dustries considered in combination. Fifteen building-material industries show a net decrease of employment from 861,953 in 1929 to 665,153 in 1937, a loss of 196,800 jobs, or 22.8 percent (table 1). If an allowance is made for decreased employment in structural-steel departments of steel works and rolling mills, not included in these figures, the total decrease amounted to about 226,000 jobs. On the other hand, in industries other than these 15, employment increased from 7,467,595 to 7,902,585, an increase of 5.8 percent. With the allowance for the influence of structural steel on these totals, the increase would be about 6.2 percent. This closely approximates the population increase from 1929 to 1937.

On a regional basis, the shifts were pronounced. Increases in total manufacturing employment were largely concentrated in Michigan and the adjoining States of Indiana and Illinois, in the States of the South Atlantic, South Central, and Southwestern regions, and on the Pacific coast. In general, the States in the northeastern section of the country had not returned to their 1929 levels of manufacturing employment by 1937.

TABLE 1.-Number of Wage Earners Employed in Industries Manufacturing Building Materials, 1929 and 1987

1 Because of changes in classification, the clay-products industry was combined with the crucibles in dustry in 1929, and with the nonclay-refractories industry in 1937 in order to obtain comparable data. Because of changes in classification, the wall-board industry was combined with the gypsum-product industry in 1937 in order to obtain comparable data

CHANGES OF CENSUS CLASSIFICATION

Before census totals can be compared, the figures for 1929 must be adjusted to be made comparable with those for 1937. The unadjusted census total for 1929, 8,838,743 wage earners, is 270,000 higher than the figure for 1937. There were, however, 7 industries included in the Census of Manufactures for 1929 which were not treated as manufacturing industries in 1937. Furthermore, though no allowance can be made for differences in the completeness of coverage, it is probable that the 1937 census somewhat understates the number of wage earners in that year.

1 The number of wage earners employed in these industries was, therefore, deducted from the 1929 total. The 7 industries, and their average number of wage earners for 1929 are coffee and spice, roasting and grinding, 9,053; flax and hemp, dressed, 47; gas, manufac tured, illuminating, and heating, 43,065; motion pictures, not including projection in theaters, 10,784; peanuts, walnuts, and other nuts, processed or shelled, 7,933; railroad repair shops, electric, 29,475; and railroad repair shops, steam, 368,681. In addition, the classification of wage earners in the turpentine and resin industry was so radically changed in 1937 that it was necessary to deduct the employment in this industry from the totals in both 1929 and 1937.

The survey of activities of manufacturing establishments for 1929 was conducted in conjunction with the regular decennial census of population. A large staff of field employees was therefore available to make a door-to-door canvass of every establishment, thus insuring practically complete coverage. For the 1937 census, however, the field force was substantially smaller, and it was necessary to depend largely upon the returns of mailed questionnaires.

In most instances, this did not result in any significant loss of coverage. There are, however, certain industries-notably in the garment-manufacturing group-in which there is

INDUSTRY TRENDS

Comparing employment in manufacturing industries in 1929 and 1937 on an industry by industry basis, one finds a limited number of industries outside of the building-material field, in which employment declined substantially. The list of those in which employment declined by as much as 5,000 people is given in the succeeding paragraphs.

Three industries show net losses in employment attributable largely to technological shifts-cigars, tires and tubes, and silk and rayon. Cigars dropped 28,000 people. This reflects an absolute decline in the number of cigars smoked and a rapid introduction of machine processes for hand processes. There is no reason to expect that there will ever again be as many people employed in the manufacture of cigars as were employed in 1929. The trend may very well continue down. Silk and rayon employed 13,500 fewer workers in 1937 than in 1929. The silk industry has been declining and will probably continue to decline. Rayon can be handled with fewer workers than equal quantities of silk. Tires and tubes dropped 20,000 people, reflecting fundamental and almost revolutionary changes in both the process of producing tires and in the life of the tires produced.

