I* I • A*«orv&rvr» INTERNATIONAL LABOUR OFFICE STUDIES AND REPORTS Series F, Second Section (Safety), No. 1 AUTOMATIC COUPLINGS AND THE SAFETY OF RAILWAY WORKERS Report on Statistics of Accidents due to Coupling and Uncoupling Operations GENEVA 192 Í PRINTED BY ALBERT KUNDIG GENEVA CONTENTS Introduction I. II. III. IV. V. VI. 3 The General Problem 8 Statistical Methods 13 Coupling and Uncoupling Accidents in Countries where Automatic Couplings are in Use (United States and Canada) 23 Coupling and Uncoupling Accidents in Countries where Couplings are Non-Automatic 30 Comparison of Coupling and Uncoupling Accident Risk in America and Europe 37 The Occupational Risk of Shunting Operators 41 Conclusion 45 APPENDIX Preliminary Note 47 Tables I and II. United States 48 Table III. Canada 50 Table IV. United Kingdom 51 Table V. India 53 Table VI. Belgium 54 Table VII. France 55 Table VIII. Switzerland 56 Table IX. Germany, Luxemburg, Netherlands, Roumania, AustriaHungary .57 Table X. Germany 59 Table XI. Sweden 60 Table XII. Norway 61 Table XIII. Finland 62 i INTRODUCTION The Fifth Session of the International Labour Conference on 29 October 1923 adopted the following resolution on the proposal of the Swiss workers' delegate, Mr. Charles Schurch: The Fifth Session of the International Labour Conference, considering the importance for the safety of railway workers of questions relating to automatic couplings, requests the Governing Body to obtain information from Governments and from international technical and industrial organisations upon the question of automatic couplings, in order to decide whether an international agreement in the matter is desirable in the interests of the workers. In compliance with this resolution, the Governing Body instructed the International Labour Office to get into touch with international organisations competent to deal with the general question of automatic couplings, and to study the safety aspect of the problem. With a view to its examination in co-operation with the international bodies most concerned, the International Labour Office deemed it desirable to undertake a preliminary study in order to ascertain, if possible, how far the need for automatic couplings is supported by the facts. As arguments in favour of the introduction of automatic couplings on rolling stock are based principally on the endeavour to reduce the number of accidents due to coupling and uncoupling operations, it is first of all necessary to ascertain how far this aim has actually been achieved in countries where automatic couplings have been introduced, and how great is the risk of coupling and uncoupling accidents in countries where they are not in use. It is the object of the present study to furnish an answer to these two questions, which have given rise to wide divergences of opinion. The chief cause of these divergencies has no doubt been the lack of definite information. It is manifest that the questions in debate can only be answered by means of comparative statistics of accidents to railway employees classified by cause. As was - 6 - pointed out at the Seventh Session of the International Railway Congress held in Washington in 1905, where the question of automatic couplings was dealt with also from the point of view of safety, "this is not a question of opinion but above all a question of statistics, and it would be desirable to obtain a little more precise information on this aspect of the question." 1 Although statistical data have later been produced, both incidentally in the course of technical discussions, and regularly in the official railway statistics of a number of countries, these data have never been systematically collected and analysed. Thus the situation is still the same as it was twenty years ago, and, as a technical railway paper recently put it, "the fact is that, generally, there is a sad lack of information as to the need for automatic couplings."2 The question being at once technical and one of international statistics, the study of it involves certain difficulties. First, the question of the risk of coupling and uncoupling accidents is complicated in that it is determined by technical conditions and connected with certain other risks inevitable in the railway service. It involves, for example, questions of brakes, heating pipes, lighting, etc. It has been argued against the assumed advantages of automatic couplings that their introduction would not remove the necessity of men passing between vehicles as long as the connecting of brake and steam-heating pipes is not automatic,8 quite apart from the fact that it would still be necessary to carry the coupling by hand and to adjust its mechanism. These technical factors, however, are outside the scope of a purely statistical report such as this and are not considered here. The second difficulty in the preparation of the present report lies in the incompleteness of the available statistics of railway accidents. The present report is based on the printed official statistics in most cases; additional unpublished information has been secured for only three countries 4. It must be pointed out that the official statistics are very meagre and differ in scope and the statistical methods employed; hence comparisons become more 1 Bulletin du Congrès international des chemins de fer, Vol. XX, No. 7, July 1906, p. 760. Brussels. 2 The Railway Gazette, Vol. XL, No. 12, 21 Mar. 1924, p. 403. London. 3 On the other hand, it should be noted that lately trains are more and more equipped with vacuum continuous brakes, and have increased in length and weight; these factors require greater strength of coupling,which would not be possible without automatic couplings. 4 See Appendix, notes to tables VII, VIII, and XIII. - 7 - or less hazardous, as in any international statistical study at present.1 Statistics have been collected for fifteen countries in all: AustriaHungary (the old Empire), Belgium, Canada, Finland, France, Germany, India, Luxemburg, the Netherlands, Norway, Roumania, Sweden, Switzerland, the United Kingdom, and the United States.2 In all cases it was endeavoured to collect statistics for a fairly long period, in most cases ten years, but for some countries it was necessary to be content with shorter periods. The principal data obtained are presented in as uniform a manner as possible in the appended tables (I to XIII), to which are attached notes indicating the scope of each series, and the statistical methods employed in compiling them. A comparative survey based on the tables is given in the following chapters. First, a general survey of the safety aspect of the question of automatic couplings is necessary in order to define the limits of the present report, as well as to give an account of the different opinions prevailing on the subject. Secondly, the scope and method of statistics of accidents to railway employees are compared. Thirdly, the results of the statistics are given separately for countries where automatic couplings have been introduced and for those where the couplings are non-automatic, while a special section is devoted to the occupational hazard of shunting operators. In conclusion, an answer to the questions put forward is attempted, so far as the information secured permits. Although the International Labour Office is well aware of the limitations and defects of the material available, and can therefore only consider this report as provisional, yet it is hoped that it will prove of interest to those dealing with problems of the safety of railway workers. 1 Comprehensive and fully comparable statistics of this kind could only be obtained by means of special questionnaires to the different railway departments, a procedure that would require considerable time. 2 It is to be regretted that the railway statistics of many countries, e.g. Italy and Spain, do not give sufficient details of railway accidents, and that certain statistics, e. g. those of Germany, have in recent years been curtailed to the detriment of their value. I The General Problem The subject of automatic couplings is not only a question of the safety of railwaymen, but also one of the technique of railway operation and finance. The advocates of a reform in the prevailing coupling system refer not only to the degree of accident risk in coupling and uncoupling operations, but also to the improvement of railway service likely to result from, and to the increasing weight of trains necessitating, the introduction of automatic couplings. The opponents of these appliances, on the other hand, fear that a considerable dislocation of traffic and inconveniences in the exchange of rolling stock would result from a change in the coupling system, and point to the expense involved in the introduction of automatic couplings, which they describe as enormous. The present paper deals solely with the safety aspect of the question, and it is generally admitted that this is its dominating aspect 1 . As a matter of fact the safety of thousands of railwaymen in every country is involved in coupling and uncoupling operations. The non-automatic system of coupling exposes the men carrying out these operations to a certain risk of accidents, particularly in goods stations and sorting yards. It is true that railway managements are alive to this risk, and have framed rules with a view to obviating it, but it can hardly be denied that such rules must be actually broken every day; indeed, a strict observance of them would practically amount to sabotage.2 Thus, shunting operators are often obliged to pass between vehicles without waiting Cf1 . Bulletin du Congrès international des chemins de fer, Vol. XX, No. 1, July 1906; Proceedings of the Seventh Session of the International Railway Congress in Washington, 1905. 2 See discussion on this and other points of safety in a series of articles (by the Editor, Mr. W. M. Acworth, and others), under the title "Lives versus Profits", in the New Statesman, Moi. I, 1913, pp. 231-232, 302-303, 335-336, 367-368, and 400. London. - 9 - for them to come to a stop, and once the coupling or uncoupling has been effected, they frequently have to pass under the buffers and between the wheels when the vehicles are again in motion. No doubt practices of this kind give rise to accidents, and it is very generally assumed — though nevertheless contested — that the accident risk in coupling and uncoupling, as well as the occupational risk of shunting operators, is relatively high. It is also generally assumed that by means of automatic couplings the safety of the railwaymen concerned would be considerably increased, if not completely secured. iNTROnUCTION OF AUTOMATIC COUPLINGS It was in the interests of safety that the question of automatic couplings was first raised in the United States. In this country, as will be shown later, the original system of coupling was extremely dangerous, and the accidents arising therefrom very frequent and severe. This attracted the attention of American engineers, who, as early as 1867, formed an association, called Master Car Builders, to study the question of automatic couplings. These studies led later to a general campaign in favour of a Federal Act prescribing the compulsory introduction of such appliances. An Act to this effect was actually passed on 2 March 1893, amended on 1 April 1896, and a supplementary Act was adopted on 2 March 1903 ( "Safety Appliances Acts"). These Acts have rendered automatic couplings compulsory on all railway lines of the United States, for most of them from 1 January 1900, and for all from 1 September 1903. The adoption of the American Acts gave ¡ise to discussions and experiments in most countries of the world. Important experiments have been carried out especially in France, as well as in Great Britain, Germany, Italy, Austria, Denmark, Sweden, and South Africa, but these have led to few, if any, practical results. At present the situation is as follows: automatic couplings are in general use in the United States, Canada, Mexico, and the Province of Natal (South Africa); they have been introduced on all engines and vehicles on the metre and narrower-gauge railways in India 1, 1 The number of engines fitted with automatic couplings in 1923 was 3,344 and of coaching and goods vehicles 65,086, while the number of engines not, so fitted was 6,396, and that of vehicles 145,447. (Communication from the India Office, London, to the International Labour Office.) - 10 -- and on a few narrow-gauge railways in Germany, France, and Great Britain. OPINIONS ON THE EFFECT OF AUTOMATIC COUPLINGS In the meantime, international bodies of technicians, persons interested in the protection of labour, and railwaymen have discussed the problem in their conferences and put forward various resolutions in favour of the adoption of automatic couplings. Some of these may be briefly recalled. The seventh session of the International Railway Congress — an association of railway boards and administrations in most countries of the world — at Washington in 1905 discussed the question of automatic couplings. It pointed out that the use of automatic couplings was established in the United States, Canada, and Mexico, but that in other countries where experiments had been made divergent opinions had been expressed on the different appliances tried; the English representatives declared that the system employed in Great Britain and Ireland gave satisfaction "both with regard to the rapidity of the service and the safety of the staff". The fifth General Assembly of the International Association for Labour Legislation, held in 1910 at Lugano, also considered this problem and invited its national sections to present to their respective governments petitions in favour of the introduction of automatic couplings on rolling stock. The international trade-union federations have for twenty years been alive to the dangers incurred by railwaymen through the absence of a system of coupling which would obviate the necessity of passing between vehicles. The International Transport Workers' Congress held at Milan in 1905 passed a resolution asking for the introduction of automatic couplings upon the rolling stock of all countries. The question has been discussed with especial frequency in international conferences since the war. It will be recalled that the programme adopted by the International TradeUnion Conference at Berne in 1919, which included the demands of the workers to be submitted to the Peace Conference, asked that a common system of automatic couplings, ' capable of adaptation to all railway vehicles, should be introduced internationally within a period of five years. Similarly, the International Conference of Railway Workers, - 11 - held in London in 1920, in a debate in the course of which different systems of automatic couplings were examined, adopted a resolution calling the attention of the different governments " t o the necessity of adopting safety appliances in order to reduce the loss of life among railway servants". Finally, during the Congress of the International Federation of Transport Workers, held in Geneva in 1921, after trial of a system of automatic couplings, a resolution was passed calling for the general adoption of an appliance for automatic coupling on all railway systems. The conclusions of organisations and individuals dealing with the question of automatic couplings from the point of view of the safety of railwaymen, however, have not been invariably favourable. In Great Britain the permanent staff of the Board of Trade was anxious, about twenty-five years ago, to compel British railways to adopt automatic couplings, but this did not meet with the approval of the inspecting officers, who were of the opinion t h a t the companies' rules provided the necessary safeguards. 