whole purpose of the classification of accidents by causes is accident prevention. The committee, therefore, in its definition of cause of accident adopted the recommendation that the accident should be charged to that condition or circumstance the absence of which would have prevented the accident; but if there be more than one such condition or circumstance, then to the one most easily preventable. The question of remote initiating cause or of personal responsibility was ignored as in most cases impossible of determination and after all often a matter of personal judgment. The immediate cause as defined by the committee is the tangible fact capable of definite ascertainment. To go further would be to venture into the field of speculation, where personal bias would too often determine the result.

The committee's original classification of causes was adopted in 1916, but was revised two years later upon the experience of members of the committee representing a considerable number of state commissions. The present classification includes eleven general headings and several hundred sub-headings, in addition to a classification of power work machines with about three hundred sub-headings.

The committee's classification of causes does not include the desirable sub-headings under the general head Poisonous Substances and Occupational Diseases. This is due to the lack of experience data which will permit the preparation of the desired lists.

Classification by location and nature of injury refers to the injury sustained at the time of the accident, as distinguished from the result or consequences of the accident. The location of injury classification follows the common anatomical divisions, beginning with the head and ending with the feet. Rules of practice were also adopted to guard against a variety of interpretations.

The classification of extent of disability refers merely to the classification into temporary total or partial disability, permanent total or partial disability, and fatal, distinction being made between those permanent disabilities which are dismemberments and those which result in impairment of function.

The classification of degree of partial disability is for the purpose of giving a measure of the partial disability where the character of the compensation act permits. Under some of the laws, however, where the compensation is based on loss of earnings,

no determination of the degree of partial disability is ever made. For the purposes of accident prevention, it is particularly important that there should be tabulations permitting some measurement of accident hazard, and for checking up the effectiveness of accident prevention measures. Tables showing the frequency of accidents have been commonly used for this purpose. Attempts to compare the hazards of different industries, or of the same industry at different times, have been somewhat unsatisfactory because of the lack of a definite basis of comparison. Two chief difficulties present themselves: (1) the lack of an accepted rule as to what accidents shall be counted, and (2) the lack of a uniform and adequate basis for the computation of accident rates.

The adoption of the standard definitions already referred to will readily remove the first difficulty. The determination of a uniform satisfactory unit of exposure as the basis for measuring the frequency of accident occurrence presents a much more serious question. The common method of compiling accident statistics was to express the frequency rate per year in terms of number of accidents per one thousand workers. This term was usually understood to be based on average number of workmen, which investigation showed to mean sometimes the usual or normal number, sometimes the average number, arrived at by an off-hand guess, sometimes the average number based upon various methods of computation. Usually no account was taken of the number of days or hours in the year during which the workmen were actually exposed to risk of accident. Of two plants showing the same accident rate on paper, one might represent an exposure of 365 days of 10 hours or 365,000 hours, the other 240 days of 8 hours each or 192,000 hours. With such a method of computation, accident rates were indefinite and meaningless because of the indefinite and variable factors.

The method recommended by the joint committee of the International Institute of Statistics and the International Congress on Industrial Insurance and Accidents, which has already been referred to, provided for a unit of exposure in terms of so-called full time workers, that is, men working 300 days of 10 hours each per year, or 3000 hours. This plan was familiar to members of the committee from its use in German and Austrian accident reports. The plan had also been used by the U. S. Bureau of Labor Statistics in its studies of accidents in the iron and steel and machine building industries, where the number of man hours

worked per year had been ascertained from accurate records in each establishment. This unit of exposure, it should be understood, is an arbitrary one and was not intended to suggest a ten hour day or a three hundred day year as the actual or ideal working day and working year in industrial plants. It is merely a unit of measure for the purpose of ascertaining accident frequency.

This unit of measure of the three hundred day worker was recommended for adoption by the committee in 1915, and was used to a considerable extent in accident statistics. Some large industrial plants adopted the method for use in tabulating their individual records for purposes of accident prevention. However, a certain amount of criticism arose, due to some dissatisfaction on the part of both employers and employees with this unit of measure. It was objected that the three thousand hour worker implied some judgment regarding the proper length of the working day and the working year. It was pointed out that the eight hour day was gradually becoming the standard working day, and that a unit of measure of twenty-four hundred hours per year would more nearly reflect the working time in some industries. The two thousand hour year was also suggested as more nearly representative for a full time year in the coal mining industry. The committee, thereupon, after careful consideration of these objections, recommended a change in the unit of measure from the three thousand hour year to a thousand hours of exposure, and at its meeting in December, 1919, recommended that both accident frequency and severity rates be computed on the basis of one thousand hours exposure instead of three thousand hours exposure, as heretofore. The committee pointed out that the thousand hour exposure was an acceptable scientific mathematical unit of measure to which any other unit of exposure was readily convertible.