There are three industries in which substitution has accounted for very substantial losses of employment. The first of these is manufactured ice from which industry 13,500 people were dropped between 1929 and 1937. Somewhere near this number of people appear to have been added in the manufacture of electric refrigerators. There were 13,500 people dropped in confectionery and ice cream. It was supposed that a fairly substantial part of the growth in the demand for confectionery and ice cream during the decade of the 1920's was attributable to prohibition. It is equally reasonable to suppose that the decline reflects, in part, the repeal of prohibition in 1933. In the manufacture of wooden boxes, 4,500 people were dropped. In this connection, it should be noted that the growth of the paper-box industry has absorbed nearly 10,000 workers.

It is not possible to account as definitely for declines of 19,000 wage earners in the furniture and wood-turning industries, of 7,300 in the production of rubber boots and shoes, and 7,100 in men's and women's hats and millinery, and of 6,800 in silverware, plated ware, and jewelry. There has been an absolute decline in the production of rubber boots and shoes. This may very well be connected with changes in our habits of living as induced by the automobile. The change in furniture and wood turning may be related to the lack of development of new homes as well as to some shift in consumer preferences. As regards hats and millinery, there may have been some change in style and consumer preferences, for there has been a continued growth in the employment of people engaged in the production of clothing generally. It is quite possible that this is one of the industries in which the use of a questionnaire in 1937, resulted in some underreporting.

The above 10 industries or groups of industries employed about 788,000 wage earners in 1929 and about 654,000 in 1937-a decrease of about 134,000 wageearner jobs. The loss of these jobs is attributed, in part, to technological changes, in part to changes in consumer tastes. In order to make the list complete, one should perhaps add cotton textiles, employing an average of 435,400 in 1937 and 441,100 in 1929. The decline, much smaller relatively than in the other industries always a large turn-over of establishments, as well as frequent shifting of plant location. In such cases, the reliance upon the mailed questionnaire resulted in a substantial understatement of activities of the industry. To the extent that this was true of the 1937 Census of Manufactures, employment in 1937 was underestimated. While the net effect that this has upon the United States total is relatively minor, it is well to keep the point in mind when considering later discussions of changes in employment by State and industry. In general, it is to be expected that loss of coverage for the reason just given will take place largely in the smallest establishments. Some idea of the effect of this loss may, therefore, be obtained from a consideration of the number of such establishments, and their relative importance as employers of labor. It will be found that while almost half of all the manufacturing establishments in 1929 employed no more than 5 workers, these plants accounted for only 3.3 percent of the total number of manufacturing wage earners. In 1937, only 41.0 percent of the establishments were of this size, and employed 2.0 percent of the wage earners. It may be assumed that part of this decline can be explained by the tendency toward elimination of the smallest sized manufacturing establishments. It thus seems safe to say that, even with a fairly large understatement of establishments, the 1937 census probably was less than 1 percent off in its wage-earner total.

'The actual decrease is somewhat less than 13.500 workers. In 1935, the Census of Manufactures made some changes in the classification of the silk and rayon industry, as a result of which some plants formerly classified in that industry were assigned to other Industries. The number of workers in such plants cannot be determined exactly.

listed, is large in absolute terms and accounts for more than 5,000 workers. Here then are possibly 139,000 jobs that disappeared on the basis of net changes of employment between 1929 and 1937.

The changes in classification within the census group, designated as "machinery, not including transportation equipment," are too large to allow one to trace changes in employment in the separate industries. The group as a whole shows little change. There was probably a significantly large decrease in the number employed on radios that might be included in the list just completed.'

We are frequently told that what is needed in the United States is a new industry. Did any new industries develop between 1929 and 1937? The census gives evidence of only two industries in 1937 that employed as many as 10,000 workers which were essentially nonexistent in 1929. In the manufacture of alcoholic beverages of one sort or another, 62,000 additional people were employed. Virtually none had been legally so employed in 1929. Nearly 17,000 people were employed in the manufacture of synthetic resins and plastics. Only 3,500 workers were classified in this industry by the census in 1929. Essentially, therefore, we may say that there was a growth of 76,000 jobs in the new industries.

