(table 8 of Volume II). For the most part, the units were renter occupied, and roughly four-fifths of them were located inside standard metropolitan statistical areas. It is not possible to determine what portion would qualify as "dwelling units" since there is no measure of the number of units in regular apartment houses or the number that are the only living quarters in the structure.

The evidence thus far suggests that the use of the housing unit concept in 1960 instead of the dwelling unit concept as in 1950 had relatively little effect on the counts for large areas and for the Nation. Any effect which the change in concept may have on comparability can be expected to be greatest in statistics for certain census tracts and blocks within metropolitan areas. Living quarters classified as housing units in 1960 but which would not have been classified as dwelling units in 1950 tend to be clustered in tracts and blocks where many persons live separately in single rooms in hotels, rooming houses, and other light housekeeping quarters. In such areas, the 1960 housing unit count for an individual tract or block may be higher than the 1950 dwelling unit count even though no units were added by new construction or conversion.

In the 1940 Census, a dwelling unit was defined as the living quarters occupied, or intended for occupancy, by one household. A household consisted of a family or other group of persons living together with common housekeeping arrangements, or a person living alone. The instructions to the enumerator did not explicitly define living quarters as separate units on the basis of cooking equipment or access. Furthermore, living quarters with five lodgers or more were included in the 1940 housing inventory but excluded in 1950 and 1960. Even though there were differences in the definitions used in each of the censuses, the overall effect of the change in definition is believed to be small. The differences, if any, would have the greatest effect on data for small areas such as census tracts or blocks.

Statistics on the number of occupied units (identified as "families" in the early censuses) are available from the Censuses of Population from 1930 back to 1890, with limited data back to 1850. The term "family" or "home" used in these censuses is essentially comparable with the current term "housing unit" (see section on "Occupied housing unit").

Comparability with 1959 Components of Inventory Change survey. In the Components of Inventory Change survey, a sample survey conducted in December 1959 as part of the 1960 Census of Housing, the "dwelling unit" was the unit of enumeration, the dwelling unit being defined as in 1950 (see 1960 Census of Housing, Volume IV, Components of Inventory Change, Part 1A–1, United States and Regions). The dwelling unit was used in the 1959 survey so that change between 1950 and 1959 could be measured on a unit-by-unit basis.

For the United States as a whole, the estimate of the number of dwelling units from the December 1959 sample survey is 58,468,000 and the count of housing units from the April 1960 Census is 58,326,000. For metropolitan areas, however, the count of housing units from the April 1960 enumeration for the 17 areas for which separate data are provided in Volume IV is slightly higher than the count of dwelling units from the December 1959 survey; there is evidence that this difference exists even after allowance for sampling variability of the 1959 estimate and dates of enumeration. Furthermore, the estimate from the April 1960 enumeration of the number of 1- and 2-room renter-occupied housing units in the 17 areas is higher than the estimate of dwelling units in the comparable category from the December 1959 survey.

As indicated above, the housing unit concept is more inclusive than the dwelling unit concept. The effect of the conceptual difference alone, however, may not be revealed by the results of two separate surveys or censuses. Other factors which affect comparability between the December 1959 and the April 1960 figures include the different enumeration procedures employed, the degree of overenumeration and underenumeration in both the survey and the census, the estimation procedure used for the 1959 The results, and the sampling variability of the 1959 estimates. results of the two enumerations are being evaluated in more detail to determine the relationship between the two concepts. Group quarters.-Occupied quarters which do not qualify as housing units are considered group quarters. They are located most frequently in institutions, hospitals, nurses' homes, rooming and boarding houses, residential clubs, missions and flophouses, military and other types of barracks, college dormitories, fraternity and sorority houses, convents, and monasteries. Group

living quarters are shared by the head and five or more persons unrelated to him. Group quarters are not included in the housing inventory; therefore, counts and housing characteristics of such quarters are not reflected in the data in any of the housing reports. The 1960 concept of group quarters is similar to the 1950 concept of nondwelling-unit quarters.