1 At the International Railway Congress at Washington in 1905, representatives of British railway companies took up a doubtful attitude with regard to automatic couplings, implying that there was no immediate need for the introduction of such appliances in the United Kingdom. A Railway Employment Safety Appliance Committee was formed in 1906 and submitted a report in 1908 in which the chairman, after having examined the statistics of accidents to railwaymen and the working conditions of British railways in general, left open the questions : (1) whether, having regard to the small number of accidents in this country due to the present method of coupling and uncoupling goods wagons, there is any such necessity for a change from the existing type of coupler as to justify the large capital outlay and the dislocation of traffic which . . . would be inevitable; (2) How the dislocation of traffic and risk of accident to men and trains during the experimental and transition stages are to be faced.2 On the other hand, French representatives at the Washington 1 Cf. "Lives versus Profits", in The New Statesman, Vol. I, pp. 231, 303; and The Railway Gazette, loc. cit. 2 Third and Fourth Reports of the Committee appointed by the Board of Trade to examine, and where necessary to test, the appliances designed to diminish danger to men employed in railway service. Appendix to Third Report, by Colonel YORKE. Cd. 4213. London, 1908. It is understood, unofficially, that expert opinion has changed very little, if at all, since the issue of this report. - 12 - Congress produced statistical data concerning accidents on the lines of the Nord Company, on the basis of which they stated that they did not see the necessity of assuming the risk of introducing a new system of coupling. The discussion on automatic couplings was continued later in France between the representatives of the railway companies and advocates of automatic couplings; the latter had their doubts as to the value of the scanty facts produced, and strongly advocated the introduction of automatic couplings as the only means of preventing the dangers incurred by railwaymen in coupling and uncoupling operations x. The French Chamber has encouraged this movement by making grants for experiments, and it is understood that at present the provision of automatic couplers is one point in the campaign of the Ministry of Public Works to reduce the number of railway accidents. The above rapid review shows clearly that, while railwaymen and social reformers have been unanimous as to the necessity for introducing automatic couplers to secure the safety of railway workers, technical men have disagreed. Even they have been almost if not completely unanimous, however, on the good results achieved in America by the adoption of automatic couplings, but they have disagreed as to their utility in European countries. In fact, they have been unable to pronounce definitely on the probable effect of the introduction in European countries, no doubt on account of the lack of definite information on the actual risk of coupling accidents in countries outside America, where this risk was admittedly exceptionally high. The resolution of the International Labour Conference quoted above in the opening paragraph implies no definite opinion on the question of automatic couplings, but emphasises the great need for more knowledge of the facts. 1 Cf. Revue générale des chemins de jer et tramways, April, 1914. Also Albert THOMAS: L'Etal et les compagnies de chemins de jer, pp. 220-235. Paris, 1914. II Statistical Methods Statistics of railway accidents serve a two-fold purpose. Their first object is to provide a measure of the accident risk in the different branches of railway traffic and the probability of losses of railway property. Secondly, they provide information as to the nature and causes of railway accidents with a view to their prevention. In order to fulfil these purposes, they have to cover not only accidents to human beings, but also damage to equipment. Further, the nature of railway traffic necessitates a clear distinction between accidents connected with or resulting from the operation of railways and movement of trains, and, on the other hand, accidents where such operation and movement were not involved. Finally, the statistics must show separately the different groups of persons exposed to the risk of any kind of railway accident, i. e. passengers, railway servants, and persons who are not passengers or servants ("trespassers" or "third persons"). On the basis of these three desiderata several different classifications of railway accidents are possible, and the various countries have employed very different methods in classifying such accidents. It would be going too far to examine these methods in detail; the present report is concerned with one group of accidents only, namely, those sustained by railway servants, and has a particular purpose: to determine the importance of coupling and uncoupling accidents within this group. CRITERIA OF COMPARISON In order to get an idea of the risk of coupling and uncoupling accidents to railway servants, three criteria have been selected as bases of comparison. The first is simply the percentage of coupling and uncoupling accidents in the total number of accidents. - 14 - to railway servants. This method, which is the most convenient for illustrating the importance of coupling and uncoupling accidents, necessitates, however, an analysis of the two series compared, since the percentage is naturally dependent on the method of calculating the total number of accidents. But even this percentage does not provide a sufficiently clear and reliable measure of the coupling and uncoupling accident risk, and it is therefore necessary, as in accident statistics in general, to relate the absolute figures of accidents to some other basis in order to calculate accident rates. Two rates have been adopted in the present study. The first is the usual accident frequency rate, i. e. the number of accidents per thousand persons exposed to the risk of these accidents. This necessitates an examination of the basic figures of the number of railway employees, which are also calculated by different methods in the statistics of different countries The second rate is one peculiar to railway accident statistics: the number of accidents per million engine-miles or train-miles (or kilometres); in order to compare this rate for the different countries further examination of the meaning of the terms "engine-mile" and "train-mile" (or "kilometre") is necessary. Thus the three principal elements to be analysed in this survey of statistical methods are: (1) the percentage of coupling and uncoupling accidents in the total accidents to railway employees (accidents covered); (2) the accident rate per thousand railway servants (railway servants covered); (3) the accident rate per million engine or train miles or kilometres (definition of engine and train mile). PERCENTAGE OF COUPLING AND UNCOUPLING ACCIDENTS However differently the various kinds of accidents to railway servants are classified or divided, approximately the same main groups always appear. There is first the group of accidents arising in connection with the operation of trains and with rolling stock, which may result in damage to railway property or in casualties to persons, passengers as well as railway servants. The first group of accidents to railway servants, then, consists of accidents arising from the operation or movement of trains, rolling stock, etc., including collisions and derailments. Another group - 15 - of accidents arising from railway operations, to which railway servants are exposed to a large extent, are those due to shunting operations, operation of brakes and switches, getting on or off engines, carriages, and wagons, walking, trespassing, or standing on the line, etc. Accidents due to coupling and uncoupling operations belong to this group, often as a special sub-heading in the group of accidents due to shunting operations. A third group of railway accidents are those not directly connected with railway working. To this group belong accidents to the administrative staff, workers in railway premises, workshops, warehouses, yards, those engaged in loading and unloading work, etc. When the group of coupling and uncoupling accidents is compared with the total number of accidents to railway servants, it is necessary to analyse the figures in greater detail. They differ not only in that some of the above-mentioned groups of accidents are excluded, but also in the definition of accidents, and most of all in standards of reporting. The first point raising difficulties in comparing the statistics of different countries is the definition of a notifiable accident. The definitions given in the statistical reports are not illuminating. Statistics of accidents to railway employees cover in Germany and Sweden "accidents occurring in the railway service" ; in Belgium and the United States "accidents arising from any operation of railways"; in the United Kingdom "accidents arising from working of railways"; in Norway "accidents in running traffic and extra trains and shunting operations"; in India accidents caused "by the travelling of trains or movement of vehicles used exclusively upon railways". It is sometimes specified (e. g. in France and Switzerland) that only accidents arising from or occurring in connection with the employment properly so called are covered by the statistics. Altogether different is the definition given in the statistics of the Union of German Railway Boards (Verein deutscher Eisenbahnverwaltungerì), in which ¡only compensated accidents are included.1 It is difficult to determine what exactly is meant by the various definitions mentioned above, since their interpretation is dependent on the regulations and practice prevailing, and these, of course, vary considerably not only from one 1 The Union of German Railway Boards includes the managements of the principal railways of Germany, Luxemburg, the Netherlands, Roumania, and Austria-Hungary, as well as of some independent private lines. See note to table IX. - 16 - country to another but also at different periods within the same country. The varying scope of the "total number of accidents to railway employees" constitutes a great obstacle to comparison. In some statistics (e. g. those of Belgium and Switzerland) accidents occurring in connection with collisions or derailment of trains are excluded from the total number of accidents to railway employees, and it might perhaps have been preferable to exclude this group from the total figures for all countries. Owing to the lack of sufficiently detailed information, however, this has not been possible, and the figures have been rendered comparable by adding collisions and derailments to the total in all cases. Other differences arise from the varying classification of accidents not directly connected with railway working, above all accidents occurring in railway premises, workshops, etc. These are not railway accidents properly so called, because they are not, as a rule, incidental to the operation and movement of engines or rolling stock. As the number of accidents to workers on railway premises is by no means negligible, it is likely to influence the total number of accidents and decrease the percentage of coupling and uncoupling accidents. Owing to the different definitions and classifications of this group of accidents, it is impossible to determine with accuracy the extent to which this fact invalidates comparison. It would appear that this group is more or less excluded from the majority of the series of statistics considered, i. e. from those of the European countries and India, but included in those of the United States and Canada. A more important obstacle to comparison of the different statistics of accidents to railway employees arises from the varying standards of reporting. The definitions of a notifiable accident referred to above are usually completed by the stipulation that a notifiable accident must be one resulting in death or a bodily injury for a certain period. It might be expected that at least the figures of fatal accidents would be comparable, but even here differences arise from the fact that non-fatal accidents subsequently resulting in death are not identically treated in all countries. In some cases it is specifically stated that accidents are counted as fatal only if death results within twenty-four hours of the occurrence of the accident (e. g. Germany, Norway, United States), while in other countries the corresponding period is more or less indefinite, e. g. in Great Britain any person so severely injured as to die at any subsequent - 17 - date after the accident, but prior to the publication of the report (about six months later), is classified as "killed". This difference in the standard of reporting of fatal accidents is not without importance, since it is understood that the number of persons dying more than twenty-four hours after the occurrence of the accident is considerable. In the case of non-fatal accidents the variation in the period for which the accident must cause disability if it is to be notifiable is still greater. Thus accidents to railway servants must be notified in Norway as soon as the injured person needs medical treatment; in the United Kingdom if they cause absence from work for one day; in the United States if they result in injury sufficient to incapacitate the employee from performing his ordinary duties for more than three days in the aggregate during the ten days immediately following the accident; while the corresponding period is six days in Switzerland and fourteen days in Germany. It would appear that the same standard as in Germany is applied in countries included in the statistics published by the Union of German Railway Boards. The French statistics cover only socalled "severe accidents" which are defined as those occurring on the principal railway lines to trains or rolling stock and resulting in the death of or injury to persons, as well as other accidents sustained by persons and causing incapacity for work for more than twenty days. No information on this point is available for Canada, Belgium, or Sweden. These differences in standards of reporting of course apply also to coupling and uncoupling accidents. In calculating the relation between the absolute figures of accidents to railway employees in general and those for coupling and uncoupling accidents, the effect of these divergences is partly neutralised, but in accident rates per number of employees or train-kilometres they constitute the chief obstacle to comparison. It may be noted that in this report accidents due to coupling are combined with those due to uncoupling. As it is understood that automatic couplings are likely to affect the number of uncoupling accidents to a considerably less degree than the number of coupling accidents, it would have been interesting to distinguish between these two groups. This distinction is, however, made only in the statistics of Belgium and Norway, and it has therefore been necessary to give both groups combined for all countries '. 