Establishments producing manufactured dairy products but also engaged in distributing fluid milk were canvassed in the census but are not included in group totals or in any of the tables in the General Summary or State Volumes. These are shown separately in the industry reports, however, primarily for the purpose of obtaining commodity statistics. Motion picture producers and establishments processing films for motion picture producers reported in the census and statistics for these establishments are shown in the Industry Volume but are not included in group totals or in any of the tables in the General Summary or State Volumes. (e) Military arsenals, Navy yards, and Government printing establishments were not included in the census. 7. Establishments Covered in the CensusComparison With Other Figures Despite the application of fairly uniform definitions of manufacturing activity, the 1947 census figures differ, in some cases rather markedly, as to region and industry from those published by other Federal agencies for the same year. These differences reflect in large part the greater emphasis in the census on securing separate reports for each establishment of companies with more than one plant, thus permitting more accurate geographic breakdowns and the exclusion from the census of nonmanufacturing activities (mining, transportation, distribution, etc.) of such companies at locations other than the manufacturing plant, and in some instances the exclusion of "general office" activities at the plant itself. The exclusion of such nonmanufacturing activities from the census is facilitated by the fact that all large manufacturers are required to submit for each establishment detailed information on input of materials and production of commodities. Also, the greater detail on production and materials consumption received in the 1947 Census of Manufactures makes possible more rigid application of the Standard Industrial Classification rules which provide for the exclusion from manufacturing of establishments producing on a custom basis or for retail sale on the premises and of establishments engaged primarily in repair work. This factor eliminated from the census a large number of nonmanufacturing establishments. In terms of employment, the exclusion of nonmanufacturing activities was less consequential. A minor portion of the difference between census and other figures arose from the fact that the 1947 census did not cover several manufacturing activities (see Section 6, item c), of which logging camps and logging contractors accounted for the largest number of employees. For these reasons, the 1947 census employment figures for the United States as a whole and for most of the States are somewhat lower than those published what higher in a comparison of census and other figures for a number of States-usually the smaller States-and in many local areas. A comparison of census and other figures for a number of cities and counties indicates a tendency for census figures to be lower in central cities and higher in surrounding areas. 8. Completeness of Coverage Coverage of the 1947 Census was checked by a carefully conducted sample survey, which provided direct measures of the completeness of the census. Estimates developed in this study, which covered all establishments within the scope of the census except sawmills, indicate that 98.2 per cent of the manufacturing employment and 98.7 percent of total wages and salaries are included in the census tables. These figures apply to the totals for manufacturing in the United States and not necessarily to the data in individual industry, area, size or other detailed tables. In particular, the undercoverage is most pronounced among small establishments and "marginal” manufacturing industries, where the appropriate classification of individual establishments as manufacturing or nonmanufacturing is frequently difficult. Also, establishments operating during only part of the census year, and establishments not under corporate control, had heavy weight in the undercoverage totals, relative to the totals for all manufacturing. The omitted establishments averaged approximately 12 employees each, compared with the census average of 60 employees. Very few large and individually important establishments were omitted and more than 98 percent of the establishments with 50 or more employees were covered. Thus, the approximately 10 precent undercoverage of establishments consisted mostly of very small establishments and had an exceedingly small impact on the employment, pay roll, value added by manufacture, and other primary manufacturing measures. The census figures on number of manufacturing establishments unquestionably are among the least important manufacturing statistics. In addition to their coverage qualifications, they tend to be inaccurate because of problems of definition particularly for small establishments. Marginal industries account for only 15 percent of all manufacturing employment, but for 35 percent of the omitted employment. Approximately onesixth of the employment undercoverage was contributed by establishments that operated during only part of the census year. Exactly comparable data are not available for the part-year operations covered in the census, but the proportion is much lower. Corporations account for 90 percent of the census employment, but for only 60 percent of the omitted em among establishments which were not included in the census mailing list. The remaining undercoverage was accounted for by manufacturing establishments which were mailed census schedules but which were not included in the census tabulations. These inThese included establishments from whom reports were not obtained, reports incorrectly classified as nonmanufacturing, etc. The check from which these results were derived included all recognizable commercial establishments in 10,000 small sample areas. These sample areas were city and town blocks and well defined rural segments which could be canvassed thoroughly without difficulty. Sanborn Maps, which show and describe individual structures, were used extensively in sampling. For places of approximately 25,000 population or more, listings of three classes of blocks were available, based on the structural descriptions on the maps. These were: 1. Factory blocks-blocks containing manufacturing establishments. 