY Budapest Budapest Székesfovàros Statisztikai Êvkônyve. Budapest Székesfovàros Statisztikai Hivatala. (Statistisches Jahrbuch der Haupt- und Residenzstadt Budapest. Budapest, Kommunal-statistisches Bureau.) ITALY Annuario statistico. Commune di Milano, Ufficio municipale di Statistica. (Statistical Yearbook. Municipality of Milan. Municipal Statistical Office.) Rome Rivista mensile di Statistica. Commune di Roma, Ufficio di Stato Civile e Statistica. (Monthly Statistical Review. Municipality of Rome, Statistical Office.) JAPAN Jiji Yearbook. NETHERLANDS Maandschrift van net Centraal Bureau voor de Statistiek. (Monthly Review of the Central Statistical Office.) — 118 — Amsterdam Statistisch Jaarboek der Gemeente Amsterdam. Uuitgegeven door het Bureau van Statistiek der Gemeente Amsterdam. (Statistical Yearbook for the Municipality of -Amsterdam. Published by the Municipal Statistical Office.) Maandbericht van het Bureau van Statistiek der Gemeente Amsterdam. (Monthly Bulletin of the Amsterdam Municipal Statistical Office.) Verslag van den Gemeentelijken Woningdienst te Amsterdam. Verslagen van Bedrijven, Diensten en Commissiën der Gemeente Amsterdam. (Report of the Amsterdam Municipal Housing Department.) The Hague Jaarboek. Statistisch Bureau der Gemeente's Gravenhage. (Statistical Office of the Municipality of The Hague Yearbook.) Maandcijfers van het Statistisch Bureau der. Gemeente 's Gravenhage. (Monthly figures of The Hague. Municipal Statistical Office.) NEW ZEALAND Monthly Abstract of Statistics. Dominion of New Zealand. NORWAY Statistiske meddelelser. Utgit av det Statistiske Centralbyra. (Monthly Review of the Central Statistical Office.) Oslo Statistisk aarbok'for Kristiania by. Utgit av Kommunens statistiske kontor. (Statistical Yearbook for the Municipality of Christiania. Municipal Statistical Office.) Statistisk Maanedsskrift. Utgit av Kristiania Kommunale statistiske kontor og folkeregister. (Statistical Monthly. Published by the Municipal Statistical Office.) POLAND Warsaw Miesiecznik Statystyczny. Majistrat M. St. Warszawy. (Monthly Bulletin of Municipal Statistics. Municipality of Warsaw.) Rocznik Statystyczny Warszawy. Magistrat M. St. Warszawy. (Statistical Yearbook for Warsaw. Municipality of Warsaw.) Lodz Annuaire statistique de la ville de Lodz. Service de statistique de la ville de Lodz. SWEDEN Sodala Meddelanden. Utgivna av K. Socialstyrelsen. (Social Review. Published by the Social Board.) Byggnadsverksamheten i rikets stadssamhällen Sveriges Officiella Statistisk. Socialstatistisk ; av. K. Socialstyrelsen. (Building activity in Swedish U r b a n Communes. Swedish ' Official Statistics, Social Statistics. Published by the Social Board.) Stockholm Statistisk ársbok för Stockholms Stad ; Stockholms stads Statistiska Kontor. (Yearbook for the Municipality of Stockholm. Municipal Statistical Office, Stockholm.) Statistisk mánadsskrift, utgiven av Stockholms Stads Statistiska Kontor. (Monthly Bulletin of the Municipal Statistical Office, Stockholm.) — 119 — SWITZERLAND Informations de statistique sociale. Publiées par l'Office fédéral du travail. Basle Statistisches J a h r b u c h des Kantons Basel-Stadt. Herausgegeben vom Statistischen Amt. Bautätigkeit und Wohnungsmarkt. Mitteilungen des Statistischen Amtes des Kantons Basel-Stadt. Die Zählung der leerstehenden Wohnungen und Geschäftslokale in Basel. Mitteilungen des Statistischen Amtes des Kantons Basel-Stadt. Statistische Viertel]ahrs-Berichte des Kantons Basel-Stadt. Herausgegeben vom Statistischen Amt. Statistische Monatsberichte des Kantons Basels-Stadt. Herausgegeben vom Statistischen Amt. Berne Statistisches Handbuch der Stadt Bern. Herausgegeben vom Statistischen Amt. Halb Jahresbericht über die Bevölkerungsbewegung und die wirtschaftlichen Verhältnisse der Stadt Bern. Herausgegeben vom Statistischen Amt, im Auftrage der Direktion des Industriellen Betriebe. Zurich Zürcher Statistische Nachrichten. Herausgegeben vom Statistischen A m t der Stadt Zürich. Die Mietpreise in der Stadt Zürich. Statistik der Stadt Zürich. Herausgegeben vom Statistischen Amt der Stadt Zürich. UNITED STATES Building Permits in the Principal Cities of the United States. Bulletin of the United States Bureau of Labour Statistics, Miscellaneous Series. U.S. Department of Labour. Bureau of Labour Statistics. Monthly Labour Review. U.S. Department of Labour, Bureau of Labour Statistics. UNION OF SOUTH AFRICA Official Yearbook. Union of South Africa.