1 According to the Belgian statistics for 1904-1913, the relation of coupling to uncoupling accidents (fatal) was 4 to 1, and according to the Norwegian 2 - 18 - Finally, it should be borne in mind that some accidents indirectlydue to coupling or uncoupling are not included in the statistics under the heading of coupling and uncoupling accidents. Thus, for example, a heading such as "untoward occurrences during shunting" (Swiss statistics) includes accidents due to men slipping, etc. while carrying out coupling; the headings "forbidden operations" or "culpable accidents" (Dutch statistics) include accidents due to breach of the rules providing that vehicles must be stationary when being coupled or uncoupled. Classifications of this kind tend in certain cases to reduce the number of accidents attributable to coupling and uncoupling operations. ACCIDENT RATE PER THOUSAND RAILWAY EMPLOYEES The above remarks refer to the absolute figures of accidents to railway employees, which constitute the basis of all the statistics. Although an idea of the relative importance of coupling and uncoupling accidents in the several countries can be obtained from these figures by means of the percentage, this method does not provide an adequate measure of the risk of coupling and uncoupling accidents. It is necessary, therefore, to compute accident rates on the basis of certain other figures. As in industrial accident statistics, the easiest method of measuring the risk of accidents to railway employees is to calculate a frequency rate, i.e. the number of accidents per thousand persons exposed to the risk of such accidents. The calculation of this frequency rate unfortunately involves considerable difficulties, and must therefore be considered rather incomplete. With the exception of those of Sweden none of the statistics available give any accident frequency rates at all. Practically all of the rates presented have therefore been calculated by the International Labour Office on the basis of figures of the number of railway employees as given in the official statistics. This basic number of railway employees, again, is not arrived at by similar methods in the various series of statistics. In general, railway employees are divided into five classes: administrative, station, train, locomotive, and workshop staff. It is clear that statistics for the years 1914-1920 it was about 11 to 4. As, however, this relation varies greatly from year to year, it is difficult to say precisely how far the number of coupling accidents exceeds that of uncoupling accidents. - 19 - all these groups should not be taken into account in calculating the accident! exposure. The whole group of administrative staff should be exempted on account of the nature of their work, as well as the group of workshop and similar employees who are not exposed to the risk of railway accidents proper. Moreover, several classes of employees belonging to other groups, such as telegraphists, telephonists, cashiers, clerks, and similar members of the station staff, should be eliminated from the calculation of accident exposure. The difficulty is that some of the statistics do not classify railway servants so as to allow for the necessary distinctions. Accident exposure comprises all railway servants, regardless of the nature of their work, in the statistics of Canada, Finland, France, Germany, India1, the United Kingdom, and the United States. Other countries adopt a classification of railway employees which excludes certain groups not exposed to the risk of railway accidents properly so called. Thus the administrative staff is excluded in the statistics of Belgium and the Union of German Railway Boards, the administrative and workshop staff in Norway, the administrative, office, workshop, warehouse, and ferry-boat staff in Sweden; and the administrative, workshop, and auxiliary staff in Switzerland. Another difference is that the statistics of Germany, the Union of German Railway Boards, and Sweden give the number of full-time employees. In the two first-named this is done very accurately; the annual average of employees on daily pay and all workers in train services is taken as equal to the number of days-duty divided by 365, and the average number of workers in line-guarding, workshop, and similar departments is taken as equal to the number of days worked divided by 300. In the United States, too, the number of railway employees covered is the average number of servants on duty. In general, however, the figures for the number of railway employees refer to the end of the year concerned. The figures given in the statistics of the United Kingdom are only approximate. On the basis of the total number of railway servants the frequency rates for all accidents to railway employees may be calculated. To calculate the frequency rate for coupling and uncoupling accidents it would naturally be necessary to compare the number of 1 The figure for India is not comparable with those for other countries, the railway staff being composed of Europeans, Anglo-Indians, and Indians; this makes the staff relatively much larger than is usual in European and American countries. - 20 - this group of accidents with the number of the group of railwaymen actually exposed to the risk, i.e., in most cases the number of shunting operators.1 Unfortunately, it is impossible to calculate this rate with any degree of accuracy. The existing statistics not only give no special rates for coupling and uncoupling accidents, but also do not specify the number of railway servants in the particular group concerned. In the absence of the necessary information, therefore, the number of coupling and uncoupling accidents is in all cases computed simply on the basis of the total number of railway servants, which of course makes the coupling and uncoupling accident rates very much too low.2 ACCIDENT RATE PER MILLION ENGINE OR TRAIN MILES OR KILOMETRES In railway accident statistics the usual frequency rate, based on the number of railway servants exposed to the risk, is not employed as the principal measure of the accident risk. For passenger accidents rates are calculated on the basis of passengermiles (kilometres) or axle-miles (kilometres) for carriages in passenger traffic ; similarly, for accidents to railway servants, the accident risk may be expressed either by means of engine-miles or trainmiles (kilometres) or axle-miles (kilometres). While some of the sets of statistics give no rates at all, others give rates per enginekilometre and train-kilometre (Germany, Union of German Railway Boards, Norway), and a few give both the rates mentioned above. In the present report the accident rate per million enginekilometres or train-kilometres has been selected as the third standard of comparison, because information on this point has been found for all countries. It has sometimes been suggested that the rate per axle-kilometre would be more satisfactory, since it allows for the influence of the number of carriages and wagons upon the frequency of accidents, but the lack of information on 1 It has been pointed out that not only shunting operators, but also porters (in the case of passenger stock) and goods guards (in the case of goods wagons) are probably exposed to the risk of coupling and uncoupling accidents. Cf. Appendix to the Third Report of the British Safety Appliances Committee, 1906 (Cd. 4230), p. 1. 2 In the last chapter of this report an attempt is made to estimate the real frequency rate of coupling and uncoupling accidents. - 21 - the number of axle-kilometres (miles) for all countries makes it impossible to calculate this rate. Moreover, the two rates are in practice very similar, as has beeen shown by test calculations for countries where both rates were available 1. With regard to differences in the definition of engine-mile and train-mile, it should be noted that " engine-mile " has a wider scope than " train-mile ", as it includes not only train-miles proper but also miles run by engines used in the administration and maintenance of service, upkeep of permanent way, shunting, etc. In the appended tables train-kilometres are given for most countries, i.e. Belgium, Finland, France, and Sweden, and for the statistics of the Union of German Railway Boards, and train-miles for the United Kingdom, the United States, Canada, and India, but nontraffic kilometres are included in some statistics (e.g. Sweden and the Union of German Railway Boards). Engine-kilometres are given for Norway (excluding kilometres run by engines used for maintenance of permanent way), Germany and Switzerland. For continental Europe the figures indicate the kilometres run by national and foreign engines on national lines. In this connection a similar reservation must be made in respect of coupling and uncoupling accidents as in the case of accident frequency rates per thousand railway servants. It is true that the number of engine-miles and train-miles represents the importance and density of railway traffic, but it must be admitted that the comparison of the number of coupling and uncoupling accidents with the number of engine-miles and train-miles is somewhat artificial. The reason for which it is given in the appended tables is that it is impossible to calculate more satisfactory rates. For example, the number of coupling and uncoupling accidents per thousand couplings and uncouplings effected would have been a better rate, but it can be calculated only for the United States 2. It may be mentioned in conclusion that railway accident statistics, unlike those of industrial accidents, rarely give any information 1 For instance, according to the Swedish statistics for 1906 to 1915, the two rates have varied as follows: Per million Per ten million v train-kilometres wagon-axle-kilometres 1906-10 (average) 0.91 = 100 0.24 = 100 1911 0.56 = 62 0.14 = 58 1913 0.87 = 96 0.22 = 92 1915 1.16 = 127 0.29 = 121 8 See table I in the appendix, and below, p. 25. - 22 - on the nature of the injury or on the duration and extent of disability resulting from accidents. It has therefore been impossible to calculate severity rates. It will be seen from the above indications that care is necessary in drawing conclusions from the figures given in the statistics for diffeient countries. It has been shown especially that the accident rates are rather inadequate for coupling and uncoupling accidents ; even the rates given differ among themselves owing to differences in the scope of the statistics and in the methods of calculating accident exposure. With these limitations, however, some general indications as to the importance of coupling and uncoupling accidents in the various countries may be given. Ill Coupling and Uncoupling Accidents in Countries where Automatic Couplings are in Use It has already been pointed out that the countries for which statistics have been secured must be divided into two groups: first, those where rolling stock is generally fitted with automatic couplings, i.e. the United States and Canada; and, secondly, those where the system of coupling is non-automatic, i.e. European countries and India. It is natural to proceed first to an examination of the statistical evidence from the first group. The statistics of the two countries being of different scope and value, they must be considered separately. UNITED STATES The results achieved in the United States by the compulsory introduction of automatic couplings more than twenty years ago constitute a lesson quoted by everyone who discusses the question. The American statistics refer to the periods both prior and subsequent to the introduction of automatic couplings, and are therefore of particular value and interest. In order rightly to understand their meaning, a brief reference to the original coupling system in the United States is necessary. It is a well-known fact that before the introduction of automatic couplings in the United States the only appliance in use was what is known as the link and pin coupling. This was a most dangerous device; the worker had to stand between two wagons while they were being brought together and hold the link in one hand between the central buffers, into the jaws of which the link had to be guided until these buffers were nearly in contact with each other, and then snatch his hand away to avoid its being crushed, while with the other hand he dropped the pin into place. Accidents with this - 24 - appliance were appallingly frequent and in America it was generally called the "man killer". The introduction of the automatic coupling soon effected a great improvement, which has been indicated by the Interstate Commerce Commission in the following manner. Had the number of railway accidents due to shunting operations increased with its previous rapidity, it would have risen from 34,456 in 1893 to 81,041 in 1907. The figure for the latter year, however, was not more than 61,871, and this in spite of the great increase in the number of railway employees (about 116,000 from 1893 to 1907) and of vehicles (about 50 per cent.) and notwithstanding the increased weight of engines and vehicles. The Interstate Commerce Commission ascribes this progress to the compulsory introduction of automatic couplings. Further evidence of this improvement is shown in tables I and II in the appendix. These are based on the statistics of the Interstate Commerce Commission, which, under the Accident Act of 3 March 1901 and later under the Accident Reports Act of 6 May 1910, receives monthly reports from common carriers engaged in interstate and foreign commerce "of all collisions, derailments, or other accidents resulting in injury to persons, equipment, or road bed, . . . which report shall state the nature and cause thereof and the circumstances connected therewith". The statistics cover all reportable accidents, including industrial and non-trade accidents to railway employees. The definition of reportable accidents differs somewhat in the periods before and after the coming