Five industries, however, experienced an increase of employment of more than 40 percent between 1929 and 1937. While none of them may be regarded as a new industry, they are certainly indicating rates of growth sufficiently large to meet the desire that is expressed when we say that we need new industries in the United States. These industries were canning, which added 43,000 workers; stamped and pressed metalware, 21,000; rayon production, 16,000; aircraft, 9,000; and asbestos products, used largely for insulation, 5,000-a total growth in the five industries of 94,000 workers.

This is a growth within what may be described as new industries of 170,000 workers-substantially more than the decrease in all of the industries that were showing a fundamental tendency to shrink.

This does not complete the list of industries with growing employment. There were at least 11 that experienced a rate of increase of employment between 1929 and 1937 two to four times as great as the increase in population. These industries and their respective increases were blast furnaces, steel works and rolling mills, 83,000; baking, 38,000; men's clothing, 26,000; shirts and collars, 10,000; women's clothing, 55,000; converted paper products 26,000;

Total wage-earner employment in the group as classified in 1929 was 1,091,269 and as classified in 1937 was 955,975. To achieve approximate comparability it is necessary to add to the figures for 1929, 14,416 wage earners making phonographs, carried in miscellan eous manufacturing in 1929; 9,784 wage earners on nonmechanical refrigerators and refrigerator cabinets, carried in the lumber group in 1929; and 1,784 wage earners on printer's and engraver's materials and in type foundries, carried in the printing and publishing group in 1929. It is also necessary to exclude from the 1929 total, 5,038 wage earners who were employed in plants classified in the gas machines and meters industry in 1929 and in the instruments and apparatus, professional and scientific industry in 1935 and, therefore, carried in the miscellaneous manufacturing group in 1937. This gives an adjusted total of 1,112,215 for 1929.

The corresponding total for 1937 is 1,100,484. This includes, in addition to the group total of 955.975 classified under machinery by the census in that year, 120,024 wage earners making foundry products and 24,485 wage earners in boiler shops who were carried in the iron and steel group by the census in 1937, but had been carried in the machinery group in 1929.

Two minor differences in classification in 1929 and 1937 are noted merely for the sake of completeness. In 1929 numbering machines were carried in hand stamps and stencils and brands. In 1937 they were carried as an unidentifiable but small part of cash registers, etc. On the other hand, taxicab meters were part of cash registers, etc., in 1929, and in 1937 were an unidentifiable part of instruments and apparatus in the miscellaneous group of industries.

It

In 1929 the census showed 328,722 wage earners in electrical machinery and radios, and 14,416 in phonographs. In 1937 it showed 257,660 in electrical machinery and 48,343 in radios and phonographs. The totals for the three industries in combination have declined by 37,135. A decrease in radio employment may be safely inferred from changes in the value of radios and tubes produced. Though the classifications may have changed, the value of the products appear to have decreased by more than one-fifth. Whether or not there has also been a decrease in employment in electrical machinery, we cannot say. should be noted that the manufacture of refrigerators (largely mechanical) employed 50,623 wage earners in 1937. In 1929 there were 16,883 in establishments primarily engaged in producing mechanical refrigerators, 9,784 in establishments producing nonmechanical refrigerators and refrigerator cabinets, and we estimate there were 5,731 engaged in manufacturing, refrigerating and ice-making machinery. These three groups, without any break-down, are combined in the 1937 total of 50,623. There was a large decrease in the production of commercial refrigerating and ice-making machinery between 1929 and 1937.

Because of a change in classification of the indutsry in 1931, the figure of 3,500 workers in 1929 is not exactly comparable to the figure of 17,000 in 1937. If the data could be made comparable, the increase in employment in the industry would be somewhat smaller than 13,500.

chemicals, 17,000; glass, 11,500; rubber goods other than tires and boots and shoes, 8,000; leather products, other than boots and shoes, 7,000; cigarettes, 5,000-a growth of 286,000 in the number of wage earners employed in 11 industries.

An adjustment should be made in the steel figure. The net growth of 83,000 conceals decreases in some branches of steel production. There are no figures on the subject, but a loss of employment for 30,000 wage earners may be estimated because of decreased production of structural shapes, concrete reinforcing bars, and nails. The net growth for the balance of the industry was then not 83,000 but 113,000 wage earners. This would raise the total for these 11 industries (excluding steel for construction) from 286,000 to 316,000. Finally, in terms of the industries which show growth of employment at a rate substantially higher than the rate of population growth, one should note knit goods with 22,500 and shipbuilding with 7,000.

