Genealogical Research Using the Bureau of Indian Affairs' Census Rolls

Records of the Bureau of Indian Affairs, 1885-1940

As a reference archivist at the Washington DC location of the National Archives whose special knowledge is in the area of the records of the Bureau of Indian Affairs, I get many questions from people who are seeking to establish their Indian heritage. This search often leads the inquirer to the Indian Census Rolls, compiled by the Bureau of Indian Affairs, between 1885 and 1940. These records are microfilmed and are available at our regional branches as National Archives and Records Administration microfilm publication M595, in 692 rolls, and at some of the many state and local history and genealogy centers.

Sometimes there are questions about these rolls that are hard to answer. How was the agent to decide what persons should be listed on his census roll? What guidelines were given? How did he determine if someone should be on his list or not? What if the grandmother was living with them but she was from another tribe? What if they said they had a son away at school? How did the census relate to questions of enrollment or tribal membership? What was the agent supposed to do about Indians not living on the reservation- were they to be included? How could a person who was on the Flandreau rolls for the Indian census in the 20s and 30s, also have had children listed in a "street directory" at the same time, in Massachusetts. How would you find out why the children were not included in the Flandreau Indian Census Roll along with the father? Are there instructions? To answer these questions, the first thing I did was to locate the original act establishing the Indian Census rolls, to see what was intended.

Introduction to Indian Census Rolls

The original Act of July 4, 1884, (23 Stat. 76, 98) was vague, saying, "That hereafter each Indian agent be required, in his annual report, to submit a census of the Indians at his agency or upon the reservation under his charge.” The Act itself did not specify the collection of names and personal information. However, the Commissioner of Indian Affairs sent a directive in 1885 (Circular 148) reiterating the statement and adding further instructions: "Superintendents in charge of Indian reservations should submit annually, a census of all Indians under their charge." He told the agents to use the plan he had prepared for gathering the information. The sample there showed columns for Number (consecutive), Indian Name, English Name, Relationship, Sex, and Age. Other information on the number of males, females, schools, school children, and teachers was to be compiled statistically and included separately in the annual report.

The first form drawn up by the Commissioner asked only for name, age, sex, and family relationship. It was so little information that these Indian Census rolls were never considered to be “private” in the same sense as the federal decennial census, and there was never any restriction against the release of the information. Gradual changes in the form of the data required and special instructions for the census are documented in National Archives microfilm publication M1121, Procedural Issuances of the Bureau of Indian Affairs, Orders and Circulars, 1854-1955, in 17 rolls.

The censuses from 1885 on were compiled by the agents using forms sent by the Bureau. There was supposed to be only one census for each reservation, except in a few cases where part of the reservation was in another state. Multiple copies were not made. The original was sent to the Commissioner of Indian Affairs. The earliest censuses were written in by hand, but typing appeared quite early. Eventually the Commissioner issued instructions on exactly how to type some entries in, and requested that the family names be placed in alphabetical sections on the roll. For a while, a new census was taken each year and the entire roll redone. Agents were told in 1921 they were supposed to list all the people under their charge, and if a name was listed for the first time, or was not listed from the last year, an explanation was required. It was considered helpful to indicate the number for the person on the previous year's census. Persons also could be designated by a number peculiar to that reservation, if it was explained somewhere, or they could be listed as "N.E.", or “Not Enrolled.” In the 1930s, sometimes only supplemental rolls showing the additions and deletions from the previous year were submitted.

The regular process of taking the Indian censuses was discontinued in 1940, although a few later rolls exist. A new Indian Census was taken by the Census Bureau in 1950, but it is not open to the public.

Naming - English or Indian Names

There were no instructions with the earliest census forms, other than to include a census of all Indians under the agent’s charge, but the Commissioner did occasionally issue a statement about the census. Primarily he urged the agents to get the information and send it in on time, without much comment. The early instructions just said to include family groups with all the people living in each household. The agent was instructed to list the Indian and English names of the head of the household and the names, ages, and relationship of the other family members. The column for Indian Name continued, but in fact, Indian names were falling out of usage and were seldom included after about 1904.

A directive in 1902 gave suggestions for how to translate Indian names to English in what would now be termed “politically correct” fashion. The usefulness of having all the family members share the same surname was pointed out, especially for the purposes of property or land ownership, so that children and wives would be known by the names of their fathers and husbands in questions of inheritance. The agents were told not to simply substitute English for the native language. It was suggested that a native name be retained as much as possible, but not if it were too difficult to pronounce and remember. If it were easily pronounced and mellifluous, it should be retained. Names of animals could be translated to the English version, such as Wolf, but only if the Indian word was too long and too difficult. “Foolish, cumbersome or uncouth translations which would handicap a self-respecting person should not be tolerated.” Complex names such as Dog Turning Round might be better rendered, for example, as Turningdog, or Whirlingdog.