According to the results of the 1960 Census of Population, approximately 4.9 million people or 2.8 percent of the total population of the United States lived in quarters not defined as housing units. In 1950, the number of persons whose quarters were not included in the dwelling unit inventory was 5.7 million or 3.8 percent of the total 1950 population. The 1950 figures are not entirely comparable with the 1960 figures; it is probable, due to the change in definition from "dwelling unit" to "housing unit," that some 1-room quarters which were housing units in 1960 would not have been dwelling units according to the 1950 definition.

OCCUPANCY CHARACTERISTICS

The questions used to ascertain occupancy characteristics of housing units are shown below. The items on persons, color, and tenure appeared as self-enumeration items on the Advance Census Report; the item on year moved and the questions related to farm-nonfarm residence for occupied units appeared as self-enumeration items on the Household Questionnaire, which was used in two-stage areas. The direct-interview questions appeared on the FOSDIC schedule. Vacancy status, duration of vacancy, and the number of acres in the place for vacant units were determined by the enumerator through information obtained from owners, landlords, or neighbors.

Occupied housing unit.-A housing unit is "occupied" if it is the usual place of residence of the person or group of persons living in it at the time of enumeration. Included are units occupied by persons who are only temporarily absent, such as persons on vacation. Units occupied by persons with no usual place of residence are also considered "occupied." For example, a unit occupied by migratory workers who have no usual residence elsewhere is considered occupied; however, if the migrants have a residence elsewhere, the unit in which they are temporarily living is classified as vacant.

A household consists of all the persons who occupy a housing unit. By definition, therefore, the count of occupied housing units is the same as the count of households in the 1960 Census of Population reports. In some cases, however, there are small differences in the counts resulting from processing procedures. (The ratio estimation procedure for population data was carried out for groups of persons on the basis of age, color, sex, head of household by tenure of unit, and nonhead; the estimation proce dure for housing data was carried out for groups of housing units on the basis of tenure, color, and vacancy status; see section on "Ratio estimation.")

The same definition for classifying a unit as occupied was used in the 1950 and 1940 Censuses of Housing. The concept of occupied units in the Censuses of Population in 1930 and earlier, although identified as "homes," "families," or "private families," are also essentially comparable with the 1960 concept of occupied housing units (or households). The counts in some of the earlier censuses included group quarters (or quasi-family groups), with each institution, rooming house, and similar quarters counted as one unit.

Population in units.-The "population in housing units, 1960" is the total count of persons living in quarters classified as housing units; it excludes persons living in group quarters. The 1960 figure for "population per occupied unit" was computed by dividing the population in housing units by the number of occupied units. When this figure is tabulated from a sample, it is subject to some sampling variability (see section on "Sample design and

By definition, the population in housing units is the same as population in households in the reports of the 1960 Census of Population. In some cases, however, there are small differences resulting from processing procedures (see discussion of households in the section above on "Occupied housing unit").

The "population in dwelling units, 1950" is the 1950 population in quarters classified as dwelling units; and the "population per occupied unit" is computed in the same way as the 1960 figure. Although the concepts are similar, the change from "dwelling unit" to "housing unit" may have introduced an element of difference between the 1960 and 1950 figures in some

areas.

The 1950 figures for places, urbanized areas, and counties in Volume I of the 1960 Housing reports are based on boundaries as of 1950. For individual SMSA's and the total inside and outside SMSA's, the 1950 figures are based on 1960 boundaries; for central cities, however, the figures are based on 1950 city limits for cities designated central cities in 1960. The 1950 figures for the urban and rural portions of a State, division, region, or the United States are based on the 1950 designation of urban and rural territory. The 1950 figures for rural-nonfarm and ruralfarm units in table 5 in the State chapters reflect the 1950 definition of farm residence; because of the change in the definition of farm residence, the 1950 and 1960 figures are not comparable. Leaders (---) are shown where it was not possible or feasible to reconstruct the 1950 figures.