2. Commercial blocks-non-factory blocks containing commercial structures. 3. Residential blocks-blocks containing no commercial structures. In addition, listings were compiled of the factory blocks in smaller towns for which Sanborn Maps also were available. Factory blocks in all large places within 50 miles of a Census Bureau District Office were sampled by selecting every 20th such block. Factory blocks in the more distant large places were sampled by selecting every other such place (63 in all) and every 10th factory block within the selected places. About 4,100 factory blocks were included by this procedure. Similar procedures were used to sample the factory blocks in the smaller towns and the commercial and residential blocks in the large places, but the sampling was confined to the 68 primary sampling units previously chosen for the Bureau's Current Population Survey. Approximately 900 town factory blocks (2 percent), 1,300 commercial blocks (0.5 percent) and 900 residential blocks (0.25 percent) were included. All other territory, including non-factory town blocks and areas not covered by Sanborn Maps, was represented by a sample of 3,000 rural and miscellaneous areal segments in the 68 primary sampling units. The names and addresses of the establishments on the census mailing list which were located in sample blocks were listed in advance of the actual field canvass. In enumerating their assigned blocks, the enumerators were required to list all other commercial establishments they found, to describe and classify the activity of each, and to interview representatives of all manufacturing, all repair, all wholesale, all warehouse and all “doubtful" establishments which had not been prelisted. (In rural sample segments, advance listing of the census mailing list establishments was not practicable, and all commercial establishments were both listed and interviewed.) The basic question asked was, "Is anything made, processed, fabricated, or assembled at this location?" When affirmative answers were received, further questions were asked regarding employment, relative receipts from nonmanufacturing activities, other business names and addresses used and, if operations were started after 1947, the name and activity of the predecessor. Because establishment classification is difficult, all the listings were reviewed. Where there was any suspicion that a nonprelisted establishment might be engaged in manufacturing whether or not there had been an interview or regardless of the answers obtained, supplementary inquiries were made in order to obtain adequate classifying information. Often when these establishments proved to be engaged in manufacturing, they also were found to have been included in the census mailing list. Such cases included errors in prelisting, establishments that had moved, establishments that had been enumerated under their trade names but that appeared on the mailing list under the owners' names, etc. Exact agreement of identifying information was required for all cases accepted as covered by the census mailing list. Complete 1947 Census of Manufactures reports were obtained for the establishments ultimately classified as manufacturers omitted from the census mailing list. In order to measure undercoverage or "losses" that occurred in the collection and processing phases of the census, all the establishments on the census mailing list for the sample areas were checked against the list of establishments included in the census tabulations. Ordinarily, considerable information was available for the establishments that did not appear in the census tabulations. Where this information conclusively showed that the establishment was out of scope, no further action was taken. Where the evidence was not absolute, additional facts were sought. For example, where respondents had clearly described their activities, and these were primarily nonmanufacturing, the case was considered satisfied. Statements that establishments were not manufacturers, unsupported by descriptions of their actual activities, however, were not accepted. Unacceptable also were explanations that the establishment had gone out of business unless a date prior to 1947 was indicated. Census pre-canvass cards, Bureau of Old Age and Survivors records, and direct inquiries to respondents were utilized in an effort to confirm the doubtful classifications. This proved highly successful. Only 2 percent of the estimated employment undercoverage is accounted for by sample establishments classified on the basis of scanty information. About 9. Change in Scope Between 1939 and 1947 "Out of business" problem cases were resolved by reference to BOASI records. All such manufactur to that agency during any quarter of 1947 were considered "losses" from the mailing list, and included in the undercoverage estimates. In 1939 and earlier biennial censuses, establishments having less than $5,000 value of products were designated as out of the scope of the Census of Manufactures. In the 1947 census, a different procedure was followed. No value size limit was used but, in ing establishments which had reported employment stead, no attempt was made to canvass establishments known to have had no employees during 1947, and all reports received from such establishments were discarded before tabulation. The basic reason for this decision was that in the mail canvass of 1947, the main reliance for coverage of small establishments was on the files maintained by the Bureau of Old-Age and Survivors Insurance of the Social Security Administration, which do not include establishments having no employees. It was not feasible to obtain complete census reports from the mailing list establishments "lost" in the course of the census. Many of them were in the out of business group. Because the necessary 1947 employment and wage figures could be obtained from BOASI records, satisfactory estimates of coverage could be developed for those two measures. The sample data were inflated to estimated totals by multiplying the individual establishment figures by the reciprocals of their sampling rates (that is, the sampling rate for the block or segment in which the establishment was located). The percentage of coverage was then calculated by computing the ratio of the census totals (excluding sawmill data) to the sum of the census figures and the estimated under coverage. The estimates of coverage given there are subject to the sampling errors specified in the footnote to the following table. The reliability of these estiThe reliability of these estimates, as in all statistical surveys, is also affected by reporting errors and errors in the application of procedures. Because of the small scale on which the survey was conducted, greater care and closer supervision could be exercised than is possible in a larger survey such as the census. It is believed, therefore, that these errors were kept to a minimum. NOTE: The estimates of undercoverage are subject to the following significant sampling errors. The chances are 95 out of 100 that the one-starred estimates will not vary from the true figure by more than 10 to 14 percent, the two-starred estimates by more than 15 to 24 percent, and the three-starred estimates by more than 25 to 40 percent. 