into force of the 1910 Act. The earlier statistics included only four classes of accidents : collisions, derailments, casualties to passengers, and casualties to employees on duty. Since 1910 there have been added to these figures accidents to employees not on duty, in and around shops, on boats or wharves, and at other places. For the year ending 30 June 1911 this meant an addition to train service accidents of 486 fatal and 43,108 non-fatal accidents. This difference must be borne in mind when the increase of accidents after 1910 is considered *. The statistics of the Interstate Commerce Commission show that the number of coupling and uncoupling accidents (fatal and nonfatal) decreased from 11,710 in 1893 to a minimum of 2,276 in 1902; the figure for 1907 was comparatively low (4,244), while 1 Moreover, the figures for 1908-1910 do not cover switching and terminal roads. - 25 - another marked drop may be noted from 1907 to 1914, the number of coupling and uncoupling accidents in the latter year being only 2,863. The decrease in coupling and uncoupling accidents was thus 80 per cent, from 1893 to 1902, 64 per cent, from 1893 to 1907 and 76 per cent, from 1893 to 1914. The rise in the absolute number of coupling and uncoupling accidents after 1902 is readily explained by the very rapid increase both in the number of railway employees and in traffic. The number and rates of these accidents have tended to decline steadily as will be seen from the figures given below. It is of interest to note first the percentage of the total number of accidents to railway employees represented by coupling and uncoupling accidents. PERCENTAGE OF COUPLING AND UNCOUPLING ACCIDENTS Year 1893 1902 1907 1914 1905-1914 (average) Fatal 15.88 4.82 6.59 5.24 6:27 Non-fatal 35.54 4.18 4.50 1.63 2.87 It is needless to comment on the large drop in the percentages, particularly those of non-fatal accidents. The following figures, giving the number of accidents per thousand coupling and uncoupling operations effected by locomotive men, firemen, conductors, and train men, are of particular interest, since this is the only rate of its kind in railway accident statistics. NUMBER OF ACCIDENTS PER THOUSAND COUPLING AND UNCOUPLING OPERATIONS Year 1893 1902 1907 Fatal 2.41 0.63 1.01 Non-fatal 62.77 9.37 13.37 These figures show a decrease, from 1893 to 1907, of about 60 per cent, in fatal accidents and nearly 80 per cent, in non-fatal accidents. The frequency rate for accidents to all railway employees and for coupling and non-coupling accidents is given below. - 26 - NUMBER OF ACCIDENTS PER THOUSAND RAILWAY Total accidents EMPLOYEES Coupling accidents Year 1893 1902 1907 1914 1905-1914 (average) Fatal Non-fatal Fatal Non-fatal 3.12 1.78 2.71 1.87 2.07 36.31 42.48 52.42 94.93 66.55 0.50 0.12 0.18 0.10 0.14 12.90 2.50 2.36 1.55 1.91 The above figures may be completed by another group of accident rates. NUMBER OF ACCIDENTS PER MILLION TRAIN-MILES Total accidents Coupling accidents Year 1893 1902 1907 1914 1905-1914 (average) Fatal Non-fatal Fatal Non-fatal 3.23 3.28 3.98 2.62 3.04 37.54 55.77 76.95 133.10 93.23 0.51 0.16 0.26 0.14 0.19 13.34 2.33 3.46 2.17 3.04 The foregoing statistics bear striking testimony to the remarkable fall in the accident risk in coupling and uncoupling operations on American railways — a fall all the more conspicuous as the general accident risk has increased considerably from year to year. The statistics thus support the official statement made by the Interstate Commerce Commission to the United States Congress in 1907. It is a noteworthy fact that, in spite of the general increase in accidents,. the accidents sustained by train service employees engaged in coupling and uncoupling cars has been constantly declining. This result shows that the Congress has done well in accepting the Safety Appliances Act, and justifies the efforts of the Government to apply this law drastically. - 27 - CANADA It was stated at the International Railway Congress in 1905 that automatic couplings were at that date extensively used in Canada l, and it is understood that the number of automatic couplings in use has increased in subsequent years. The statistics presented for Canada (table III) only cover the period from 1912 to 1920. As no figures are available for the period prior to the introduction of automatic couplings in Canada, the statistics are much less interesting than those for the United States. As a preliminary to comparing the figures for Canada, where automatic couplings are in fairly general use, with those for other countries where couplings are non-automatic, certain observations should be made. The Canadian figures relate to two groups of accidents: (a) those resulting from movement of trains, locomotives, or cars, and (6) those in which the movement of trains, etc. was not involved. Coupling and uncoupling accidents belong to group (a). As details regarding the standard of reporting of accidents and any changes in it are not available, it is impossible to give a satisfactory explanation of certain facts. It will be seen from table III that the total number of fatal accidents decreased during the period covered, while the total number of non-fatal accidents increased very much, from 2,924 in 1912 to 7,719 in 1920. At the same time, the number of fatal coupling and uncoupling accidents fell while that of similar nonfatal accidents varied considerably from year to year, no regular movement of increase being noticeable. Consequently the percentages of total accidents to railway employees represented by coupling and uncoupling accidents remained practically stable in the first group (fatal), and decreased in the second. This will be seen from the following figures. PERCENTAGE OF COUPLING AND UNCOUPLING ACCIDENTS Year 1912 1920 1912-1920 (average) Fatal 7.7 7.1 7.5 Non-fatal 7.1 3.1 3.9 1 By Mr. J . E . M U H L F E L D . Cf. Bulletin du Congrès international de fer, Vol. X X , No. 7, J u l y 1906, p. 778. des chemins - 28 - Apart from the figure for non-fatal accidents in 1912 — an exceptional year in this respect — the percentage of coupling and uncoupling accidents has been stabilised at rather over 7 for fatal and at just under four for non-fatal accidents. While fatal accidents decreased and non-fatal ones increased, the number of railway employees and train mileage remained remarkably stable during the period under consideration. Thus the accident rates varied in the same sense as the actual number of accidents, as will be seen from the following tables. ACCIDENT RATE PER THOUSAND EMPLOYEES Total accidents Coupling accidents Year Fatal 1912 1920 1912-1920 (average) 1.50 1.08 1.31 Non-fatal 18.2 49.7 28.9 Fatal Non-fatal 0.12 0.08 0.10 1.33 1.56 1.11 ACCIDENT RATE PER MILLION TRAIN-MILES Total accidents Coupling accidents Year Fatal 1912 1920 1912-1920 (average) 2.32 1.42 1.85 Non-fatal Fatal Non-fatal 29.00 65.60 40.55 0.18 0.10 0.13 2.06 2.05 1.55 The principal feature of the above tables is that the marked increase in the rates of non-fatal accidents in general is accompanied by a decrease in the corresponding rates of coupling and uncoupling accidents. It may be assumed that the stabilisation of the percentage of coupling and uncoupling accidents at a relatively low figure, and the decrease in the rates of coupling and uncoupling accidents, concurrently with an increase in the rates for all accidents to railway employees, is partly due to the introduction of automatic couplings in Canada. - 29 - COMPARISON The coupling and uncoupling accident risk in the United States and in Canada may be compared on the basis of the percentages and rates of these accidents, taking for the United States the averages for the period 1905-1914, and for Canada the averages for 1912-1920. The risk appears on the whole to be rather similar in the two countries. The percentage of coupling and uncoupling accidents is, in Canada, 7.5 for fatal and 3.9 for non-fatal accidents, a little higher than in the United States, where the corresponding percentages are 6.3 and 2.9. On the contrary, the coupling and uncoupling accident rates are higher in the United States. The rate per thousand employees is in the United States 0.14 for fatal and 1.91 for non-fatal accidents, while the corresponding figure for Canada are 0.10 and 1.11. Similarly the rate per million train-miles is in the United States 0.11 for fatal and 3.04 for nonfatal accidents, the corresponding figures for Canada being 0.13 and 1.55. It will be seen from these figures that the differences between the two countries are of minor importance, except in the case of the rate for non-fatal coupling and uncoupling accidents per million train-miles. IV Coupling and Uncoupling Accidents in Countries where Couplings are Non-Automatic The second object of the present report is to note the accident risk in coupling and uncoupling operations in countries where the coupling system is non-automatic. These countries include the United Kingdom and India (tables IV and V), though a considerable number of railways in the latter country use automatic couplings, and all the countries of continental Europe (tables VI to XIII). It may be stated at the outset that the non-automatic coupling system is not identical in all countries, and is different on the various kinds of vehicles. It should be particularly noted that the coupling system on passenger stock is usually different from that on goods wagons. Precise distinctions in this respect are made only in the statistics of Great Britain, which distinguish accidents connected with coupling and uncoupling of passenger stock from those due to coupling and uncoupling of goods stock. Here this is not taken into account, but it is understood that the system of non-automatic couplings with double buffers is used in all the countries of continental Europe. It is not proposed to examine the statistics country by country, or year by year, but to compare the figures for all the countries covered and for the total periods given in the appended tables. This is the best method of presentation for purposes of international study, in spite of the fact that the different methods of compiling the statistics make it rather difficult. These differences may, however, be considerably reduced if a clear distinction is drawn between the figures of fatal and non-fatal accidents. Those of fatal accidents, in spite of differences in the regulations concerning their notification, may be regarded as fairly comparable. The figures of non-fatal accidents are much less comparable owing to their different scope and methods, as explained in chapter II. The method of presentation is based on the three criteria previously mentioned: (1) percentage of coupling and uncoupling accidents; (2) accident rate per thousand railway employees and (3) accident rate der million train-kilometres. - 31 - PERCENTAGE OF COUPLING AND UNCOUPLING ACCIDENTS The first method of estimating the importance of coupling and uncoupling accidents is to calculate the percentage of the total number of accidents to railway employees represented by these accidents. These percentages have been calculated on the basis of the averages given in the appended tables. In the following pages the countries are arranged so as to group together the statistics which are most nearly comparable, at the same time taking into consideration the geographical position of the different countries. It will be remembered that in the computation of the total number of accidents the principal differences in method are : (1) t h a t the standard for reporting accidents as fatal is death within twenty four hours in the statistics of Germany, the Union of German Railway Boards, and Norway, while elsewhere it is not specified; (2) that the standard for reporting non-fatal accidents varies from disability for one day (United Kingdom) and six days (Switerzland) to 14 days (Germany and the Union) and 20 days (France), while the standard of reporting is not precisely specified in some of the statistics (Norway, Belgium, Finland, and Sweden); (3) that accidents not directly connected with railway working, especially accidents in workshops, etc., are excluded from the total to varying degrees. PERCENTAGE OF COUPLING AND UNCOUPLING ACCIDENTS IN VARIOUS COUNTRIES Country United Kingdom India Belgium . . . . France . . . . Switzerland . . Luxemburg . . Netherlands . . Roumania . . . Austria-Hungary Germany 1 . . . Sweden . . . . Norway . . . . Finland . . . . Period covered Fatal accidents Non-fatal accidents 1911-1920 1919-1920 1909-1913 1919-1923 1913-1922 3.27 6.18 8.00 9.76 11.38 11.33 11.15 20.81 16.39 11.70 1905-1909 7.89 11.02 1905-1909 1905-1909 1906-1916 1913-1920 1911-1920 14.57 15.58 14.18 19.22 2.63 14.69 14.26 22.21 19.00 1.74 1 For Germany, two series of figures are available (tables IX and X). In this and the following tables the figures for the period 1905-1906 are given in view of their comparability with other statistics of the Union of German Railway Hoards. - 32 The above figures go to show that the very restricted group of coupling and uncoupling accidents represents, in the total number of railway accidents of all descriptions, by no means an insignificant proportion. Taking first only fatal accidents, it seems to be a fact that the accident risk in coupling and uncoupling in the United Kingdom is very slight as compared with the total number of accidents to railway servants. The percentage would be little decreased even if the accidents occurring on railway premises were added to the total. It will be seen, however, that the percentages for other countries show a considerably higher degree of accident risk in coupling and uncoupling operations. In continental Europe coupling and uncoupling accidents generally amount to from 10 to 15 per cent, of the total when non-train accidents are included therein, and to from 15 to 20 per cent, if accidents due to the movement of trains and rolling stock only are included in the total. The highest percentages are in Austria-Hungary, Germany, Sweden, and Norway, i.e. in countries in central and northern Europe, where no doubt the hard climate during the winter renders shunting operations exceptionally dangerous1. In view of differences in statistical method it is natural that the percentages of non-fatal accidents should vary more than those of fatal ones. The non-fatal percentage appears to be appreciably higher than the fatal in most of the countries concerned, while the two figures are practically equal in three countries (AustriaHungary, Switzerland, and Norway). The non-fatal percentage varies between 11 and 22, the maximum being reached in Sweden. The percentage is also very high in France (16), Norway (19), and Belgium (21) \ The fact that the proportion of non-fatal coupling and uncoupling accidents is sometimes greater than and sometimes equal to that of the fatal group, is sufficiently explained by the differences in the systems of reporting non-fatal accidents, which make it impossible to draw any general conclusions as to the severity of these accidents. Only detailed classification of coupling and uncoupling accidents by the nature of the injury and the extent of the resulting disability — a classification which is not generally given — would provide the information necessary for such conclusions. This question must therefore be left on one side. 1 The very low percentages for Finland are exceptional, but owing to the restricted scope of the statistics and the absence of information on methods employed, it is impossible to account for this. 33 - ACCIDENT RATE PER THOUSAND RAILWAY EMPLOYEES The accident rate per thousand railway employees is the principal measure of accident risk available for this study. As was pointed out in chapter II, however, this rate is more dependent on the statistical methods employed than the percentage of accidents, since, in addition to the difference in the figures of accidents, differences in the calculation of accident exposure also have a bearing upon the reliability of the rates. The chief differences in the calculation of accident exposure have been found to be the following: (1) the statistics of the United Kingdom, India, and France give the total number of employees, i.e. including large numbers of persons not exposed to the accidents covered; (2) the administrative staff is excluded from the statistics of Belgium and the Union of German Railway Boards ; and (3) administrative and workshop staff, etc. are excluded from the statistics of Switzerland Sweden, and Norway. ACCIDENT RATES PER THOUSAND RAILWAY EMPLOYEES IN VARIOUS COUNTRIES 1 Fatal accidents Non fatal accidents Country All United Kingdom India Bulgium France Switzerland . . Germany . . . Luxemburg • • Netherlands . . Roumania . . . Austria-Hungary Sweden Norway Coupling and uncoupling All Coupling and uncoupling . 