The "population per occupied unit" in the 1940 Census of Housing is essentially comparable with the concept of "population per occupied unit" (or "population per household") in the 1950 and 1960 Censuses. Similarly, the concept of population per unit (identified as "population per family" or "population per private family") in the Censuses of Population in 1930 and earlier are essentially comparable with the current concept of population per occupied unit (or household). The population count in the publications of 1940 and some of the earlier censuses, however, is the total population and includes persons in institutions, rooming houses, and similar quarters. Likewise, counts of units (families) for some of the censuses include group quarters or quasifamily groups. Thus, comparability is affected in small areas having an appreciable number of persons in group quarters (quasi-family groups).

In text table F in the United States Summary chapter, total population was used for all years in the computation of the pop

Persons (P2).-All persons enumerated in the 1960 Census of Population as members of the household were counted in determining the number of persons who occupied the housing unit. These persons include not only occupants related to the head but also any lodgers (roomers, boarders, partners, wards, foster children) and resident employees who shared the living quarters of the household head.

The median number of persons for occupied housing units is the theoretical value which divides the distribution into two equal parts-one-half the units having more persons and one-half having fewer persons than the median. In the computation of the median, a continuous distribution was assumed, with the whole number of persons as the midpoint of the class interval. For example, when the median was in the 3-person group, the lower and upper limits were assumed to be 2.5 and 3.5 persons, respectively. In table 26 of the State chapters of Volume I, the median for all occupied units was computed from the 100-percent count of units, whereas the medians for owner- and renter-occupied units were computed from the 25-percent sample of units. The sample rate for medians in other tables is given in table I.

Comparable data on the number of persons in the unit are available from the 1950 and 1940 Censuses of Housing. In certain areas, comparability may be affected to some small extent by the change in concept from dwelling unit to housing unit. Data are available also from the Censuses of Population in 1930 and earlier; however, data for some of these censuses pertain only to persons related to the head.

Persons per room. The number of persons per room was computed for each occupied housing unit by dividing the number of persons by the number of rooms in the unit. The tabulation form contained a terminal category of "10 or more" rooms; for purposes of the computation, the terminal category was given a mean value of 11. Similar data are available from the 1950 and 1940 Censuses.

Color (P5).-Occupied housing units are classified according to the color of the head of the household into two groups, white and nonwhite. Data for units with nonwhite houshold heads are published for areas having a specified number of such units (see section on "Description of tables"). The color group designated "nonwhite" consists of such races or ethnic groups as Negro, American Indian, Japanese, Chinese, Filipino, Hawaiian, Eskimo. Aleut, Korean, Asian Indian, and Malayan. Persons of Mexican birth or ancestry who are not definitely of Indian or other nonwhite race are classified as white. Persons of mixed racial parentage are classified as nonwhite.

The concept of race, as it has been used by the Bureau of the Census, is derived from that which is commonly accepted by the general public. The use of self-enumeration in the 1960 Census may have affected the accuracy of the data on color as compared with earlier censuses. Whereas formerly the classification was obtained in most cases by the enumerator's observation, in 1960 it was possible for members of the household to classify themselves.

Data on color of head of household are available from the 1950 and 1940 Censuses of Housing. Except for 1910, the number of units (families) by color of household head is available from the Censuses of Population from 1930 back to 1890 (data for 1910 being available only for the Southern States).

Heads of Spanish surname and Puerto Rican heads.-In order to provide data for housing occupied by Spanish- and Mexican-Americans for areas of the United States where most of these persons live, white household heads of Spanish surname were identified in five Southwestern States (Arizona, California, Colorado, New Mexico, and Texas). In all other States, Puerto Rican heads of households were identified. Puerto Ricans comprise persons born in Puerto Rico and persons of native parentage with at least one parent born in Puerto Rico (as determined by population items on place of birth). Native persons include persons born in the United States, the Commonwealth of Puerto Rico, or a possession of the United States. Data are published in the State chapters for areas having a specified number of units with white household heads of Spanish surname or units with Puerto Rican heads (see section on "Description of tables"). No data for these units are published in the United States Summary chapter.

Housing data were published in 1950, but not in earlier censuses, for units with white household heads of Spanish surname in the five Southwestern States. Housing data for units with Puerto Rican heads are not available in censuses prior to 1960.