1 Marginal industries are those which include many establishments engaged both in manufacturing and nonmanufacturing activity. A detailed analysis of the operations of each such establishment is necessary to determine whether it is engaged primarily This change in procedure has not, however, appreciably affected the comparability of the figures for 1947 with those for earlier years except for the figures on number of establishments. In 1939, establishments with no employees amounted to fewer than 5,000 and accounted for about $125 million out of a total of $25 billion value added by manufacture. However, a substantial portion of this value added was accounted for by a small number of jobbers of the type also included in 1947. On the other hand, a partial study of the 1947 returns indicates that some 10,000 establishments with products valued at less than $5,000 and accounting for about $15 million value added were tabulated in that year. Moreover, the price level of manufactured goods approximately doubled between 1939 and 1947, and therefore a minimum size limit of $10,000 for 1947 would correspond in terms of manufacturing activities to the $5,000 size limit used in 1939. In 1947, there were tabulated some 23,000 establishments shipping products valued at less than $10,000 and accounting in total for less than $75 million value added by manufacture. Accordingly, it will be seen that, on balance, about 18,000 of the 67,000 increase in the number of establishments between 1939 and 1947 is accounted for by the exclusion of no-employee establishments and the inclusion of other establishments whose scale of activities was so limited as to have been eliminated from the 1939 census, but that other census figures are not thereby appreciably affected. Establishments primarily engaged in the following activities were included in manufacturing industries in 1947 but were considered as nonmanufacturing in 1939: Coffee and spice, roasting and grinding The 1947 tabulations included 400 coffee roasters (in- $75 million of value added; and 163 tobacco stemmers and redryers, who employed nearly 26,000 persons and contributed $72 million to the value added by manufacturing. The 1947 census excluded the following types of establishments which were included in the 1939 census: Dried fruit packers Retail establishments manufacturing ice cream Independent retail bakeries, selling through a single retail outlet and not operating any house-to-house routes . Logging camps and logging camp operations Contractors and building materials dealers primarily engaged in mixing and delivering ready-mixed concrete Quarrying departments of establishments both quarrying limestone and producing hydraulic cement and lime. The hydraulic cement and lime plants of the combined operations were retained in manufacturing Quarries also engaged in producing cut-stone and stone products Machine shops engaged exclusively or almost exclusively in repair work followed in the preparation of this system of industry classification: (1) The classification should conform to the existing structure of American industry; (2) The reporting units to be classified are establishments (for definition of establishment, cf. Section 5 above), rather than legal entities or companies; each establishment is to be classified according to its major activity; (3) To be recognized as an industry, each group of estab.lishments must have significance from the standpoint of the number of establishments, volume of business, employment and pay rolls, and other important economic features. In addition to the above principles, it is important that an industry comprise a group of establishments: (a) whose output consists, to a relatively large extent, of the products defining the industry; and (b) whose output of the products defining the industry accounts for a relatively high proportion of the total output of those products by establishments in all industries. Finally, (c) the output of the bulk of the establishments should be distributed among the various products or groups of products defining the industry; if any considerable number of establishments concentrated on the production of a single product within the group of products defining the industry, such concentration would constitute the basis for a further subdivision of the industry. Except for logging camps and logging camp contrac- 10. Definition of Industries An industry consists of a group of establishments primarily engaged in the same line or similar lines of economic activity. In the manufacturing field, the line of activity is generally defined in terms of the products made or the processes of manufacture used. On this basis, there may theoretically be thousands of manufacturing industries corresponding to the different types of products and the processes used in their manufacture. However, industries established in this manner would be too numerous to deal with and, further, would not generally satisfy the criteria considered essential in the creation of a good system of industry classification. Except for certain relatively minor modifications, the 453 manufacturing industry classifications used in this census were derived from the 1945 Standard Industrial Classification Manual issued by the Bureau Into Industries An establishment is assigned to, or classified in, an industry, generally on the basis of the principal products made. The products made by each establishment are grouped according to the industries to which they "belong" (see Appendix C, Industry Descriptions) and the group of products