Derogatory nicknames were to be dropped.

The Agent’s Jurisdiction—Who Was Included?

For years little guidance was given to help the agent determine whom to include. In 1909, he was asked to show how many resided on the reservation and how many allotted Indians were living on their allotments. That information was not included on the census roll itself, but as part of the annual report. He was urged to take pains to make the numbers accurate.

It wasn’t until 1919 that any clarifying instructions about whom to include were added. The Commissioner directed superintendents and agents in Circular 1538, "In enumerating Indians who are not attached to your jurisdiction, they should be classified by tribal affiliations, in which case they should be designated by approximate blood relationship." He was referring to people living in the jurisdiction, but not from that reservation or tribe, rather than people not present and living off reservation. If they were listed with a family, the agent should tell what family relationship they bore to an enrolled person, and what tribe or jurisdiction they actually belonged to. The Commissioner pointed out that both parents might not be members of the same tribe, for example, one Pima and one, Hopi. The parents had the right to determine with which tribe the children should be identified, and agents were instructed to show the parents’ selection as the first one, with a hyphen and the second tribe, as in Pima-Hopi.

Very likely the only thing new by 1919 was to be sure to indicate the formal tribal affiliation of all. Formerly it might simply have been assumed from the census that the grandmother living with the family was actually a member of that tribe and reservation. Or she might not have been listed, because she really did belong with another tribe. Or if more than one tribe resided within a jurisdiction, the distinction might not have been made. In urging accuracy, the Commissioner said in 1921, “It does not seem to be generally appreciated that the census rolls are often the basis of the property rights of the Indian enrolled. An allotting agent looks to the census roll to determine who are entitled to allotments. An examiner of inheritances secures much of his information …from the census rolls.” (Circular 1671). But in many ways it was the decision of the Superintendent or Agent as to whether someone should be included in the census.

Changes to the Indian Census

Between 1928 to 1930 that the BIA Indian Census underwent a real change. The format was changed, there were more columns, new information required, and instructions printed on the back. The forms used for 1930 and thereafter showed the following columns 1) Census number-Present, 2) Last, 3) Indian Name –English, 4) Surname, 5) Given, 6) Allotment, Annuity Identification Numbers, 7) Sex, 8) Date of Birth - Mo., 9) Day, 10) Year, 11) Degree of Blood, 12) Marital Condition (M, S,) 13) Relation to Head of Family (Head, Wife, Dau, Son). The format was changed to the wider landscape orientation of the page.

Reservation and Nonreservation Indians

One important change for 1930 concerned people who did not live on the reservation. The understanding was that the agent was to include all his enrollees, whether there on the reservation or elsewhere, and no residents who were enrolled on another reservation. They should be recorded on another agent’s list.

Circular 2653 (1930) says "A special survey of absentees is to be made at each jurisdiction and their addresses determined." The Commissioner goes on to say, "names of Indians whose whereabouts have been unknown for a considerable number of years are to be dropped from the rolls with the approval of the Department. The same pertains to bands of Indians of whom no census has been made for an extended time and who have no contact with the Service, viz., the Stockbridges and Munsees, the Rice Lake Chippewas and the Miamis and Peorias. These will be enumerated in the 1930 Federal census."

Cooperation with the federal officials who were conducting the 1930 decennial census was requested, but it is clear they were two different censuses taken in the same year, by two different government bureaus, with different instructions. However, some 1930 BIA censuses have penciled information that may correlate to the federal 1930 census data. For example, the 1930 census for Flandreau has handwritten numbers in the columns for county. The instructions shed no light on this. But, the since same number appears sometimes with several names having the same surname, it looks like it could be the family number from the federal census for that county, or perhaps a postal code or other correlating number. Although the agents were cooperating with the federal census takers, they were taking their own census. If the federal census takers figured the number of Indians counted on a reservation as a member of a tribe, they did not want to recount the same people living off reservation. Sometimes there might be notes done on the form to check off and make sure that people were not being counted twice.

The Commissioner directed the superintendents in Circular 2676 that the "census must show only Indians at your jurisdiction living on June 30, 1930. Names of Indians removed from the rolls since the last census, because of death or otherwise, must be entirely omitted." A later amendment altered this to state, “The census must show only Indians enrolled at your jurisdiction living on April 1, 1930. This will include Indians enrolled at your jurisdiction and actually living on the reservation, and Indians enrolled at your jurisdiction and living elsewhere.” He was still hammering on this theme in Circular 2897, when he said, “Dead Indians reported on Census Roll as was done by some agencies last year will not be tolerated.” He also took care to define the meaning of the Superintendent's area of jurisdiction to include “Government rancherias and public domain allotments as well as reservations.”

