In a rational system of wages, in which they increased with needs, there would be increments with age and service up to the age of 40 years; and in a rational organization of training and work a man would be progressively worth more as he approached the prime of life. As it is, the majority of men who have energy and a sense of responsibility can earn more after 35 than before 25 years. The suggestion from the tables is that the minimum wage for men over 20 should allow for the support of one child, and that every industry should be organized so that promotion to higher grades of wages should come in the ordinary course to all capable men in the first ten years of work.

In Table IX the distribution of households according to the number of children and irrespective of age, summarized at the bottom of Table VII, is shown for each borough separately. In every borough the numbers descend regularly from the category no children under 14 years (three-fifths of all in Bradford) through successive categories.

TABLE IX.-WORKMEN CLASSIFED ACCORDING TO THE NUMBER OF CHILDREN DEPENDENT ON THEM.

Since the publication of the first number of Economica, the results of the investigation by Mr. B. Seebohm Rowntree and Mr. Frank D. Stuart have been issued in a book entitled The Responsibility of Women Workers for Dependants (Clarendon Press). The tables in it are based on particulars for 13,637 women workers, the cases being taken from eleven towns, and owing to the large number of cases and the detailed information about each, the authors have been able to compile very interesting tables giving for women in eight age-groups the percentage having dependants, and to give with greater particularity than has before been attained the relative prevalence of various causes of women's responsibility.

The resulting figures differ from those of the investigation described in Economica, much as did those of the smaller inquiry described in The Human Needs of Labour, and as with the latter, the unlikeness to ours is so great as to call for some consideration. Reference to the data shows that a change in the age limit for women earners from 16 to 18 years only affects the proportions in our tables very slightly indeed, and as we may regard the data in each inquiry as representative of the situation where they were collected, the possible reasons left for the discrepancy are a difference of conditions in the towns chosen, in the date of the investigation, and in the principles guiding the tabulation. Of these, the difference between the collections of towns is not likely to be very great, and while between 1914 and 1919 the distribution of earners and non-earners by sex and age has undoubtedly altered, the definitions and the methods of tabulation employed would yet appear to be the chief cause of divergence. Mr. Rowntree's investigators ascertained whether women were in receipt of earnings which made it possible to help those for whom they were responsible; in the "general" classification in Economica the question was simply whether they had responsibility, not whether they could meet it. Mr. Rowntree finds 1,180 dependants on 13,637 workers, i.e., 8 per 100; Miss Hogg, in Economica, found about 20 per 100. On the other hand, Mr. Rowntree (p. 17) finds that of the working women concerning whom information was obtained, 12 per cent. "wholly or partially supported others"; Miss Hogg arrives equally (p. 85) at 12 per cent. in the general classification as having partial or total responsibility; in the "particular" classification for five towns, which one would expect to be closer to Mr. Rowntree's, the relative number having partial responsibility is considerably greater. Students of this problem will find it necessary to analyse minutely the methods employed, which are fully described in both the originals.

The Survival of Small Firms

By A. L. BOWLEY,

Professor of Statistics, University of London.

A VERY considerable amount of new information relating to the structure and organization of business and industry was collected by the Government Departments during the war, a great part of which served its immediate purpose and has either not been preserved or will never be analysed nor published. It is the object of this note to rescue a bye-product of the work of a semi-official committee that investigated the first effect of the war on unemployment in London.

In August, 1914, inquiries were made of employers as to the number of persons (male or female) employed in July and in August. Three lists of businesses were used, as follows: The Board of Trade, in connection with the Census of Production then in progress, had a list of all firms (about 1,700) known to be employing 100 or more persons (50 or more in the building trade), and answers were received from nearly 1,000 of these. The Home Office had an index of about 42,000 factories and workshops, and questionnaires were sent to a great number of these selected at random, and about 3,300 were returned. A committee in the city obtained similar information from about 1,300 firms. The whole inquiry was initiated, organized and carried out with great rapidity, and the first report on unemployment was issued exactly four weeks after the declaration of war. An examination of the returns shows that approximately 5,500 firms, employing 340,000 persons, had sent reports. Those from the city related principally to banking, insurance, brokers, dealers, shopkeepers and commercial houses generally; the results in these groups are shown separately in Table I. The Home Office list related mainly to firms employing less than 100 persons, and was, therefore, supplementary to the Board of Trade's list. After careful scrutiny it was estimated that for small firms (employing less than 100) the sample averaged 1 in 10, and for large firms 2 in 3 of all persons employed in London (1,100,000) in the very wide range of occupations included. Before calculating the relative numbers for the table the numbers for the small firms were multiplied by 10, and for the large firms by one and a half, to rectify the inequalities of the samples; these factors are only approximately correct, and the table shows more accurately the distribution within the two groups obtained by dividing it at 100 employees, than in the mass taken as one group. Even when allowance is made for this roughness the results are striking; 72 per cent. of the persons work for firms

who employ less than 100; 988 out of 1,000 of the commercial and 972 out of 1,000 of the industrial firms (each firm taken as a unit) employ less than 100. More than half the firms employ less than 20 persons, and this statement is not affected by the roughness of the approximation.

In Table II the number of firms is shown for certain industries following the Census classification; here no adjustment has been made for the inequality of the samples. Each category covers, in fact, a fairly wide range of separate kinds of manufacture.

The Board of Trade has collected similar information for the country as a whole continually for more than six years, for the purpose of studying changes in employment, but so far has made no systematic publication of the results. It is to be wished that they should issue tables for the various industries similar to those here offered for London.

TABLE II.-Actual Number of Firms reported in cERTAIN INDUSTRIES.

NOTE.-Owing to the inequality of the samples, the numbers in the last column should be reduced (by dividing by various unknown factors whose average is 6) to make it comparable with the earlier columns.