Tenure (H12).-A housing unit is "owner occupied" (reported as "owned or being bought" on the enumeration forms) if the

fully paid for. The owner need not be the head of the household. A cooperative apartment unit is "owner occupied" only if the owner lives in it.

All other occupied units are classified as "renter occupied," including units rented for cash as well as units occupied without payment of cash rent. Units rented for cash (reported on the direct-interview form as "rented") are units for which any money rent is paid or contracted for. Such rent is commonly paid by the occupants but may be paid by persons not living in the unit-for example, a welfare agency. Units for which no cash rent is paid include units provided by relatives not living in the unit and occupied without rental payment, units provided in exchange for services rendered, and units occupied by a tenant farmer or sharecropper who does not pay any cash rent. "No cash rent" appears as a category in the rent tabulations. In county tables for ruralfarm units in the State chapters, the category appears under "rent status."

Essentially the same definitions of tenure were used in the 1950 and 1940 Censuses of Housing and in the Censuses of Population from 1930 back to 1890.

Year moved into unit (P12).-Data on year moved into unit are based on the information reported for the head of the household. The question refers to the year of latest move. Thus, if the head moved back into a unit he had previously occupied, the year he moved back was to be reported; or if he moved from one apartment to another in the same building, the year he moved into his present unit was to be reported.

The intent of the question is to establish the year the present occupancy of the unit began, as indicated by the year the household head moved into the unit. The year the head moves is not necessarily the same year other members of the household move, although in the great majority of cases the entire household moves at the same time. The statistics roughly reflect turnover in occupancy of units but do not necessarily indicate the total number of changes in occupancy that have occurred in a given period.

Data on year moved into the unit were not collected in censuses prior to 1960.

Vacant housing unit.-A housing unit is "vacant" if no persons are living in it at the time of enumeration. However, if its occupants are only temporarily absent, the unit is considered occupied. Units temporarily occupied entirely by persons having a usual place of residence elsewhere are classified as vacant (the unit at their usual residence is considered occupied). A vacant unit may be furnished or unfurnished; it may be offered for rent or sale; it may have been rented or sold but the new occupants have not moved in; or it may be held off the market for the owner's occasional or future use, for speculation, or for other

reasons.

Newly constructed vacant units are included in the inventory if construction has reached the point that all the exterior windows and doors are installed and the final usable floors are in place. If construction has not reached this point, the unit is excluded. Dilapidated vacant units are included, provided they are still usable as living quarters; they are excluded if they are unfit for human habitation. Vacant units are defined as unfit for human habitation if, through deterioration or vandalism, most of the doors and windows are missing and the floors are unsafe.

Vacant quarters are excluded from the housing inventory if there is positive evidence (a sign, notice, or mark on the house or in the block) that the unit is to be demolished. Vacant quarters condemned for reasons of health or safety so that further occupancy is prohibited are likewise excluded from the inventory. Also excluded are quarters being used for commercial or business purposes, quarters used for the storage of hay, machinery, business supplies, and the like; and vacant trailers.

With few exceptions, the same general instructions were used in 1950. In the 1960 Census, however, the instructions for enumerating certain vacant units were more specific than in 1950, particularly the instructions regarding units to be demolished, units unfit for human habitation, and units being used for nonresidential purposes. It is possible also that comparability is affected in some areas by the change from "dwelling unit" to "housing unit."

Information for vacant units was first collected in the 1940 Census of Housing. The counts and categories are not entirely comparable with those in 1960. The 1960 and 1950 counts are considered more inclusive.

Year-round or seasonal occupancy (H7).-Year-round housing units are units which are usually occupied or intended for occupancy at any time of the year. A unit used only occasionally throughout the year is considered a year-round unit. In resort areas, a unit which is usually occupied on a year-round basis is also considered a year-round unit.

Seasonal units are intended for occupancy during only a season of the year. Included are units intended for summer or winter recreational use, such as beach cottages and hunting cabins; units held for herders, loggers, and cannery workers; and units intended for migratory workers employed in farmwork during the crop season. Although units held for migratory workers were separately identified by the enumerator, they were included with other seasonal units in the tabulations.