0.62 0.71 1.17 0.02 0.04 0.09 8.61 1.34 8.98 1.02 0.15 1.29 . . •) .> .) . 0.97 0.99 0.11 0.15 45.85 2.38 5.46 0.34 0.99 0.08 2.33 0.26 0.70 1.00 0.62 0.59 0.10 0.14 0.12 0.02 5.25 5.64 4.92 1.78 0.77 1.25 0.93 0.03 It will be seen from the above table that, despite differences in statistical methods, the fatal accident risk is strikingly similar in the different countries. It varies from 0.6 in the United Kingdom, 1 The periods covered are the same as in the preceding table, except for the United Kingdom, where it is 1911-1914. 3 - 34 - Finland, and Norway to 1.2 in Belgium. In connection with the rates of fatal coupling accidents the figure for the United Kingdom calls for special attention, as it is by far the lowest. The basis of this rate, i.e. the number of railway employees, includes, it is true, workers on railway premises who are not exposed to the danger of railway accidents proper, these being the only ones included in the total of accidents. As has been pointed out above, however, this is also the case in the statistics of some other countries where the rates of fatal coupling and uncoupling accidents are markedly higher than in the United Kingdom. The rates for the countries of continental Europe vary between 0.02 and 0.15 per thousand railway employees. The mode appears to be approximately 0.11 to 0.12. Finland, where the figures are very low in this case also, Luxemburg, the Netherlands, and Roumania, have the lowest coupling and uncoupling accident risk, and Sweden, where the general rate was nearly as high as in Belgium, has twice as high a rate for coupling and uncoupling accidents as the latter country. Germany is at the top with the relatively high rate of 0.15 coupling accidents per thousand railway employees. A first perusal of the column of non-fatal accident rates gives the impression that these rates vary to such an extent that any comparison would be futile. It is, however, to be noted that the figures for Switzerland and Finland are so exceptional that they cannot be sufficiently explained by the statistics ; the standard of reporting in the former is higher than in many other countries, and the number of employees, though representing only the train service men properly so called, is roughly comparable with that in Sweden and Norway, for example. Further, it is needless to try to compare conditions prevailing in India with those in European countries; the exceptionally low rate for India may partly be explained by the very large number of railway employees previously mentioned. If Switzerland, Finland, and India are eliminated, the variation of the general non-fatal rates may be given as from 2.3 (Luxemburg, Netherlands, Roumania) to 9.0 (Belgium), while the rates of coupling and uncoupling accidents vary from 0.26 (Luxemburg, Netherlands, Roumania) to 1.29 (Belgium). These differences can be explained to a great extent by the difference in the standards of reporting. There is no doubt it is the one-day standard of reporting that makes the rate for coupling and uncoupling accidents in the United Kingdom higher than in many other European countries. - 35 - ACCIDENT RATE PER MILLION TRAIN-KILOMETRES The final comparison to be made is that between the accident rates per million train-kilometres. The term "train-kilometres" may not seem quite accurate, since some statistics cite engine or locomotive kilometres. But, though the bases of these rates cannot be deemed strictly comparable, the differences may be considered as of minor significance. It is also to be noted that the rates are per million ivam-kilometres also for Anglo-Saxon countries, the rates per million train-miles having been divided by 1.6. ACCIDENT RATES PER MILLION TRAIN-KILOMETRES IN COUNTRIES VARIOUS 1 Fatal accidents Non-fatal accidents Country A,ii United Kingdom India Belgium France Switzerland . . Germany . . . Luxemburg . . Netherlands . . Roumania . . . Austria-Hungary Sweden Norway Finland 1 Coupling and uncoupling All Coupling and uncoupling . 0.60 1.76 0.90 0.02 0.11 0.07 8.11 3.65 6.92 0.96 0.41 1.43 . . 1 0.72 0.98 0.09 0.15 34.17 2.35 4.00 0.33 1.03 0.08 2.41 0.27 0.95 0.88 0.30 0.76 0.14 0.13 0.06 0.02 7.11 5.16 2.38 2.28 1.04 1.19 0.45 0.04 ] . The tables given in the appendix show a curious coincidence: the number of employees is usually approximately equal to the number of thousand, train-kilometres, though in India and AustriaHungary it is much greater, and in Switzerland and Norway less. Rates based on the two absolutely different factors are thus more or less similar, which facilitates comment. Taking first fatal accidents, it will be seen that the variation in rates for coupling and uncoupling accidents is wider than that in the rates for all accidents. The coupling and uncoupling rate in the United Kingdom (0.02) especially differs from the others, 1 T h e periods covered are t h e same as in t h e preceding table (United Kingdom, 1911-1914). - 36 - being lower than in most other countries and seven times lower than in several central and northern European countries. The corresponding rates for continental Europe vary from 0.02 to 0.15, that is to say, between exactly the same limits as the rates per thousand employees. The lowest rate in continental Europe is in Finland, and the highest, as in the previous case, in Germany. The mode may be fixed at 0.11 to 0.12 coupling and uncoupling accidents per million train-kilometres, i.e. practically at the same figure as the rate per thousand employees. With regard to non-fatal accident rates, the observations with regard to Switzerland made in connection with the preceding table hold good here. The Swiss rate is all the more exceptional, as in this case its basis may be considered as equivalent to that of other countries. On the other hand, the rate for India appears in this case more or less normal. The general non-fatal accident rates of the United Kingdom and of Austria-Hungary appear particularly high. The rate of non-fatal coupling and uncoupling accidents varies in continental Europe (except in Switzerland and Finland) from 0.27 (Luxemburg, Netherlands, Roumania) to 1.43 (Belgium). The limits of variation are very much the same as those for rates per thousand railway servants. The outstanding feature of the comparisons made in the preceding pages is the exceptional position of the United Kingdom. This is particularly noteworthy in the comparison of fatal accidents. The British percentage of these accidents is lower than in any other country, and similarly the rates for coupling and uncoupling accidents are in most cases 3 to 5 times lower than the corresponding figures for countries with the highest rates. In the rates of non-fatal accidents the United Kingdom is less conspicuous, but this is no doubt due to the one-day standard of reporting. The comparatively low figures of coupling and uncoupling accidents in Great Britain were attributed by Colonel Yorke in 1908 to the abolition of the D shackle, to better lighting, better discipline among the men, and particularly to the universal adoption of the shunting pole, which renders it unnecessary for the men to go between goods wagons for the purpose of coupling; it should, however, be noted that this pole cannot be used for passenger stock \ 1 Appendix to Third Report of the Railway Employment Safety Appliance Committee. Cd. 4213, p. 6. V Comparison of Coupling and Uncoupling Accident Risk in America and Europe The preceding chapters have dealt, on the one hand, with the progress in the safety of coupling and uncoupling operations in those countries where automatic couplings have been introduced, and, on the other hand, with the risk of coupling and uncoupling accidents in countries where the couplings are non-automatic. There remains one question. Different opinions have been expressed as to the safety of coupling and uncoupling operations in America since the introduction of automatic couplings as compared with the conditions prevailing in European countries. The next task is therefore to compare the risk of coupling and uncoupling accidents in these two groups of countries. In chapter III it was shown by statistics that the introduction of automatic couplings in the United States has effected a vast improvement in the safety of railwaymen. Since their use was made compulsory, during the period from 1894 to 1903, the risk of coupling and uncoupling accidents has decreased strikingly, while, at the same time, the risk of railway accidents in general has risen constantly 1, concurrently with an increase in the volume, density, and speed of traffic as well as other circumstances. The proportion of accidents due to coupling and uncoupling as compared with the total of accidents to railway employees of all classes, which in 1893 attained the startling figures of 16 per cent. for fatal and 36 per cent, for non-fatal accidents, dropped during the first decade after the introduction of automatic couplings to 5 and 4 per cent., and during the subsequent decade to 5 and 2 per cent., respectively. The decrease in the relative importance of coupling and uncoupling accidents was thus 70-90 per cent. 1 This applies to the pre-war period covered in table II. More recent statistics for the United States show that the general accident risk also has considerably diminished during recent years. - 38 - Similarly, the rates of coupling and uncoupling accidents per thousand railway employees and per million train-miles decreased from 1893 to 1914 by about 75-85 per cent. This evidence is clear enough ; but it has not always been admitted that anything is to be learned from it by other countries. This argument is not altogether unfounded. As was pointed out in chapter III, the non-automatic coupling system of the United States railways was extremely dangerous. In no other country have coupling and uncoupling accidents formed one-third or more of the total of railway accidents, as they did in America. Nowhere have the American frequency rates of coupling and uncoupling accidents before the introduction of automatic couplings — 0.5 fatal and 12.5 non-fatal accidents per thousand employees and practically the same numbers per million train-miles — been equalled. The point is that the introduction of automatic couplings, while it diminished the exceptionally high risk prevailing before, has not brought this risk down to a sufficiently low level. It has been argued, in fact, by many writers that the risk of coupling and uncoupling accidents in America with automatic couplings is still higher than in Europe with the non-automatic system of coupling with double buffers. The question is whether this contention is borne out by statistics. This point has been dealt with especially by British and French writers and they have usually tended to answer the question in the affirmative, mainly on the basis of the statistics for the United Kingdom, which indeed are very elaborate. The preceding analysis has shown that the figures for the United Kingdom are actually lower than those for the United States. While the proportion of coupling and uncoupling accidents to the total of accidents to railway employees was in the United States (average for the period 1905-1914) 6.3 per cent, for fatal and 2.9 per cent. for non-fatal accidents, the corresponding figures for the United Kingdom (average for the period 1911-1920) are 3.3 and 2.5 *. Further, while the average rates of coupling and uncoupling accidents per thousand employees were in the United States 0.14 fatal and 0.91 non-fatal, the corresponding rates for the United Kingdom (1911-1914) are only 0.02 and 1.02. Similarly, the rates of coupling and uncoupling accidents per million train-miles in the United States rates were 0.19 fatal and 2.68 non-fatal, 1 Including also accidents occurring on railway premises, as in the United States. - 39 - while the United Kingdom rates are considerably lower, namely 0.03 and 1.53. These comparisons, based on the statistics of the United Kingdom, may be completed by certain data given in France, where the question of automatic couplings aroused considerable interest before the war. The proportion of coupling and uncoupling accidents to the total accidents to railway employees was given as 3.4 per cent. for the ten years ending 1913 — a figure that may be obtained from a number of different publications 1. Similarly, the only rates for coupling and uncoupling accidents given for France, namely 0.06 fatal and 0.57 non-fatal accidents per thousand railway employees for the period 1898 to 1904 2, were considerably lower than the corresponding rates for the United States. The evidence of the statistics of the United Kingdom and France has been often used as a generalisation and applied to European countries as a whole. It is not surprising, therefore, that the opinion came to be generally held that the accident risk in coupling and uncoupling operations in Europe was insignificant. It has been shown, however, that conditions in the United Kingdom are quite exceptional as compared with those indicated by the statistics of the countries of continental Europe, and that recent French official statistics s negative the figures quoted above. It is not proposed to discuss here the value of the statistics of continental countries; the reader is referred to chapters II and IV. It will be sufficient to note here the principal results of the survey of conditions in continental Europe and to compare them with corresponding figures for America. The proportion of coupling and uncoupling accidents to the total accidents to railway employees vary within the following limits (percentages) : PERCENTAGE OF COUPLING AND UNCOUPLING ACCIDENTS Continental Europe North America (United States and Canada) 1 Fatal 3-19 6-7 Non-fatal 2-24 3-4 Revue générale des Chemins de fer et des Tramways, Apr. 1914. Albert THOMAS; op. cit., p. 223. R. MÜLLER and J. TOMSCHIK: Eisenbahnerschutz, Anträge und Erläuterungen, p. 12. Vienna, 1914. 