The agents were urged to be careful to remove names of those deceased, and to include names of those who were still “under their jurisdiction” but perhaps on a rancheria or public domain allotment. The implication is that the information for previous years could be erroneous. Also it is clear that the jurisdiction did include some people living on allotments in the public domain, whose lands were no longer considered as a part of a reservation. However, spouses of Indians who were themselves not Indian, are not listed. Charles Eastman’s wife, a non Indian, does not appear on the Flandreau census along with her husband.

By 1930 many Indians had gone through the allotment process and received patents for their lands, now considered as part of the public domain, as opposed to lands reserved for a reservation. Agents were told to consider Indians living on allotted lands on the public domain as part of their jurisdiction. Some censuses made that distinction, reservation and nonreservation Indians. For example, the Grande Ronde - Siletz present day membership criteria mentions the “public domain” rolls of 1940 prepared by the Grand Ronde-Siletz Agency, Bureau of Indian Affairs.

A revised census form was used in 1931, prompting the Commissioner to give further instructions in Circular 2739. The 1931 census had the following columns: 1) Number 2) Name: Surname 3 )Given Name 4 )Sex: M or F 5) Age at Last Birthday 6 )Tribe 7) Degree of Blood 8) Marital Status 9) Relationship to Head of Family 10) At Jurisdiction where Enrolled, Yes or No 11) At Another Jurisdiction, [its]Name 12) Elsewhere, Post Office 13) County 14) State 15) Ward, Yes or No 16) Allotment, Annuity, and Identification Numbers

The members of a family were defined as 1, Head, father; 2, wife; 3, children, including step children and adopted children, 4, relatives, and 5, "other persons living with the family who do not constitute other family groups." A grandparent, brother, sister, nephew, niece, grandchild, or any other relative living with the family should be listed and the relationship shown. A column was included to list roomers or friends living with the family, if they were not listed as heads of households on another census sheet. A single person living at home could only be a "Head" if the father was dead and the oldest child was serving in that capacity. The agent was also told to report all tribes making up the jurisdiction, not just the predominant one.

Further instructions on residence said, If a person resided at the reservation, column 10 should say Yes, and columns 11 through 14 left blank. If an Indian resided at another jurisdiction, column 10 should be No, and column 11 should indicate the correct jurisdiction and state, and 12 through 14 left blank. "When Indian resides elsewhere, column 10 should be NO, column 11 blank, and columns 12, 13, and 14, answered. County (column 13) must be filled in. This can be obtained from the Postal Code." Children at school but technically still part of their families were to be included. They were not to be reported at another jurisdiction or elsewhere.

There is evidence that the census takers were unclear themselves on whether to list someone who was not present. The Commissioner kept after them about mistakes. "Please see that columns 10 to 14 are filled in as directed, as two people spent over two months correcting the errors in these columns last year."

Roll Numbers—Is it an "Enrollment Number?"

The number in the earliest censuses was a consecutive number that could change from one year to the next for the same person. Although agents had been asked as early as 1914 to tell the roll number on the previous roll especially in the case of alterations, they were specifically asked in 1929 to indicate what number the person was on the previous roll. It seemed that 1929 became the benchmark number in some cases, and the person continued to be defined by that number on future rolls. Instructions for the 1931 census said: “List alphabetically, and number names on roll consecutively, with no duplicate numbers…” That set of numbers was followed by the column indicating the number on the previous roll. In most cases, the “ID number” was that: the consecutive number on the 1929 roll. So there was a new Consecutive Number each year, and an Identifying Number from a base roll, and an Allotment Number, if the allotting had been done. Using Flandreau as an example, in year 1929 the “allot-ann-id numbers” (in unnumbered column 6) given are identification numbers starting from 1 to 317 end, and these id numbers correspond exactly to the column for the present order on the list.

So, the id number was derived from the order on the list in 1929, and was carried over to subsequent years. In 1930, the id number was that 1929 consecutive order number.

The Concept of Enrollment

It is clear that by this time, there was an accepted concept of “enrollment” being employed, even though there were no official membership enrollment lists existing for many tribes. A few tribes had been involved in government supervised enrollment lists, usually relating to legal questions in which the federal government owed the tribe moneys as determined by the courts. In that case, the federal government had a vested interest in determining who was a legitimate member, to whom money was owed, and who was not. Apart from those special cases, the Superintendents and Agents had been occupied for years with the allotment process, identifying those who were eligible to receive an allotment, and they had been involved yearly in the distribution of goods and money and checking the eligible names off an annuity roll. Many tribes had accepted Annuity Roll numbers, and Allotment Roll numbers. At the discretion of the Superintendent, those that did not could have an assigned Identifying Number.