Essentially the same definitions were used in the 1950 Census. In 1950, however, units which were temporarily occupied by persons having a usual residence elsewhere (classified as "nonresident" units) were shown as a separate category and, although they were treated as vacant units, they were not classified by year-round or seasonal use as in 1960. It is believed that most of the "nonresident" units in 1950 would have been classified as seasonal.

Data for year-round units are presented by condition and vacancy status in most tables. For seasonal units, however, no detail is shown.

Vacancy status (H13).-Status of a vacant unit is reported as of the time of enumeration. In the tabulations, the five categories described below apply to vacant units intended for yearround occupancy and in sound or deteriorating condition.

The count of available vacancies constitutes a measure of vacant units on the housing market. It consists of units which are for year-round occupancy, are in either sound or deteriorating condition, and are offered for rent or for sale. Excluded are seasonal units; dilapidated units; and units already rented or sold, held for occasional use, or not on the rental or sale market for other reasons.

The 1960 category "available" is comparable with the 1950 category "nonseasonal not dilapidated, for rent or sale." The separate categories "for sale only" and "for rent" for both years also are comparable since essentially the same concepts were used; in 1950, however, characteristics were not tabulated for the two groups separately, except for value and rent. The categories provided in the 1940 Census reports are not comparable with the 1960 or 1950 categories.

Units available for sale only are the available vacant units which are offered for sale only; they exclude units offered "for sale or rent." A vacant unit in a cooperatively owned apartment building is included if the individual unit is "for sale only." A vacant unit in a multiunit structure which is for sale as an entire structure is included if that unit is intended to be occupied by the new owner and if the unit is not also for rent.

Units available for rent are the available vacant units which are offered for rent and those offered for rent or sale at the same time.

The category tabulated as rented or sold, awaiting occupancy consists of vacant units which are for year-round occupancy and in either sound or deteriorating condition, and which have been rented or sold but the new occupants have not moved in as of the date of enumeration.

The category tabulated as held for occasional use consists of vacant units which are for year-round occupancy and in either sound or deteriorating condition, and which are held for weekend or other occasional use. The intent of this question was to identify homes reserved by their owners as "second homes." Because of the difficulty of distinguishing between this category and seasonal vacancies, however, it is possible that some second homes were classified as seasonal and therefore are included in the category "seasonal."

The category tabulated as held for other reasons consists of vacant units which are for year-round occupancy and in either sound or deteriorating condition, and which are held off the market for reasons not specified above. For example, the category includes units held for a caretaker or janitor, units held for settlement of an estate, and units held for personal reasons of the

owner.

The categories "rented or sold, awaiting occupancy," "held for occasional use," and "held for other reasons" when combined are comparable with the 1950 category "nonseasonal not dilapidated, not for rent or sale" (and some portion of the 1950 "nonresident" category).

Homeowner vacancy rate. The percentage relationship between vacant units available for sale and the total homeowner The total inventory is termed the homeowner vacancy rate. homeowner inventory consists of owner-occupied units and vacant units available for sale. This rate more adequately describes the sale market than a rate based on total housing units.

Rental vacancy rate.-The percentage relationship between vacant units available for rent and the total rental inventory is termed the rental vacancy rate. The total rental inventory consists of renter-occupied units and vacant units available for rent. This rate more adequately describes rental market conditions than a rate based on total housing units.

Duration of vacancy (H38).-The duration of vacancy is the length of time (in months) from the date the last occupants moved from the unit to the date of enumeration. The data, therefore, do not provide a direct measure of the total length of time units remain vacant. For newly constructed units which have never been occupied, the duration of vacancy represents the time period since the date construction was completed.

No information on duration of vacancy was collected in previous censuses.

Acres in the place (H17).—In the State chapters of Volume I, data on number of acres in the place are presented for vacant units in rural territory. The categories are "10 acres or more" and "less than 10 acres" (see section on "Farm-nonfarm residence").