All in all, therefore, in industries employing more than 10,000 workers in 1937 and growing substantially more rapidly than population, there was an increase of about 516,000 jobs in manufacturing-76,000 in 2 new industries; 94,000 in 5 industries, with increased employment of over 40 percent; 316,000 in 11 industries, basically old, but still showing a net growth of employment 2 to 4 times as great as population; and about 30,000 in knit goods and shipbuilding. Against this "new industry" gain of 516,000 is to be offset the loss of 139,000 jobs in industries with substantial decreases of employment opportunity. The net gain is 377,000 jobs in relatively large industries showing pronounced trends of employment.

The striking changes in manufacturing employment are thus confined to a relatively small list of industries. Decreases occurred in the total for 15 industries manufacturing building materials: 862,000 in 1929 to 665,000 in 1937. In the 11 industries with large decreases noted there were 1,229,000 in 1929 and 1,090,000 in 1937. The 20 industries for which increases were noted employed 1,862,000 in 1929 and 2,347,000 in 1937.

The remaining manufacturing industries showed smaller absolute changes than those that have been discussed. Some, of course, showed small decreases. Others increased. In general, outside of the building-materials industries and of the industries just listed as changing considerably, there were small gains. In the aggregate, they employed 4,378,000 in 1929 and 4,466,000 in 1937, a net gain of about 88,000 wage earners.

This means that in all manufacturing industries outside of the buildingmaterials field, there was a net increase in employment of about 465,000 wage earners, a gain of 6.2 percent. This large segment of employment-all of manufacturing outside of building materials-expanded almost exactly in proportion to population.

STATE TRENDS OF EMPLOYMENT

When the same adjustments were made in each State as were made in the national figures, it was found that manufacturing employment was greater in 1937 than in 1929 in 23 States and the District of Columbia, while declines were found in 25 States (table 2). The increases ranged from 1.1 percent in New Jersey to 27.4 percent in Michigan. Pennsylvania showed a drop of less than 0.1 percent, while the greatest decline (19.9 percent) was experienced in Arizona. In general, the States in the northeastern section of the country had not returned to their 1929 levels of manufacturing employment by 1937. This condition was true in four of the six New England States, in two of the three Middle Atlantic States, and in Ohio and Delaware. On the other hand, consistent gains were registered in the South. With the single exception of Mississippi every Southern State east of the Mississippi River showed an increase. Most of these gains were quite large; in only two cases were they less than 10 percent.

Not including radios, nonmechanical refrigerators, or ice-making machinery discussed above (p. 1311, footnote 5). These figures include structural steel in blast furnaces, steel works, and rolling mills. They do not include mechanical refrigerators discussed above (p. 1311, footnote 5), wall board and gypsum which in combination increased from 7,462 to 11,590 (see table 1), nor agricultural implements and tractors for which a 1929 to 1937 comparison is not possible.

This figure represents the difference of 435,000 that is the net sum of changes in industries (including all of the iron and steel industry) other than building materials, plus an allowance of 30,000 decline in structural steel made in plants classified in steel works and rolling mills that should be attributed to a decline in building materials.

Two other regions show important increases. Michigan, as has been noted, had an increase of 27.4 percent. Small increases occurred in Illinois and Indiana. This gain reflects increased employment in the automobile and steel industries. The Pacific coast likewise experienced increased employment, notably in California and Oregon. Finally, as part of the growth of southern industry, one should note the increase of employment in the Southwestern States of New Mexico, Oklahoma, and Texas.

In general, with the exceptions just noted, employment declined in the States west of the Mississippi River. The Great Plains and Mountain States have a relatively small number of manufacturing wage earners and the declines involve relatively few workers. Increases occurred, however, only in Nevada and North Dakota, the only two States in the Union with less than 3,000 manufacturing wage earners in 1937. The decreases in the States bordering the Mississippi are significant.

TABLE 2.-Number of wage earners employed in manufacturing industries, by State and geographical division, 1929 and 1937