tributed by periods of consumption. Bare amounts paid for rental tell little unless something can be known of the accommodation secured. Clothing must be distributed by kinds and by periods of service,-a very difficult matter. The purchasing power of money ought to be taken into account, though it usually is not.

At last the schedule is planned, including many items the investigator has no direct or special use for at the time, but that he thinks are "interesting" and likely to throw unexpected light on the social problem when gathered, and now he goes forth to collect his information.

Difficulty number two is promptly met with,-that of the possibility of access to the information desired. He cannot, like the astronomer, make first-hand observation of his material which can be verified by other first-hand observers, but is obliged to accept the statements of other people-his very objects of study-who may be, and usually are, of differing degrees of intelligence, of willingness to tell the truth, and of differing personal bias which will lead some to understate some things and others to overstate them, and vice versa. They, in turn, are not giving information entirely at first hand, but must depend, more or less, upon memory for matters that puzzle anybody to recall definitely. And, finally, the investigator has to translate their concrete contributions of items into his " units," and present them in such shape and semblance of order as to afford ground for some definite conclusion about something large or small.

Now the third difficulty is encountered. Owing to the first two drawbacks, the scope of a budget investigation ought to be especially large to eliminate errors and give truly representative details. As a matter of fact, it is usually, on account of the time and labor necessary for its preparation, especially small. Data are usually gained for from one hundred to five hundred families,—a range entirely inadequate as a basis for conclusions.

In a recent investigation where it was frankly acknowledged that the number of families investigated (two hundred) was

too small to "furnish conclusive deductions for all workingmen's families," comparison was made with broader investigations to prove the general correctness of the one in question. It might be urged, however, that, where the small investigation is merely confirmed by the large one, it is unnecessary, and, where it is not confirmed, it is too small to allow us to determine whether it is showing new and typical phenomena or simply individual variations due to scarcity of examples.

Where it is found impossible (as it usually is) to make the complete enumeration of facts that would give the whole truth for a class, for workingmen in a given country, for instance, at a given period, or for a given income class, or race, or neighborhood, the investigator has to fall back upon a selection of examples that again works confusion.

In the investigation above referred to, the two hundred families studied were found in one special neighborhood in New York City, but we are nowhere given definitely the principle of selection whereby, as the preface claims, "they were sorted out by a method as sound as could possibly be chosen," and consequently are unable to judge, except from the word of the investigator, whether the results of the investigation do or do not "give a comparatively true insight into the social, economic, and industrial life of a large class of workingmen's families in any city neighborhood of similar character."

As the actual group of two hundred families chosen included eight nationalities and eleven different sizes of family (ranging from two to twelve) with incomes ranging from $200 to $2,500 and occupations of great variety, including unskilled, skilled, and trading occupations, it is difficult to see how so small a group as a whole can be taken as a type, in all its diversity, of the "working" population as a whole, and still less how, even under proper classification and subdivision, to show the effect of variation in one condition at a time (as variation of income within one race- and size-of-family group, or variations of race within one income-and size-of- family group, or of size of families within the same income-and race-group), any

*More," Wage-earners' Budgets," New York, 1907.

safe conclusion can be drawn from groups which in many cases are so small as to reduce to individual instances.

In the budget investigation just referred to, some attempt is made to so classify the material presented as to show income and expenditure according to the different elements involved, but we never succeed in getting only one variable at a time. If we have a classification showing variation in income-groups within one size-of-family group, the race elements are mingled indiscriminately, so that we cannot tell whether it is the grade of income reached (which is shown), or the different preponderance of different race elements in different income-groups (which is not shown where size-of-family and income-groups are compared) that is related to the expenditure.

In like manner, in another set of tables, the reader thinks he has arrived at some wise conclusion as to the effect of race traits on expenditure by comparison of different races within the same income-group, when he is suddenly brought to a standstill by observing that the different race-groups show different sizes of family, a circumstance which has its own separate effect on both the amount and the proportion of expenditure.

The inquiry into standards of living recently undertaken by the New York Conference of Charities and Correction avoided some of these defects by confining the investigation as nearly as possible to families of approximately the same size (which might be considered as the average type,-e.g., father, mother, and three children), and within a small range of income so selected as to bring the families within the lower levels of what was assumed to be "normal" subsistence.

But this investigation succeeded in securing for detailed study only two hundred and thirty schedules, covering three income classes and eight race-groups.

Realizing the difficulties of the undertaking, the cautious investigator of budgets will disclaim any attempt at elaborate generalization, and will even, with studied modesty, claim to make none, offering his study as merely a bit of social history to be used by some one else who may find it helpful.

But it is interesting to see that these very investigators

cannot, after all, refrain from generalization; and some of the attempts at interpretation are rather amusing. To take one more instance from More's "Workingman's Budgets," the deduction is made in the final summary that "thrift seems to be most marked in nations in which the preponderance of the income is from the husband," on the basis of one table giving percentage of income derived from the husband and another giving the average surplus in families where there was a surplus shown, both by nationalities.

It occurred to the present writer that the amount of income per person for the families of different races might have something to do with it, too. For instance, Norwegians stood at the head of the list as to amount of surplus and also per cent. of income from husbands. But the group, as a whole, consisted of four families of an average size of 3.5 and an average income of $1,171.75. Such families ought to show a surplus as compared with a group where the average income is $850 and the average size of family 6.0. It might be objected that these very facts are closely related to the percentage of income from the husband, this growing naturally less as one goes down in the scale of prosperity.

To test this and to see also if the generalization was justified, let us arrange the race-groups in the order, respectively, of average income per person, of average surplus where a surplus existed, and of percentage of income from the husband, also giving the number of families in each group, to indicate the weight to be attributed to the facts they show, on the basis of the tables given in the report.

Comparison of the middle column with the right and left hand columns, respectively, will indicate which generalization was the better founded. It should be recalled that irregularity in the correspondence between order according to average income and order according to surplus may be allowed, as indicating thrift as a purely race trait, aside from income; and the chief irregularities are explicable on that basis, notably for Italy, which has a race reputation for thrift.

In the right-hand pair of columns, however, the hypothesis on which they are founded leaves no room for irregularity in order. As a matter of fact, the correspondence fails after the very first item. Omitting race-groups where the number of families is less than four helps matters somewhat; but even then the correspondence is not close, failing notably for the Germans and French.

But, after all, what is the value of these studies without some generalization? What is looked for in these inquiries into standards of living is something beyond mere bits of social history. The chief inducement for taking up social research is to get some light on social causation or, at least, social tendency; and, in particular, circumstances of life should be related to the well-being of the family. It is not enough to get bare items of income and expenditure. We want to translate them into terms of well-being. This task may be approached in various ways.

The mere fact of survival on a given income is a test of the minimum degree of well-being, or, rather, one should say, of a certain degree; for it may be questioned whether those persons who are actually wiped out by death are of as much concern to society at large as those who drag their lives along through various stages of physical and moral disability.

That a family can barely survive on $500 a year can be learned in a simpler fashion than by a budget inquiry; what we really want to know is what grade of well-being that amount can secure, and under what conditions.

Certain approaches to this problem may be made statistically. For instance, we can reduce the amount of money spent