Comparability with the Housing Vacancy Survey.-Data on many of the vacancy characteristics in Volume I are available for the United States (by inside and outside standard metropolitan statistical areas) from the Housing Vacancy Survey (HVS). This nationwide survey, covering a sample of approximately 42,000 units (occupied and vacant) is conducted monthly by the Bureau of the Census in conjunction with its Current Population Survey. Data are published quarterly in Current Housing Reports, Series H-111, Housing Vacancies. The concepts and definitions used in the HVS in 1960 were essentially the same as those used in the April 1960 Census.

Results showed that 9.1 percent of the total housing units in the United States were vacant according to the April 1960 Census and 10.2 percent were vacant according to the HVS for the first quarter 1960. These vacant units include units available for sale or rent, seasonal units, dilapidated units, and other types that are held off the market. Vacant units available for sale amounted to 0.9 percent of the total housing inventory in the April 1960 Census and 0.6 percent in the HVS; the rates for vacant units available for rent were 2.5 and 2.6 percent, respectively. Characteristics of available vacancies, such as median sale price, median rent, median number of rooms, and the proportion with all plumbing facilities agreed fairly closely.

Differences between the April 1960 Census and the HVS may be attributed to several factors. Sampling variability of the figures from the HVS and those based on a sample from the April 1960 Census is one factor. Other factors include survey techniques; methods used in weighting the sample estimates; the more extensive training, control, and experience of the HVS enumerators than of the census enumerators; and the use of hourly rate payments in the HVS and of piece-rate payments in the census.

Farm-nonfarm residence (H17, H18, H42).—In rural territory, occupied housing units are subdivided into rural-farm and ruralnonfarm units on the basis of number of acres in the place and total sales of farm products in 1959. All vacant units are included in the rural-nonfarm inventory. Number of acres in the place was reported for vacant units but no information was obtained on sales of farm products.

Occupied housing units are classified as farm units if they are located on places of 10 or more acres from which sales of farm products amounted to $50 or more in 1959, or on places of less than 10 acres from which sales of farm products amounted to $250 or more in 1959. The remaining occupied units, including those on "city lots," are classified as nonfarm units. (A house

is on a city lot if it is located in a community or closely built-up

XXX

cupied units for which cash rent is paid also are classified as nonfarm if the rent does not include any land used for farming, even though the place on which the unit is located qualifies as a farm.

For purposes of determining farm residence, the residential "place" consists of the land which the respondent considers to be the farm, ranch, or property. The place may consist of more than one tract of land; the tracts are usually adjoining although they may be separated by a road, creek, or another piece of land. If there is more than one house on the place, all have the same farm-nonfarm classification (except that, on farms, units for which cash rent is paid are always classified as nonfarm if the rent does not include any land used for farming). For example, if the operator lives in one house and a hired hand lives in a separate house on the same place, both housing units have the same classification.

Sales of farm products refer to the total amount (gross amount) received from the sale of crops, vegetables, fruits, nuts, livestock and livestock products (milk, wool, etc.), poultry and poultry products, and nursery and forest products produced on the place and sold at any time during 1959.

Farm-nonfarm residence in 1950 was determined for both occupied and vacant units in rural territory by the respondent's answer to the question, "Is this house on a farm or ranch?" Units on farms whose occupants paid cash rent for the house and yard only were classified as nonfarm, similar to the procedure in 1960. Units on institutional grounds and in summer camps and tourist courts were classified as nonfarm in 1950, regardless of the answer to the above question. Housing units in such places in 1960 are classified according to the answers to the questions on acreage and sales of products.

Farm residence was determined in the 1940 Census of Housing also. Counts of farm units (families) are available for each of the census years from 1930 back to 1890. Comparability is affected not only by the change in the definition of farm residence but also by the change in the delineation of urban and rural territory.

In 1950 and 1940, a count of farm dwelling units located in cities and other urban territory was obtained, but in the 1960 Census of Housing no effort was made to identify urban-farm units. In 1950, the count of urban-farm units (occupied and vacant) amounted to only about 96,000 in the United States as a whole.