accounting for the principal portion of the total value of shipments of the establishment determines its industry classification. This group of products is said to be the primary products of the establishment as well as of the industry which it defines; all other products made by establishments classified in the industry are referred to as secondary products. In some instances, a knowledge of the types and value of products shipped is not sufficient for determining the proper industry classification of an establishment. It may also be necessary to know the process used in the manufacture of the products, or the materials used, or other characteristics of the operations of the establishment. For example, an establishment primarily engaged in producing insulated copper wire is classified in Industry 3392, Wire Drawing, if it purchases copper rods, draws the rods into wire, and insulates the wire, but in Industry 3631, Insulated Wire and Cable, if it purchases copper wire While some establishments produce only the primary products of the industry in which they are classified, this is rarely true of all the establishments in an industry. The general statistics (employment, pay rolls, value added by manufacture, etc.) shown for an industry, therefore, reflect not only the primary activities of the establishments in that industry, but also their activities in the manufacture of secondary products and, for that matter, their auxiliary nonmanufacturing activities. For this reason the industry statistics shown throughout the census volumes and, in particular, in tables 1 through 4 of each industry report in the Industry Volume, usually cannot be directly related to the product statistics (table 6 in each industry report) showing shipments by all industries of the primary products of that industry. The extent to which industry and product statistics can be matched with each other is indicated in a transition table (table 5) which shows on the one hand the proportions by value of the primary and secondary products shipped by the industry and on the other, the value of the primary products of the industry made as secondary products in other industries. In a classification system dividing the manufacturing field into some 450 separate industries to which establishments are assigned on the basis of their principal activity, a certain amount of overlapping production by one industry of products of other industries, is bound to arise. Such overlapping is particularly prevalent in such industry groups as apparel, furniture, and metal fabricating, where establishments within each group, employing the same basic types of machinery and fabricating operations, produce a wide variety of products belonging to a number of different industries. For 1947, the average amount of overlapping for all industries was approximately 10 percent in both directions, compared with from 15 to 20 percent for the furniture and metal fabricating (except transportation equipment) groups of industries. The greater the degree of overlapping, the less meaningful are the relationships between general and product statistics for the particular industry. Measures of overlapping production for each of the 1947 industries are shown in Appendix C, Industry Descriptions; these measures are derived from table 5 of the Industry reports. In a number of instances, the Standard Industrial Classification was modified for use in this census to minimize the extent of overlapping for certain industries. (See 1947 Census Deviations from Standard Industrial Classification, Appendix E.) As indicated in Section 5, Definition of Establishment, the census attempts for some industries to secure separate reports for large plants compile "cleaner" industry statistics by reducing the extent of overlapping production in each of the industries affected. As the number of separate industry classifications for manufacturing has increased over the years there has been a greater likelihood of year-to-year shifts assigned on the basis of the principal products male. of the industry to which a particular establishment is An establishment whose value of output consists of 55 percent creamery butter and 45 percent natural cheese in one year is classified in the creamery butter industry for that year; if, in another year, the proportions are reversed (45 percent creamery butter and 55 percent natural cheese), it appropriately belongs in the natural cheese industry. Such shifts, when affecting a number of large establishments, may be reflected in an apparent abnormal increase or decline in activity for the industries affected, whereas, in fact there may have been little change. Census experience in applying the industrial classification system to establishment reports indicate that on the whole the system does conform to the structure of industry. However, a number of industries were found to be of relatively little economic significance. Some of them, such as Industry 2014, Sausage Casings; 2296, Linen Goods; 2864, Natural Dyeing Materials; and 3335, Primary Refining of Magnesium, were merged with other industries. (For a complete list, see Comparison of 1947 Census with Standard Industrial Classification, Appendix E.) No attempt was made to make similar combinations for a number of other relatively unimportant industries. There were also difficulties encountered in the assignment of industry classifications to certain types of establishments. In some of the borderline areas between manufacturing and other major divisions of the classification system, it is not possible to attain the same degree of precision in assignment of establishments to industry classifications, nor as complete industry coverage as in other areas of the manufacturing division. This is particularly true for manufacturing industries, products of which are also made to a significant extent by establishments engaged primarily in nonmanufacturing activities. Since no reports were obtained from such primarily nonmanufacturing establishments which fall outside the scope of the Census of Manufactures, national totals for the output of these products as published in the table 6's will be understated to the extent of their production. Where this type of understatement is important, e.g., venetian blinds, awnings, millwork, dried fruit, and prepared feeds, it is pointed out in the textual material preceding the tables for each of |