2 These figures refer to the lines of the Nord Company only; they were presented to the International Railway Congress at Washington, 1905. Cf. Bulletin du0 Congrès international des Chemins de ]er, Vol. XX, No. 7, p. 761. Statistics supplied to the International Labour Office by the courtesy of the Ministry of Public Works. - 40 - It should be noted that the American figures include all nontrain accidents, which are partially or totally excluded from the European figures. A similar rough comparison of rates for coupling and uncoupling accidents per thousand railway employees is as follows: RATES PER THOUSAND EMPLOYEES Fatal Continental Europe 0.02-0.15 North America (United States and Canada) 0.10-0.14 Non-fatal 0.03-5.46 1.11-1.91 In this connection it may be well to quote the modes of the rates for continental European countries which were given above as approximately 0.11—0.12 for fatal and 08—1.2 for non-fatal accidents. The rate of coupling.and uncoupling accidents per million trainkilometres is as follows: RATES PER MILLION TRAIN-KILOMETRES Continental E u r o p e N o r t h America (United S t a t e s and Canada) Fatal 0.02-0.18 0.08-0.09 Non-fatal 0.04-4.0 0.97-1.11 The modes of the series of rates for the countries of continental Europe may be given as 0.11—0.12 (fatal) and 1.2 (non-fatal), which are slightly higher than the rates for the American countries. While, therefore, the general accident risk to railway employees is very much higher in America than in Europe — and this position of the American railways as the greatest sources of accidents in the world should be strongly emphasised — coupling and uncoupling accidents there represent a smaller proportion of the total of accidents to railway employees than in continental Europe, while the coupling and uncoupling accident risk, especially in respect of fatal accidents (which are the only comparable factor), on the other side of the Atlantic is equal to or less than that in continental Europe. VI The Occupational Risk of Shunting Operators It is not denied that the frequency rates given throughout this report do not give a true idea of the real risk of coupling and uncoupling; they serve only as a basis of comparison. It would therefore be of vital interest to ascertain the degree of accident risk in the same way as in industrial accident statistics, i.e. the rate of coupling and uncoupling accidents per thousand men exposed to these accidents. The difficulty is that it is not generally possible in railway accident statistics to determine the exposure to coupling and uncoupling accidents, because these accidents are distributed among different groups of employees who often have other duties besides coupling and uncoupling. For instance, the American statistics, which give frequency rates for the different groups of railwaymen, distribute coupling and uncoupling accidents among train men in general, train men in yards, yard train men (switching crew), and others. Shunting operators are not given separately in these statistics. The pre-war statistics of the United Kingdom, however, include a table which is of particular interest. This table, given below, shows the number of men employed by railway companies in certain occupations, and the average proportions of men killed and injured in accidents other than train accidents in which the employment of trains and railway vehicles was concerned, for the ten years ending 1913. The groups of occupations have been arranged in ascending order of fatal accident risk. The following table furnishes eloquent proof of the fact that the occupational risk of shunters is far greater than that of any other group of railwaymen.1 1 It is desirable to indicate the classification in the British statistics of accidents occurring during shunting operations, which is as follows: (1) while coupling or uncoupling vehicles; (2) from coming in contact, whilst riding on vehicles, with vehicles standing on adjacent lines; (3) while passing over, - 42 - AVERAGE PROPORTION OF ACCIDENTS AMONG DIFFERENT CLASSES OF RAILWAY SERVANTS IN THE UNÌTED KINGDOM, 1904 TO 1913 Class of servants Numbers employed on 31 December 1913 Average proportion for the ten years ending with 1913 Killed Labourers Firemen Engine drivers Guards (passenger) Porters P e r m a n e n t way men Labourers ( p e r m a n e n t way) Guards (goods) and brakesmen Shunters 1 1 1 1 1 53,780 27,343 29,869 8,427 61,648 52,517 ) 19,427 j 16,588 15,095 Injured in in in in in 1,620 1,439 1,436 1,403 1,269 1 in 748 1 in 428 1 in 1 in 504 425 1 in 1 in 1 1 1 1 1 in 320 in 41 in 65 in 66 in 94 19 15 Converted to the usual frequency rates, it appears that the number of shunters killed was 2.35 and the number of injured 66.7 per thousand employed. By way of comparison it may be noted that the highest accident frequency rates in the British compensation statistics from 1919 to 1921 are those for miners, 0.93 killed and 105.1 injured per thousand employed, and for dockers, 0.80 killed and 67.1 injured per thousand employed1. This means that the occupational risk of shunters in the case of fatal accidents exceeds that connected with the most hazardous industries, and in the case of non-fatal accidents equals the risk connected with any other occupation except mining. These facts are all the more noteworthy since the percentages and general accident rates presented above have shown that, taken as a whole, coupling and uncoupling accidents are less frequent in the United Kingdom than elsewhere. Similar statistics are not available for other countries. It is, however, interesting to note the following figures for Germany, under, or standing on buffers; (4) from falling off, or when getting on or off engines, wagons, etc.; (5) while tracking, spragging, or chocking wheels; (6) while attending to ground points; (7) while moving vehicles by capstan, turn-tables, props, levers, etc. ; (8) other accidents. 1 HOME OFFICE: Workmen's Compensation Statistics and Proceedings under the Workmen's Compensation Act 1906 and the Employers' Liability Act 1880. The statistics cover all accidents disabling for one week or more. - 43 - giving the percentage distribution of the total number of accidents to railwaymen among the different groups \ PERCENTAGE DISTRIBUTION OF ACCIDENTS TO RAILWAYMEN, BY GROUPS, IN GERMANY, 1921 Grade of Employees Station supervisory staff and telegraphists Pointsmen Shunting staff Higher grade permanent way men Locomotivemen Trainmen Other higher grades Platelayers Goods porters Other lower grades All grades Killed Injured 1.39 5.07 27.01 3.76 12.06 29.81 2.97 9.70 3.50 4.73 100 1.20 3.90 28.20 2.60 13.70 34.10 3.00 7.10 2.50 3.70 100 As the total number of fatal accidents to railway employees in 1921 was 538, and the number of non-fatal accidents (disabling for more than 14 days) 1,283, the number of fatal accidents to shunters was 145 and that of non-fatal accidents 362. As, on the other hand, shunters formed approximately 5 per cent, of the total number of railwaymen, which was 629,080 (excluding administrative staff, but including workshop staff), the number of shunters was about 31,400. This gives accident rates of 2.58 killed and 11.53 severely injured per thousand shunters employed. By way of comparison it may be noted that according to German industrial accident insurance statistics • the highest rates recorded were (in 1920) 5.27 fatal and 6.54 non-fatal accidents (disabling for 13 weeks or more) in the trade association of seamen, and 2.01 fatal and 8.88 non-fatal in the trade association of miners 2. It is thus evident that shunters are exposed to a more serious occupational risk than that in most dangerous occupations, except navigation. In Switzerland it is understood that the average number of shunters during the period from 1913 to 1922 was about 2,500 8. It will 1 Table secured from Germany by the International Transport Workers' Federation (Amsterdam), supplement to the News Letter, No. 4, Apr. 1924. The figures given in this publication correspond with the official statistics. 2 Amtliche Nachrichten des Reichsversicherungsamts, Vol. 38 (1922) No. 1 ; "Unfallversicherung, Nachweisung für das Jahr 1920". 3 Supplement to the News Letter of the International Transport Workers' Federation, No. 4, Apr. 1924. - 44 - be seen from table VIII that the average numbers of coupling and uncoupling accidents were 3.3 fatal and 160 non-fatal. Consequently, the rate of accidents to shunters in the decade concerned were 1.32 killed and 64.0 injured (disabled for more than 6 days) per thousand employed. In the Netherlands the number of shunting operators was 2,250 in 1910 and 4,000 in 1920; the average for the period may be taken as the mean of the two figures, i.e. 3,125. The average number of coupling and uncoupling accidents was 3.4 fatal and 8.3 non-fatal 1 . The corresponding frequency rates are thus 1.09 killed and 2.65 injured per thousand employed. Thus the fatal accident rate per thousand shunters employed was as follows : Germany United Kingdom 2.58 2.35 Switzerland Netherlands 1.32 1.09 The number of shunters injured per thousand employed was as follows : United Kingdom Switzerland 66.7» 64.0 Germany Netherlands 11.5 2.7 The statistics are not complete enough to account for the differences in the above rates, but from the statistics for the United Kingdom and Germany, which may be considered quite reliable, it is evident that the occupational risk of shunting operators is strikingly high, exceeding, especially in the case of fatal accidents, the risk in the most hazardous industries. 1 Supplement to the News Letter of the International Transport Workers' Federation, No. 4, Apr. 1924. CONCLUSION The foregoing chapters embody an attempt to estimate and compare the risk of coupling and uncoupling accidents, as well as the occupational risk of shunting operators, in the several countries of America and Europe. It is now possible to sum up the statistical evidence provided by this report. It has been shown that the actual effects of automatic couplings, when introduced, have been most beneficial to railwaymen. The important change that has been carried out on the United States railways has, in the matter of safety of railwaymen, led to an improvement that may be given as about 70-80 per cent, in fatal and 80-90 per cent, in non-fatal accidents due to coupling and uncoupling. With regard to countries where coupling is non-automatic the importance and degree of accident risk due to the present system of coupling and uncoupling have been analysed at some length. In this respect there is a difference between the United Kingdom and continental Europe. In the former country the risk of coupling and uncoupling accidents, as expressed by the proportion of such accidents to the total accidents to railway employees, and by the rates of coupling and uncoupling accidents in relation to the total number of railwaymen and to the total number of kilometres run by trains during the year, appears to be very considerably lower than in the United States since the introduction of automatic couplings. In most of the countries of continental Europe, on the contrary, the risk in coupling and in uncoupling accidents, as expressed by the corresponding percentages and rates, is higher than either in the United Kingdom or the United States. In view of these facts it is particularly interesting to note, on the basis of statistics for the United Kingdom itself, as well as for certain other countries, that the occupational risk of men performing coupling and uncoupling operations is exceedingly high. It would be a conservative estimate to take this risk generally as - 46 - 1.5-2.5 killed, and 50-60 injured, of which about 10 are severely injured, per thousand shunting operators per year. In conclusion, a rough estimate may be attempted of the actual number of accidents for which the non-automatic system of coupling is responsible in the eleven European countries passed under review in this report. Simply adding the average annual numbers of fatal accidents recorded in the appended tables gives the figure 225. Corresponding calculation of non-fatal accidents would be useless on account of the different standards of reporting; but applying the accident rates for injured shunters given above, an approximate figure may be reached of from 4,500 to 9,000 men injured, and 900 to 1,500 severely injured, in connection with coupling and uncoupling operations 1. These figures may be taken as indications of the strikingly high numbers of railwaymen killed and injured in coupling and uncoupling accidents every year in eleven European countries alone. 1 The method for arriving at the latter figures is as follows. Applying the rates of 1 . 5 - 2 . 