The definition of farm residence adopted for 1960 employs more restrictive criteria than the 1950 definition. One reason for the change was to make the definition of farm residence essentially consistent with the definition of a farm used in the agricultural census, beginning with the 1959 Census of Agriculture. The effect of the 1960 definition is to exclude from the farm inventory, those housing units located on places considered as farms by the occupants but from which agricultural products are not sold or from which sales are below the specified minimum. On the other hand, the 1960 definition brings into the farm inventory a smaller group of housing units which meet the criteria but which are not considered by the occupants to be located on farms.

A test conducted in the Census Bureau's Current Population Survey (CPS) in March 1960 indicated that at that time the change in definition of farm residence resulted in a net reduction of 1.0 million occupied units (households) on farms, representing approximately one-fifth of the farm units under the 1950 definition. The CPS estimate of 4.1 million occupied farm units under the new definition, however, was 0.5 million greater than the April 1960 Census figure of 3.6 million. No similar estimates of the difference are available by State or smaller area. impact of the change in definition has probably been very uneven from one area to another.

The

Although there is no conclusive evidence on the relative validity

that in the April 1960 Census, several factors must be taken into consideration in evaluating the figure, 0.5 million. These factors include the more extensive training, control, and experience of the CPS enumerators than of the census enumerators; the use of hourly rate payments in the CPS and of piece-rate payments in the census; the use of self-enumeration in the census; the methods used in weighting the sample estimates; and the difference between the sampling variability in the CPS and in the 25-percent sample in the 1960 Census. There is also a difference between the definition of urban territory in the census and that in the CPS. In the March 1960 CPS, the boundaries of urban areas used were still those of the 1950 Census of Housing and did not include the annexations and other substantial expansions of urban territory that were incorporated in the 1960 Census of Housing. In the 1960 Census, the determination of farm residence was limited to rural territory as defined in 1960. The effect of this difference was to classify an unknown but presumably small number of units as rural farm in the CPS that are treated as urban in the reports of the 1960 Census. Finally, for some of the units in the CPS sample in March 1960, the determination as to farm residence had been made as early as December 1959, whereas the determination for the units in the April 1960 Census was made as of the time of enumeration. In view of the continued decline in the number of farms, it is probable that a number of places that qualified as farms in December would no longer have been so classified in April.

There are differences also between the number of farms as enumerated in the 1959 Census of Agriculture and the number of occupied rural-farm housing units from the 1960 Census of Housing. According to the 1959 Census of Agriculture, there were 3.7 million farms in the United States and 3.4 million farm operators living on the farms they operated. The number of occupied rural-farm housing units according to the 1960 Census of Housing (3.6 million) was 3.8 percent below the number of farms but 3.6 percent above the number of resident farm operators. Even if there had been no errors of enumeration, the number of occupied rural-farm housing units from the Census of Housing would not equal exactly either the number of farms or the number of operators living on the farms they operated. The Census of Agriculture, for example, includes farms in urban territory. Moreover, there are two or more occupied housing units on some farms and no occupied housing units on other farms. In addition, there may be occupied farm housing units even where there is no resident farm operator. Finally, the Census of Agriculture was taken in the fall of 1959, and evidence from other sources suggests that there was some decline in the number of farm units between the time of the Census of Agriculture and the Census of Housing. Data from a study in which schedules from the 1960 Sample Survey of Agriculture were matched with those from the 1960 Census of Housing suggest that approximately 13 percent of the units occupied by farm operators in rural areas were not classified as farm units in the 1960 Census of Housing. The nature of the matching procedure was such that it was not possible to identify and count the complementary group of units classified as farm in the housing census but as nonfarm in the agriculture survey.

STRUCTURAL CHARACTERISTICS

The comparison of the wording and explanations that appeared on the self-enumeration and direct-interview forms is given below. Type of housing unit and number of units in structure were always to be determined by the enumerator through observation and supplementary questions asked of the respondent. Number of rooms appeared as a self-enumeration item on the Advance Census Report, and all the other self-enumeration items appeared on the Household Questionnaire, which was used in two-stage The direct-interview questions appeared on a FOSDIC schedule. Although the question on year built appeared on the

areas.