So, the concept of eligibility for services was apparently equated to a status of enrollment even if there were no actual enrollment list. The questions of eligibility were tied to allotment lists, annuity rolls, and prior census rolls.

The landscape changed again in 1934, when legislation was passed called the Indian Reorganization Act. Under this act, tribes were encouraged to specifically set up a constitution that gave recognized criteria for determining membership and enrollment. A quick survey of Indian Tribal Constitutions on the Internet shows that a number actually did adopt the BIA census as the base roll, for membership.

Degree of Blood

Degree of blood was not required on the early rolls. When it was included, for a short period, blood quantities were artificially compressed into only three categories that may have led to confusion in later years when more specific categories were required. The 1930 Indian census did not allow more than three distinctions to be made in amount of blood because a mechanical reading device was to be used. Circular 2676 (1930) said about the new census form, Form 5-128, that it “must be filled out in absolute conformity to instructions on reverse. This ruling is necessary because a mechanical device has been installed in the Office for tabulating the data….Thus for degree of blood then symbols F for full blood; ¼+ for one fourth or more Indian blood; and – ¼ for less than one fourth. No substitution of more detailed information is permissible in any column.” Later, in 1933, the agents were told to use the categories F, 3/4, ½, 1/4, 1/8. Still later, they were urged to be exact if possible.

If someone were going to use the 1930 blood quantum information in retrospect it could lead to mistakes. Obviously, you can’t go from an artificially compressed category and return with greater detail, and be accurate.

Accuracy of the Indian Censuses

What can be said in retrospect about the accuracy of the Indian Censuses? Even with the instructions, agents were sometimes confused as to whether they should list the names of people who were away. If the agent had the address, and knew the person was still maintaining ties with the family, he would probably consider the persons as still under his jurisdiction, and count them in his census. But if persons had been away for several years, the agent was supposed to remove them from the roll. He was supposed to report the reason the person was removed and get the OK from the Commissioner. The Commissioner instructed the agents to remove the names of people who had died, or who had been away for years. He was very annoyed at the agents for failing to be accurate. His constant harping suggests there were continuing inaccuracies. In the end, the Indian Census Rolls may, or may not be considered a list of all those people who were officially considered "enrolled." Some tribes did adopt them as a base roll.

But, it is also clear that the numbers had varying meaning. Very likely you could, at least by the mid 1930s, equate the presence of a name on a roll as indicating sustained presence in the tribal jurisdiction of that Agent with a status of membership understood. As early as 1914, the Commissioner started asking that the numbers on the roll should indicate the number of the person on the roll the year before. That indicates that although the roll was freshly numbered each year, with minor variations gradually occurring due to births and deaths, it was nevertheless reflective of a continuous group of people. This is the way most rolls look, until the 1930 changes.

Understanding the Indian Census—An Example

How could a person who was on the Flandreau rolls for the Indian census in the 20s and 30s, also have had children listed in a "street directory" at the same time, in Massachusetts?

There are several possibilities. Theoretically, if the children were living in his household on the reservation, they should have been counted as members of his family on the BIA census. This is also true, if the children were away attending school, but lived with him otherwise; they should have been counted. If he was separated from his wife and she took the children to Massachusetts, they would be part of her household and would not be counted on the reservation census with the man. If she was not an enrolled member of that tribe or reservation and lived away with her children, she would not be counted, nor the children, in the agent's count for the census of that reservation for that year. If the mother was a member of a different tribe or reservation, the children might have been counted on that other reservation's census. Agents were instructed to list people who lived on the reservation but were not members of that tribe. But they were not counted in the total census count. The point was that a person should not be counted twice, and the agent had to include some information that would help resolve the issue.

They were supposed to indicate what tribe and which jurisdiction the person was from. They usually would give the general address of people who were away. When the census was submitted, it would be easier to figure out if someone had been left off of one or included on another when they shouldn't be. The Commissioner of Indian Affairs was less concerned about factual names than concerned that the total number be accurate. That is not to say that the exact identity of persons was not important; it was. The Commissioner noted that the censuses would be useful in making annuity rolls, and in determining issues of inheritance, so he wanted them to be correct.

Free Online Access To Indian Census Rolls

Access NARA microfilm M595 (Native American Census Rolls, 1885–1940) online for free as digitized images at Internet Archive.