5 shunters killed per thousand employed, the total number of men directly exposed to the risk of coupling and uncoupling accidents may be assumed to be between 90,000 and 150,000. Applying the rates of injuries to shunting operators (50 - 60 per thousand employed) the annual average number of men injured by coupling and uncoupling accidents may be estimated at between 4,500 and 9,000. Applying, finally, the rates of severely injured shunters (10 per 1000 employed) the annual average number of severely injured shunters may be estimated at between 900 and 1,500. These figures are only a rough indication. APPENDIX PRELIMINARY N O T E In the following tables detailed information is given on accidents to railway employees, and particularly on accidents due to coupling and uncoupling operations. The countries and the periods covered by these tables are as follows: Table Table Table Table Table Table Table Table Table I. II. III. IV. V. VI. VII. VIII IX. Table Table Table Table X. XT. XII. XIII United States, 1893 to 1907. United States 1900 to 1914. Canada, 1912 to 1920. United Kingdom, 1911 to 1920. India, 1919 and 1920. Belgium, 1909 to 1913. France, 1919 to 1923. Switzerland, 1913 to 1922. Germany, Luxemburg, Netherlands, Roumania, AustriaHungary, 1905 to 1909. Germany, 1912 to 1921. Sweden, 1906 to 1916. Norway, 1913 to 1920. Finland, 1911 to 1920. Each table is accompanied by a brief note on the scope and methods of the series concerned, giving details, where possible, as to (1) railways covered, (2) persons covered, (3) engine or train kilometres or miles, (4) accidents covered, and (5) sources. To the figures drawn from official statistics are added averages for the period covered and accident rates, calculated by the International Labour Office. There are two rates, namely, (1) number of accidents per thousand railway employees covered, and (2) number of accidents per million.engine or train-kilometres or miles. For the Anglo-Saxon countries the latter rate is given on the basis of both train-miles and train-kilometres. TABLE I. — UNITED STATES PROGRESS AND RESULTS OF THE COMPULSORY INTRODUCTION OF AUTOMATIC COUPLING ON UNITED STATES RAILWAYS, 1 8 9 3 TO 1907 Year ending 30 June 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 Number of Total number of vehicles locomotives, provided with wagons, and automatic carriages couplings 1,308,734 1,313,570 1,306,260 1,333,599 1,333,466 1,362,408 1,412,619 1,488,501 1,590,417 1,681,410 1,797,260 1,845,304 1,891,228 2,010,584 322,238 357,621 408,856 545,583 678,725 909,574 1,137,719 1,404,132 1,549,840 1,648,530 1,770,558 1,823,030 1,871,590 1,989,796 — — Number of couplings and uncouplings1 Number of accidents due to coupling and uncoiîpling Fatal Non-fatal 433 251 291 229 214 209 196 228 198 143 253 278 241 311 299 11,277 7,240 8,137 8,457 6,283 5,433 5,281 3,970 2,768 2,113 2,788 3,441 3,109 3,501 3,945 179,636 160,633 157,731 162,876 161,397 170,708 178,851 192,198 209,043 225,422 253,660 253,834 265,235 285,556 295,000 Decrease of coupling and uncoupling accidents since 1893 Number (killed and Percentage injured) — — 4,219 5,282 3,024 5,113 6,068 6,233 7,512 8,743 9,454 8,669 7,991 8,357 7,896 7,460 36 28 25 44 52 53 64 75 80 74 70 71 69 64 i Effected by locomotive men, firemen, conductors, and train men. TABLE II. UNITED STATES ACCIDENTS TO RAILWAY EMPLOYEES, DISTINGUISHING THOSE DUE TO COUPLING AND UNCOUPLING, ON UNITED STATES RAILWAYS, 1900 TO 1914. Year ending 30 June Number of railway employees 1900 1901 1902 1903 1904 1,017,653 1,071,169 1,189,315 1,312,537 1,296,121 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 Train mileage (000 omitted) Total number of accidents Number of coupling and uncoupling accidents Fatal Non-fatal Fatal Non-fatal 888,704 877,000 905,000 959,000 995,000 2,550 2,675 2,969 3,606 3,632 39,643 41,142 50,524 60,481 67,067 228 198 143 253 278 3,970 2,768 2,113 2,788 3,441 1,382,196 1,521,355 1,672,074 1,458,244 1,528,806 1,732,435 1,702,164 1,748,380 1,864,303 1,740,226 1,038,441 1,078,000 1,139,000 1,129,149 1,112,452 1,221,852 1,237,500 1,236,768 1,236,902 1,242,080 3,361 3,929 4,534 3,405 2,610 3,382 3,602 3,635 3,715 3,259 66,833 75,701 87,644 82,487 75,005 95,671 126,039 142,442 171,417 165,212 241 311 299 238 161 206 209 192 195 171 3,109 3,501 3,945 3,113 2,330 2,978 2.966 3,234 3,360 2,692 1905-14 (average) 1,635,018 1,167,214 3,543 108,815 222 3,123 2.17 66.55 0.14 1.91 3.04 93.23 0.19 2.68 2.17 66.55 0.14 1.91 Accident 5 per tho usand employees > (average) Accident 3 per mil' ion trainmiles (average) Accident s per million trainkilome tres (average ) - 49 - NOTE TO TABLES I AND II Railways Covered. All steam railways in the United States. of lines operated: 249,852 miles (1912). Total length Employees Covered. Average number of all railway employees on duty (an employee is considered on duty when he is rendering services or holding himself in readiness at his post of duty) or lawfully on the premises of the carrier. The figure includes shopmen, stationmen, trackmen, bridge and building men, and other industrial employees, as being exposed to non-train accidents (see below). . Train-Mileage. Miles run by freight, passenger, mixed, special, work service (non-revenue), and yard service trains. Accidents Covered. The railway companies are required to report monthly on all accidents occurring in the operation of railways to the Interstate Commerce Commission (Accident Reports Acts of 1900 and 6 May 1910). These reports cover accidents arising from the operation of a railway and resulting in casualty causing the death of a person within twenty-four hours after the accident, or injury to an employee sufficient to incapacitate him from performing his ordinary duties for more than three days in the aggregate during the ten days immediately following the accident. Accidents to railway employees include two groups: (1) Train service accidents, i.e. those arising in connection with the operation of trains, locomotives, or cars that result in casualties to persons. These are subdivided into accidents to employees while engaged in coupling or uncoupling (locomotives and cars, or air hose, steam hose, and safety chains); while operating locomotives, brakes, and switches; from coming into contact with fixed structures; while getting on or off cars or locomotives; at highway grade crossings; and miscellaneous. (2) Non-train accidents, i.e. those connected with the operation of a railway that are not incidental to the operation of trains, locomotives, or cars; accidents occurring in connection with the construction or repair of buildings, tracks, or other equipment; with the movement of vehicles in shops, engine-houses, etc., in industrial work; in the loading and unloading of cars; on lorries or wharves, at stations, freighthouses, engine-houses, shops, or tracks, etc. The figures for the years' 1908-1910 do not cover switch and terminal routes. Sources: INTERSTATE COMMERCE COMMISSION, BUREAU OF STATISTICS: Rules governing Monthly Reports on Railway Accidents — Annual Reports on the Statistics of Railways in the United States. — Quarterly Accident Bulletins; Railroad Accidents in the United States. DEPARTMENT OF LABOUR, BUREAU OF LABOUR STATISTICS: Statistics of Industrial Accidents in the United States (Bulletin No. 339), pp. 28-36. - 50 TABLE III. - CANADA ACCIDENTS TO RAILWAY EMPLOYEES, DISTINGUISHING THOSE DUE TO COUPLING AND UNCOUPLING, ON CANADIAN RAILWAYS, 1912 TO 1 9 2 0 Total number of accidents Year Number of railway employees Train mileage (000 omitted) Fatal 1912 1913 1914 1915 1916 1917 1918 1919 1920 155,901 178,652 159,142 124,142 144,770 146,175 143,493 150,777 155,177 100,930 113,437 107,895 93,218 111,075 115,797 109,857 103,832 117,834 234 324 224 115 174 209 178 174 167 1912-20 average) 152,137 108,208 Accidents per thousand employees (average) Accidents per million trainmiles (average) Accidents per million trainkilometres (average) Number of coupling and uncoupling accidents Non-fatal Fatal Non-fatal 2,924 3,407 3,161 2,573 4,332 4,596 • 5,352 5,432 7,719 18 27 16 9 12 16 9 11 12 208 182 129 78 135 187 182 172 242 200 4,388 15 168 1.31 28.86 0.10 1.11 1.85 40.55 0.13 1.55 1.16 25.35 0.08 0.97 NOTE TO TABLE III Railways Covered. Government and private steam railways. operation: 39,384 in 1920. Mileage in Employees Covered. Railway employees of all categories. Train-Mileage. Not defined. Accidents Covered. Total number of accidents as classified in two groups: (a) accidents resulting from movement of trains, locomotives, or cars, and (b) accidents other than those resulting from the said movement. Accidents due to coupling and uncoupling of carriages form a sub-group in group (a). Details regarding the standard or reporting, etc. are not available. Source: The Canada Year Book, 1918 to 1920. - 51 - TABLE IV. — UNITED KINGDOM ACCIDENTS TO RAILWAY E M P L O Y E E S , D I S T I N G U I S H I N G THOSE DUE COUPLING AND UNCOUPLING, ON THE RAILWAYS OF GREAT BRITAIN Year 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1911-14 (average)1 1911-20 (average)1 Average number of employees of railway companies AND IRELAND, Train mileage (000 omitted) 1911 Total number of accidents Fatal Non-fatal TO 1920 Number of coupling and uncoupling accidents Fatal Non-fatal 5,280 5,533 5,727 5,025 5,120 3,685 . 2,823 2,697 3,834 4,407 12 12 16 13 10 11 6 7 19 15 611 679 687 560 584 312 283 248 492 543 767,796 348,900 378,000 379 336 413 416 407 408 329 292 316 372 625,943 426,675 386 5,391 13 634 648,724 415,480 367 4,413 12 500 0.62 8.61 0.02 1.02 0.93 12.97 0.03 1.53 0.60 8.11 0.02 0.96 608,750 643,135 428,600 412,500 435,300 430,300 Accidents per thousand employees (average 1911-1914) Accidents per million trainmiles (average 1911-1914) Accidents per million trainkilometres (average 1911-1914) 1 FROM TO Where figures are missing, the figure for the latest preceding year is taken. NOTE TO TABLE IV Railways Covered. All steam railways in the United Kingdom. Total route length: 23,734 miles (in 1920). Employees Covered. Approximate number employed by railway companies (contractors' employees are excluded), including those employed on railway premises. This number is only calculated every second year, and was not estimated during the years 1915 to 1919. Train Mileage. Number of miles run by engines on rails, excluding shunting, assisting and line miles. The obligation to report mileage was suspended during the war years 1915 to 1918. Accidents Covered. The railway companies are required under the Regulation of Railways Act 1871 (Section 6) to report on all accidents which occur in the working of railways or on railway premises; the obligation to report was partially suspended during the war years. Moreover, the Public Safety and General Purposes Department of the Ministry of Transport conducts enquiries into certain accidents. - 52 - As regards railway servants, only those accidents which cause the servant injured to be absent for at least one whole day (standard set by an Order of 21 December 1907) from his ordinary work have to be reported. Any person killed in an accident, or so severely injured as to die at any date subsequent to the accident, but prior to the date of the report (about six months after the expiration of the year concerned) is included in the table as "killed". Accidents to railway servants properly so called are classified in two groups: killed or injured (1) from accidents to trains, rolling stock, permanent way, etc., including collisions, derailments, accidents due to failure of engines or rolling stock, fires in trains, and others, i.e. accidents to engine drivers, firemen, guards, and brakesmen; (2) by other accidents in which the movement of trains and railway vehicles was involved, divided into two main groups: (a) accidents occurring during shunting operations, and (b) others (from falling off trains; when getting on and off engines, vans, etc.; from coming in contact with over-bridges, etc.; while attending to the machinery; working on permanent way; walking, crossing or standing on the line; from being caught between vehicles; from falling or being caught between trains and platforms, etc.). The third group, accidents on railway premises not due to train accidents or to the movement of trains and railway vehicles, "are not as a rule attributable to railway working" and are excluded from the total. Accidents "occurring during shunting operations while coupling and uncoupling vehicles" form a heading in group 2. Sources : General Report to the Board of Trade (Ministry of Transport) upon the Accidents that have Occurred on the Railways of the United Kingdom during the Years 1911 to 1920. Returns of Accidents and Casualties as Reported by the Several Railway Companies in the United Kingdom during the years ending 31 December 1911 to 1920. - 53 - TABLE V. — INDIA ACCIDENTS TO RAILWAY EMPLOYEES, DISTINGUISHING ACCIDENTS DUE TO COUPLING AND UNCOUPLING, ON INDIAN RAILWAYS, 1919 AND 1920 Total number of accidents Number of coupling and uncoupling accidents Year ending 31 March Number of railway employees Train mileage (000 omitted Fatal Non-fatal Fatal Non-fatal 1919 1920 684,976 713,078 158,588 162,161 456 450 842 1,023 34 22 109 98 698,577 160,375 453 933 28 104 1919-20 (average) Accidents per thousand employees (average) Accidents per million trainmiles (.average) Accidents per million trainkilometres (average) 0.71 1.34 0.04 0.15 2.82 5.82 0.17 0.65 1.76 3.65 0.11 0.41 NOTE TO TABLE V Railways Covered. Principal railways open for traffic in India (State lines worked by the State, most of the State lines worked by companies, and of companies' lines subsidised by the Government, Indian State lines worked by Indian States, etc.). Total route mileage: 36,735 (1920), of which about 3/4 are covered by the accident statistics. Employees Covered. Servants of all races (European, Anglo-Indians, and Indians) employed on all open railway lines in India. Train Mileage. Total of coaching, goods, mixed, and miscellaneous miles run by trains on all railway lines in India. Accidents Covered. Persons reported as killed or injured in India by the travelling of trains or movement of vehicles used exclusively upon railways. Information on the standard of reporting not available. Source: GOVERNMENT OF INDIA RAILWAY BOARD: Railways in Administration Report for the Years 1918-1919 and 1919-1920, Vol. II. India, - TABLE 54 VI. — - BELGIUM ACCIDENTS TO RAILWAY E M P L O Y E E S , DISTINGUISHING THOSE DUE TO COUPLING AND UNCOUPLING, ON BELGIAN STATE RAILWAYS, 1909 TO 1913. Year Total number of accidents Number of Number trainof railway Severe kilometres employees (ooo Ininomitted) Fatal jury jury Coupling and uncoupling accidents Fatal Severe injury Injury 1909 1910 1911 1912 1913 57,146 58,754 59,902 61,435 62,953 73,709 75,689 77,380 80,224 82,828 42 71 82 72 83 108 117 89 80 98 405 394 406 465 534 5 8 7 3 5 14 9 6 3 4 87 91 113 102 130 1909-13 (average) 60,038 77,966 70 98 441 5.6 7.2 105 1.17 1.63 7.35 0.09 0.12 1.17 0.90 1.26 5.66 0.07 0.09 1.34 Accidents per t h o u s a n d employees (1909-1913). Accidents per million trainkilometres (1909-1913). NOTE TO TABLE VI Railways Covered. All steam railways owned by the State. Effective length: 4,368 kilometres (1913). Employees Covered. Workers employed on the State Railways, excluding administrative staff (fonctionnaires and employés). Train-Kilometres. The figures refer to kilometres run by national and foreign engines in passenger and goods traffic on the lines owned by the Belgian State. Accidents Covered. Accidents which occur in the operation of railways and resulting in death or bodily injury. The non-fatal accidents are divided into two groups, namely, severely injured (blesses) and injured (contusionnés). The total number of accidents includes three groups: (1) accidents exclusively due to fault of the employee injured (about two-thirds of the accidents in 1909-1913); (2) accidents inherent in the particular work of the injured person and thus unpreventable (about one-third of.the accidents in 19091913); (3) accidents attributable to defects in equipment, material, etc. (one accident in 1909-1913). The Belgian statistics give separate figures for coupling and uncoupling accidents, but they are combined in the table. Source: MINISTÈRE DES CHEMINS DE FER, MARINE,POSTES ET TÉLÉGRAPHES. Chemins de fer et service de l'électricité. Marine, postes, télégraphes et téléphones. Compte rendu sur les opérations pendant les