This method furnished an accurate measure of the frequency of accident occurence. The true measure of the hazard of the industry, however, is not expressed by the mere number of accidents per thousand hours' exposure. Entirely different hazards might be expressed by the same accident frequency rate, since in such rates the most trivial lost time accident is given quite as much weight as a death or permanent total disability.

Fully 95 per cent of all lost time accidents leave no permanent impairment of earning capacity, and a great majority of them

cause only a few days' disability. A single death results in more economic loss than several hundred average temporary disabilities. Furthermore, no fixed relation exists between the number of deaths and of permanent and temporary disabilities in various industries. While deaths are infrequent in the textile industries, and in many of the lighter machine trades, they are many times more frequent among coal miners, railway trainmen, seamen, and on construction work. Accident rates which do not take these differences into account are inaccurate and misleading.

The committee, accordingly, sought some method of reducing temporary disabilities, permanent disabilities, and deaths to some common denominator expressive of economic loss or industrial hazard. The time loss was readily fixed upon as the most significant, stable, and convenient expression. It was naturally suggested, since the temporary disabilities, which constitute 95 per cent of all accidents, are of necessity measured in the record and compensated on the basis of days lost. Furthermore, no other measure seems readily available. The physical results could hardly be reduced to a common denominator. The compensation cost would not serve the purpose, as it would vary greatly from state to state, and change with amendments to the law. The wage loss, likewise, would vary from locality to locality and from time to time. The time loss is apparently the only available unit of measure which is both significant and stable.

For the purpose of expressing the hazard indicated by deaths and permanent disabilities in terms of time loss it was assumed that the injured workman of average age, if not disabled, would have had a working life expectancy equal to about two-thirds of his full life expectancy as shown by the American Experience Table. The approximate average age of workmen fatally injured was found to be thirty-three years and a reasonable time loss equivalent was fixed at twenty years, or six thousand working days. An equal time loss was assigned to permanent total disability, while proportionate losses were assigned for permanent partial disabilities upon a percentage rating of degree of disability expressive of the judgment of the committee after a study of foreign and American disability schedules and American experience.

The use of this table of time loss equivalents for weighting deaths and permanent disabilities to show accident hazard permits the computation of a single rate expressive of accident severity in terms of days lost per one thousand hours' exposure. It is thus

relatively easy to compare accident experience from year to year, from industry to industry, and from state to state. The method is equally suitable for use in any establishment with a sufficient number of employees to make a rate significant and has been so used.

The scale of weighting recommended is intended merely as a standard for comparing severity of accident injuries and the accident hazards of employment, but is not set up as a basis for awarding compensation.

The sixteen standard tables and the standard method of comparing compensation costs recommended by the committee need only be mentioned here. Three of the tables recommended-those relating to accident frequency and severity rates, and that designed to compare time losses resulting from the various causesare equally useful for a state jurisdiction or for the individual employer, and have been so used. Some of the other tables, while intended primarily to assist in analyzing experience for ascertaining costs, determining premium rates, and other administrative purposes, are applicable to individual uses.

The Committee on Statistics is continuing its work, for it is in no sense ended. A solid foundation has been laid which, it is believed, will stand without important changes (except perhaps such as are to be expected in the classification of industries).

The usual difficulties between acceptance in principle and adoption into full operation will necessarily make progress in standardization somewhat slow. Commissions are obliged to take the position that the adjustment and payment of claims must take precedence over any other work. In most states, therefore, additional appropriations must be made available before material changes or expansion of statistical work is possible.

A considerable number of commissions have made a beginning in putting the committee's recommendations into effect and year by year better and more adequate statistics are put forth, including valuable special studies. The questions concerning proposed amendments to the laws and their probable effect, with which commissions and legislators are constantly confronted, will gradually force a continuous study and analysis of the facts of actual experience in a systematic manner-which is the purpose of all statistical work.