In many instances the chain of causes appears to return upon itself, the effect becoming a cause and forming, in a sense, a "vicious circle." A man is destitute because of unemployment, unemployed because of physical weakness, weak because of nature of abode and insufficiency of nutritious food, confined to insanitary dwelling-places and poorly fed and depressed because of his destitution. What point in this circle should be selected for tabulation? Obviously, the latitude that must be given to individual judgment puts agreement out of question. The individual judgments, though issuing in a numerical statement with all the delusive aspect of exactness that "figures" give, are the outcome of a struggle in the investigator's mind between indeterminate factors. Prejudice or, in more courteous phrase, the "personal equation" decides what part of a circular causal movement will be designated or how far back of the immediate causes the inquiry will be pushed. Moreover, the investigator's judgment is not likely to be entirely undisturbed by strong feeling. In regard to the subject of intemperance, for instance, an American until recently found it difficult to think calmly. He abominated either the liquor traffic or the frenzied language of the temperance advocate. That the bias of one student will exactly offset that of another is by no means certain.* The nature and amount of popular discussion given to any factor largely determines the bulk it assumes in each investigator's mind. Intemperance as a cause of destitution cannot be disregarded. "Nature of abode," however, has been overlooked. The small percentage of cases attributed to it (2.2 and 1 at the highest in Warner's and Lindsay's tables, respectively, and nothing for most cities) will astonish any one acquainted with the life of tenement districts. The attention given to housing conditions in recent years and the impression made upon the public mind will possibly be reflected in future tabulations, and result in some approach to a correct

* Professor Warner refers to the tendency of bias to become corrected in grand totals, and then suggests that variation is, after all, rather slight. Thus "causes indicating misconduct vary only between 7.5 and 32.5." This only is almost ludicrous. If variations of 25 points are to be regarded as trivial, the use of numerical data can have little meaning.

percentage. Such tables are, however, at best but records of the more or less confused impressions and reactions of more or less prejudiced minds. They indicate the course of popular thought and feeling, of subjective rather than of objective conditions. They can never give a rightly proportioned picture of the facts. How widely they depart from the reality it is idle to inquire.

A statistical blank such as that formerly recommended by the National Conference is, by its very nature, misleading. Consisting of a list of causes, headings under which data are to be entered, its use can be defended only on the assumption that the list is exhaustive so far as causes that can be considered principal are concerned, or that the omitted causes are a negligible quantity. Such listing leads to the oversight of causes not listed. The caption "Causes not named or unknown" is an insufficient safeguard. Placed at the end of the list, it attracts too little notice. Attention is inevitably centred upon the more conspicuous headings. The causes indicated by them occupy the mind at the outset of the inquiry, and, as in all human observation, what is expected and, therefore, almost desired, is, of course, discovered. Nothing would be gained, however, if each student were to make his own list of causes. Uniformity of terminology (that is, a statistical blank) is necessary if there is to be any tabulation of the combined results of different investigators. The very means necessary to a statistical study, however, renders it impossible to indicate the true proportions of all the factors creating destitution. A blank form, therefore, should not be used in a study of the entire subject. At best, a blank could exhibit the relative prominence-not relative importance of a small number of causes. For such a purpose the proposed blank of the National Conference could be adapted by the omission of causes that are inconspicuous.* The difficulty of securing accurate and complete data and of interpreting

* Dr. Ayers states that the committee of the Conference revising the statistical blank in 1898 was guided by the opinion that those headings in the old form that had shown only very small percentages should either be omitted or joined with other headings."

them, however, makes it clear that even the attempt to compare only a few conspicuous causes can lead to no trustworthy conclusions.

The case-counting method, we may conclude, is, and always will be, a complete failure. Comparing causes, not by any possible accurate measure of their effects, but by counting cases and assuming that they are of equal value; taking into consideration only a small proportion of the total of cases without assurance that those taken are truly representative of the whole; based for the cases actually investigated on incomplete data and on the untenable assumption that a principal cause may be picked out from a tangled network of causes; limited artificially to a printed list of causes suggested by the dominant interests of the time in advance of any real investigation,-it throws light neither on what is fundamental nor on the relative importance of the superficial phenomena and it gives no adequate study of any distinct class of the destitute or trustworthy comparison of any special groups of causes. It follows, as a matter of course, that it can yield no conclusion of sufficient generality to be of service in establishing principles of poor-relief. A few factors, indeed, in such a study acquire prominence,—intemperance, matters of employment, sickness. That these are in numerous cases among the immediate causes of distress may be granted. General observation will lead to that conclusion, and statistical proof is unnecessary. The only legitimate inference, however, is that of all the forces at work they are the conspicuous ones. It is not proved that they are the most important.