années de 1909 à 1913. — 55 — TABLE FRANCE VII. ACCIDENTS TO RAILWAY E M P L O Y E E S , COUPLING AND UNCOUPLING, DISTINGUISHING ON THE FRENCH THOSE DUE TO RAILWAYS, 1919 TO 1923 Year 1919 1920 1921 1922 1923 Total of severe accidents Severe coupling and uncoupling accidents Fatal Non-fatal Fatal Non-fatal 417 336 344 411 387 712 702 559 678 615 39 47 36 31 34 111 119 95 102 109 379 653 37 107 1919-1923 (average) NOTE TO TABLE VII Railways Covered. The principal steam railways of France, i. e. lines of the State, the Ceinture, Nord, Orléans, Midi, Paris-Lyon-Méditerranée, Est, Alsace-Lorraine. Total length: 41,181 kilometres (1912). Employees Covered. All railway servants in service. Figures not available. Train-Kilometres. Kilometres run by passenger, mixed, goods, ballast, and service trains. Figures not available. Accidents Covered. The statistics only cover "severe accidents", i. e. (1) those occurring on principal lines to trains or vehicles and resulting in death or injury to persons; (2) those sustained by persons and causing incapacity for work for more than 20 days. Accidents not connected with the working of trains properly so called, and all accidents occurring on railway premises, are excluded. Source: Information supplied by courtesy of the Ministry of Public Works. - 56 - TABLE VIII. — SWITZERLAND ACCIDENTS TO RAILWAY E M P L O Y E E S , DISTINGUISHING THOSE D U E TO COUPLING AND UNCOUPLING, ON ALL SWISS RAILWAYS, 1913 TO 1922 Year 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 Total number of accidents Number of coupling and uncoupling accidents Number of railway employees covered Number of enginekilometres (000 omitted) Fatal Non-fatal Fatal Non-fatal 30,313 27,168 28,168 28,096 28,117 28,130 30,077 31,773 30,767 29,753 56,760 51,411 44,984 45,186 36,650 26,956 28,369 35,486 36,172 38,336 48 37 29 25 20 16 22 30 27 33 1,688 1,448 1,347 1,445 1,639 1,194 1,307 1,382 1,059 1,168 7 8 2 1 3 2 5 1 2 2 183 182 136 145 191 148 157 191 122 140 29,836 40,031 29 1,368 3.3 160 0.97 45.85 0.11 5.46 0.72 34.17 0.09 4.00 1913-22 (average) Accidents per thousand employees (average) Accidents per million enginekilometres (average) NOTE TO TABLE VIII Railways Covered. Swiss Federal and private, normal-gauge and narrowgauge lines, as well as rack-and-pinion lines. Length (in 1921): normalgauge lines, 2,808 kilometres Federal and 885 private ; narrow-gauge lines, 74 Federal and 1,520 private; rack-and-pinion lines, 117 kilometres. Tramways (496 kilometres) and funicular railways (49 kilometres) are excluded. Employees Covered. Workers classified under the headings: (1) maintenance of permanent way and line-guarding departments ; (2) station and train staff. Excluded: administrative staff, locomotive, workshop, and auxiliary departments. Engine-Kilometres. Number of kilometres run by national and foreign engines, including electric engines, on Swiss lines. Accidents Covered. Accidents occurring in employment properly so called, excluding those occurring in auxiliary work, interior office service, workshops, construction, etc., and resulting in death or incapacity for work for more than 6 days. Accidents are classified in two groups: (1) collisions and derailments of trains; (2) other train service accidents, which are subdivided under 60 headings according to cause, of which coupling and uncoupling of vehicles is one. Only suicide and attempted suicide are excluded. Sources: DÉPARTEMENT FÉDÉRAL DES POSTES ET DES CHEMINS DE FER: Statistique des Chemins de fer suisses pour les années de 1913 à 1922. Information supplied by courtesy of the Federal Department of Posts and Railways. - 57 - TABLE IX. GERMANY, LUXEMBURG. NETHERLANDS, AUSTRIA-HUNGARY ROUMANIA, ACCIDENTS TO RAILWAY E M P L O Y E E S , D I S T I N G U I S H I N G THOSE D U E TO COUPLING AND UNCOUPLING, ON GERMAN, LUXEMBURG, DUTCH, ROUMANIAN, AUSTRIAN AND HUNGARIAN L I N E S BELONGING TO THE UNION OF GERMAN RAILWAY BOARDS, 1 9 0 5 TO 1 9 0 9 Country Germany Year Coupling and of Total number of Number Number train uncoupling accidents of kilometres accidents (000 railway Fatal Non-fatal Fatal Non-fatal employees omitted) 1905 1906 1907 1908 1909 574,434 615,366 660,469 663,816 655,358 588,164 625,480 667,300 665,560 663,750 600 655 714 599 531 1,457 1,526 1,709 1,474 1,372 97 105 118 83 87 212 232 235 213 192 Average 633,889 642,051 629 1,508 98 215 0.99 2.38 0.15 0.34 0.98 2.35 0.15 0.33 Accidents per thousand employees Accidents per million trainkilometres Luxemburg Netherlands Roumania 1905 1906 1907 1908 1909 66,281 71,058 72,659 73,299 75,118 64,734 67,544 70,921 70,995 71,695 65 80 86 56 68 145 171 182 162 177 5 3 14 3 3 16 15 20 17 24 Average 71,683 69,258 71 167 5.6 18.4 0.99 2.33 0.08 0.26 1.03 2.41 0.08 0.27 Accidents per thousand employees Accidents per million trainkilometres AustriaHungary (including BosniaHerzegovina) 1905 1906 1907 1908 1909 303,264 316,372 341,570 382,255 388,480 225,348 238,754 258,200 272,900 282,590 159 224 278 273 283 1,015 1,528 1,996 2,277 2,272 24 37 36 39 41 138 210 291 347 347 Average 346,388 255,558 243 1,818 J35.4 267 0.70 5.25 0.10 0.77 0.95 7.11 0.14 1.04 Accidents per thousand employees Accidents per million trainkilome tres - 58 - NOTE TO TABLE IX Railways Covered. The German, Luxemburg, Dutch, Roumanian, Austrian, and Hungarian State Railways, and a number of private railways in these countries belonging to the Union of German Railway Boards ( Verein Deutscher Eisenbahnverwaltungen), which published, before the war, statistics of its own. The length of both normal and narrow-gauge lines covered by these statistics was (in 1909) as follows: A. Germany (State and private) kilometres 57,228 B. Luxemburg, Netherlands, (State and private) Nordabant-Deutsche (private), Chimay line (private), Roumania (State) and WarsawVienna line (private) 7,293 C. Austria (State and private) Hungary (State and private), common (private), and BosniaHerzegovina (military line) 41,206 Total for the Union 105,727 Employees Covered. Officials and workers in track maintenance and lineguarding; station train despatch; train, locomotive, and workshop departments; administrative staff excluded. The figures give the number of fulltime employees, which is equal to the number of days worked divided by 365 for officials on daily pay and for employees in train service, and equal to the number of work-days divided by 300 for workers and helpers in line-guarding, workshops, etc. Train-Kilometres. Number of kilometres run by national and foreign engines on national lines,-including electric engines and non-traffic trains, but excluding shunting. Accidents Covered. All accidents compensated by the accident insurance institutions in the respective countries. The total number of accidents to railways employees includes the following groups: (1) train accidents (collisions and derailments); (2) incautious getting on or off vehicles in motion; (3) shunting of vehicles and formation of trains ; (4) coupling and uncoupling of vehicles ; (5) loitering or trespassing on railways ; (6) miscellaneous. Since 1910 no details as to accidents to railway employees have been given. Source : Statistische Nachrichten von den Eisenbahnen des Vereins Deutscher Eisenbahnverwaltungen fur die Rechnungsjahre 1905-1909. - 59 - TABLE X. — GERMANY ACCIDENTS TO RAILWAY EMPLOYEES, DISTINGUISHING THOSE DUE TO COUPLING AND UNCOUPLING, ON GERMAN RAILWAYS, 1912 TO 1921 Year 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 Average number of railway employees Number of enginekilometres (000 omitted) Total number of accidents Fatal Number of coupling and uncoupling accidents Non-fatal Fatal Non-fatal 743,944 786,466 764,028 735,254 759,589 820,462 922,355 1,132,185 1,105,557 1,067,437 794,179 814,800 674,668 682,652 677,205 629,136 526,222 424,682 471,179 531,423 682 747 633 704 900' 1,215 1,025 761 589 538 1,457 1,406 1,181 1,315 1,704 2,223 1,596 1,477 1,209 1,283 120 120 86 118 124 151 177 180 109 81 217 170 146 160 217 267 205 235 188 151 883,728 622,615 779 1,485 127 196 0.88 1.68 0.14 0.22 1.25 2.39 0.20 0.31 1912-21 (average) Accidents per thousand employees (average) Accidents per million enginekilometres (average) NOTE TO TABLE X Railways Covered. German normal-gauge lines; length: 59,110 kilometres (State) and 3,578 kilometres (private) in 1917. Employees Covered. Total of railway officials and workers, i.e. (1) administrative staff; (2) track maintenance and guarding staff; (3) station train despatch, and train staff; (4) engine-house and workshop staff. The number of railway employees is an annual average, which is equal to the number of days spent in service divided by 365 for employees on daily pay and for workers in train-service, and to the number of work-days divided by 300 for workers and helpers in line guarding, workshops, or other departments. Engine-Kilometres. Number of kilometres run by national and foreign engines including electric engines, on national lines in traffic service (Nutzkilometer). Accidents Covered. Accidents occurring in railway service, excluding workshops, and resulting either in death within twenty-four hours after the occurrence of the accident, or in bodily injury causing incapacity for work for more than 14 days. The classification of accidents is as given in the note to table IX. Unfortunately the German statistics published since 1918 have combined groups (3) and (4). Source: Statistik der im Betriebe befindlichen Eisenbahnen Deutschlands nach den Angaben der Eisenbahnverwaltungen, compiled by the Reichs-Eisenbahnamt from 1880 to 1918, and by the Reichsverkehrsministerium since 1918. - TABLE 60 - SWEDEN XI. ACCIDENTS TO RAILWAY E M P L O Y E E S , DISTINGUISHING THOSE D U E TO COUPLING AND UNCOUPLING, ON SWEDISH STATE RAILWAYS, 1906 TO 1916 Total number of accidents Number of coupling and uncoupling accidents Averageof number railway employees Number train- of kilometres (000 omitted) Fatal Non-fatal Fatal Non-fatal 1906-10 (average) 1911 1912 1913 1914 1915 1916 18,373 18,771 23,730 24,918 25,334 26,631 36,818 21,960 22,884 24,170 25,330 26,110 30,169 35,378 20 13 27 22 15 35 39 99.6 94 118 131 137 210 211 3.2 1 3 6 2 4 5 15.2 26 30 26 34 37 54 1906-16 (average) 25,202 26,572 24.4 143 3.5 31.7 1.0 5.64 0.14 1.25 0.88 5.16 0.13 1.19 Year Accident s p e r t h o u s a n d employee 3 (average 1906-1916) Accident s per million train-kilometres (average 1906-1916) NOTE TO TABLE XI Railways Covered. State railways, excluding lines on concession as well as private lines run by the State. Length of railways covered: 4,986 kilometres in 1916. Employees Covered. Railway employees on ordinary, special, and casual rolls, excluding employees in railway offices, workers in railway workshops and warehouses, and employees on steam ferry-boats. Average number of employees, i.e. number of days worked divided by 300. Train-Kilometres. Number of kilometres run by national and foreign locomotives on national lines in traffic service, including auxiliary kilometres. Accidents Covered. The total number of accidents includes those due t o : (1) derailment or collisions of trains; (2) getting on or off vehicles in motion; (3) shunting and formation of trains; (4) coupling and uncoupling of vehicles; (5) incautious trespassing on rails; (6) falls from trains; (7) other circumstances in train service. Accidents not connected with operation of trains and vehicles (accidents on railway premises) are excluded. Source: KUNGL. JARNVAGSSTYRELSEN: Statens järnvägar är 1916. officiella Statistik, Järnvägar, Stockholm, 1918. Sveriges — 61 — TABLE XII. — NORWAY ACCIDENTS TO RAILWAY E M P L O Y E E S , DISTINGUISHING THOSE DUE TO COUPLING AND UNCOUPLING, ON NORWEGIAN STATE RAILWAYS, 1913 TO 1920 Year ending 30 June of Number of Number enginerailway kilometres employees (000 omitted) Total number of accidents Fatal 1913 1914 1915 1916 1917 1918 1919 1920 4,296 4,591 4,675 4,756 5,114 5,471 6,021 6,891 10,610 11,141 10,951 11,898 11,102 9,644 10,327 11,671 1 1913-20 (average) 5,227 10,918 Accidents per thousand employees (average) Accidents per million enginekilometres (average) Non-fatal N U mber of coupling and | uncoupling accidents Fatal Non-fatal 1 4 4 6 4 2 4 3 6 12 19 31 41 24 47 16 16 1 5 9 8 3 3 3 3.25 25.8 0.6 4.9 0.62 4.92 0.12 0.93 0.30 2.38 0.058 0.45 1 1 1 NOTE TO TABLE XII Railways Covered. State railways only. Length: 2,821 kilometres in 1920. Private railways (length, 466 kilometres) are excluded. Employees Covered. The regular staff employed on State railways, grouped in the following categories: (1) station staff; (2) engine and train staff; (3) line maintenance, guarding, and cleaning staff. Administrative staff and shopmen are excluded. Engine-Kilometres. Number of kilometres run by national and foreign engines on national lines, excluding trains run for the maintenance of lines. Accidents Covered. Accidents to railway employees on duty occurring in running traffic trains and extra trains, and during shunting operations, and resulting in death within twenty-four hours after the occurrence of the accident or in injury necessitating medical treatment. The total number of accidents- to railway employees includes the following groups: accidents occurring (1) on trains in motion; (2) while getting on or off vehicles; (3) during formation of trains, (a) by coupling, {b) by uncoupling, (c) others; (4) when trespassing on railways; (5) miscellaneous. Source: HOVEDSTYRET FOR STATSBANENE: De offentlige Jernbaner — Norges Jernbaner. Beretning for ärßt 1 juli-30 juni, 1912-1913 to 1919-1920. (Norges offisielle statistikk.) — 62 — TABLE XIII. — FI NLAND ACCIDENTS TO RAILWAY E M P L O Y E E S , D I S T I N G U I S H I N G TO COUPLING AND UNCOUPLING, IN Year Number of railway employees Total number of accidents Number of trainkilometres (000 omitted) Fatal Non-fatal 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 14,815 14,783 15,544 16,711 20,393 23,405 23,008 19,139 20,517 24,105 13,821 14,255 14,503 13,643 17,312 19.528 18,333 9,418 13,927 16,035 10 10 7 7 17 22 19 5 11 6 22 28 25 49 37 78 44 17 22 22 Average 19,242 15,078 11.4 34.4 Accidents per thousand employees (average 1911-1920) Accidents per million trainkilometres (average 1911-20) NOTE THOSE DUE F I N L A N D , 1 9 1 1 TO 1920 Numberof couplingand uncoupling accidents Fatal Non-fatal 2 2 1 2 2 0.3 0.6 0.59 1.78 0.016 0.03 0.76 2.28 0.02 0.04 TO TABLE XIII Railways Covered. State Railways open to traffic. Total traffic length 3,990 kilometres in 1921. Employees Covered. The whole staff employed on State Railways. Train-Kilometres. Total number of kilometres run by national and foreign trains on national lines. Accidents Covered. Total number of reported accidents to railway employees, including collisions and derailments, but excluding accidents in workshops. Source : Information supplied by the courtesy of the Statistical Office of the Railway Board. Suomen Rautatiehallituksen kertomus vuodelta 1920 (Suomen virallinen tilasto XX. Rautatietilasto No. 50.) Helsingfors, 1923.