Warner appears to draw a general conclusion from his statistical study by selecting from the list of causes those of large percentages (intemperance, sickness, etc.) and showing that these imply weakness, physical, mental, or moral. He concludes that "the commonest exciting cause of the poverty that approaches pauperism is incapacity. Weakness of some sort is the most typical characteristic of the destitute classes." Almost every cause in the entire list, however, whether numeri

cally prominent or not, could, by a similar method of reasoning, be shown to rest on weakness. The numerical data, therefore, are not necessary to the conclusion. Given merely a list of causes, such as may be obtained by general observation, an analysis of the list will show personal weakness as an allpervading element. Objection may be made to Warner's conclusion, moreover, not only on the ground that it is reached by unnecessary and faulty statistical operations, but that it is incomplete. Analysis carried further will show that weakness is relative to the forces with which personality must contend; i.e., the environment. The fundamental factors could be grouped under "personality" and "environment," but which group is the more important it is not possible to determine.*

The writer does not wish to imply that all study of concrete cases of destitution is without value. It is to the attempt

* Some publications of later date than those considered above indicate that doubts are beginning to arise in regard to the value of the case-counting method. Professor T. S. Adams (Adams and Sumner, "Labor Problems," 1905, p. 150), citing Warner's data, remarks that such figures can never reveal the fundamental or original causes of poverty, and they must be used with great caution." Professors Mayo-Smith and Giddings, in reporting as a committee in 1899 on an investigation of this kind, say: "The committee does not lay any particular stress on the results of this inquiry. The question seems to be too difficult for the rough method of statistics." (Seventeenth Annual Report of the C. O. S. of New York City, p. 61.) A report of the Committee on Social Research of the New York Charity Organization Society (Twenty-third Annual Report, p. 86) declares a study based on opinions or impressions of charity organization society agents in regard to principal causes of poverty to be "unscientific." "It is felt that much more valuable results will follow from a study of actual conditions, such as sickness, lack of work, and drunkenness. . . . By carefully noting the associated conditions in every case, and tabulating the number of instances of such associated conditions, it will be possible to arrive, in the course of time, at a knowledge which shall be based on facts, and not on impressions." In future reports the committee hopes to establish "correlations" between "conditions and circumstances," which will form a basis for conclusions as to causes." Here appears a lingering hope that statistical study may yet lead to conclusions. In recent years the Committee on Statistics of the National Conference of Charities and Corrections has not countenanced the case-counting method. "Most of us," writes one member, "have come to see that the old method of studying causes of poverty was an unfortunate one. . . . It was unfortunate for two reasons: first, because it meant reliance on opinions, not on facts; and, second, because the burden of deciding whether it was intemperance, lack of work, unwise philanthropy, inefficiency, or illness in a given case that brought the family to dependence, and the conviction that the decision could not be of much value, did much to make statistics in general hateful to charity workers." (Lilian Brandt in Proceedings of the National Conference, 1906, p. 426.) The schedule drawn by the committee and adopted at the Minneapolis meeting in 1907 did not use the old blank as a model. In fact,

it was merely "designed to exhibit the general activities of societies dealing with needy persons in their homes," and calls for such data as number of cases dealt with, expenditures, number of volunteer workers, etc.

to measure causes that objection is made. If not cribbed and confined by a statistical schedule to specifically listed causes, it is possible that the study of concrete data may lead to the discovery of neglected factors or causes as yet unknown. A detailed study of individuals or families, a monographic method like that of the Le Play School, may prove useful. In providing a starting-point for an effort to trace causal antecedents in the chain and network of forces, such studies may contribute more to the satisfaction of the scientific temperament and to the illumination of the problem of poor-relief than any attempt at the measurement of the area over which given causes act. When we consider how causes are interlaced, how interdependent and inseparable are individual and environment, racial trait and social structure, it appears impossible to make any statement of the importance of causes in terms of relative magnitude. The only general conclusions of value, in the opinion of the writer, that can be drawn from a study of the causes of poverty, refer to the permanency or removability of the different forces at work. Nothing is really permanent, but what in a comparative sense may be called permanent factors constitute fixed points, hard facts, to which the practice of poor-relief must accommodate itself. The removable, on the other hand, should be studied with reference to the method, and perhaps in some cases the desirability, of removing. This is the point of view suggested, when it is borne in mind that, after all, the chief motive in a study of the subject is the desire to find guiding principles in the relief of human suffering.