The Effect of Land Allotment on Native American Households During the Assimilation Era∗

Home , Bertrand (steamboat), Burke Act, Dawes Act, Diminishment, Indian colony, Paper township

Christian Dippel† Dustin Frye‡

July 5, 2020

Preliminary and Incomplete

Abstract

In the early twentieth century, the federal government broke up millions of acres of tribally owned reservation lands and allotted them to individual Native American households. The allotted land was initially held in trust (with limited property rights), and households had to prove themselves “competent” in order to be given full ‘fee- simple’ property rights. In short, allotment was a conditional transfer program aimed at cultural assimilation. We study how Native American households responded to the incentives created by allotment.

Keywords: Indigenous Economic Development, Cultural Assimilation, Conditional Transfer Pro- grams, Property Rights, Record Linkage JEL Codes: N10, O1, Z1

∗We thank Randall Akee, Terry Anderson, Leonard Carlson, Marie Duggan, Donna Feir, Paola Giuliano, Miriam Jorgensen, Bryan Leonard, Adriana Lleras-Muney, Mindy Miller, Steve Nafziger, Dominic Parker, Matthew Snipp, and seminar participants at the Hoover Institution, WEAI conference San Francisco, the EHA conference Atlanta, the NBER Ipf Meetings, and Claremont McKenna for helpful conversations. We thank David Cruse, Michelle Garabetian, Adam Krueger, Jason Kwan, Zach Lewis, Nika McKechnie, Josh Mimura, Lilian Stanwick, Diana (Zhoutong) Wu, Nina Zacharia and especially Chengcheng Zhang for excellent research assistance. We thank Matthew Gregg for sharing reservation-year aggregate schooling data. Dippel gratefully acknowledges support from a UCLA Senate COR grant. †University of California, Los Angeles, CCPR, and NBER. ‡Vassar College “The time has arrived when we should deﬁnitely make up our minds to recognize the

Indian as an individual and not as a member of a tribe. The General Allotment Act is a

mighty pulverizing engine to break up the tribal mass. It acts directly upon the family

and the individual.” — Teddy Roosevelt’s Address to Congress, 1901

1 Introduction

Toward the end of the 19th century, with the conclusion of the Indian Wars and the closing of the frontier, reformers and the U.S. government turned their attention towards the cultural assimilation of Native Americans, ninety percent of whom were living on the reservations created in the previous decades. This is signiﬁed by the famous 1892 quote: “kill the Indian in him, and save the man.”1 Assimilation efforts were centered on land allotment policies that broke up tribally owned reservation lands into individually owned homestead-sized land allotments. As the Bureau of In- dian Affairs (BIA) commissioner noted: “if there were no other reason [for allotment], the fact that individual ownership of property is the universal custom among civilized people of this country would be a sufﬁcient reason for urging the handful of Indians to adopt it.” Allotment was the cor- nerstone of federal Indian policy beginning with the passing of the General Allotment (or ‘Dawes’)

Act in 1887 until it ended with the passing of the Indian Reorganization (or ‘Howard-Wheeler’)

Act (IRA) in 1934 (Carlson, 1981, p18).

When a reservation was opened for allotment, all families on the reservation were given allotments, and these allotments were held in a trust managed by the local Indian agent (the BIA’s local superintendents in charge of a reservation). Trust-status meant allottees could not sell or collater- alize an allotment. In order to obtain full ‘fee-simple’ legal title for their allotment, allottees had to be declared “competent” by the BIA agent (Carlson, 1981; Banner, 2009; Otis, 2014). In short, In- dian allotment was designed as a conditional transfer program aimed at cultural assimilation. The

ﬁrst ‘treatment arm’ was an unconditional transfer program: receiving an allotment gave the allottee the unconditional right to use the land for their own purposes, as well as the right to leasing rents. The second treatment arm was only obtained conditional on proving one’s “competence.”

1 Quote from a speech by Capt. Richard Pratt, founder of the ﬁrst Indian boarding school. Appendix-Figure A1 shows one of the many “before/after” pictures one ﬁnds in association with the Assimilation Era.

1 Allotment’s conditional transfer arm (full title) was worth almost 20 times annual per capita incomes in our data, orders of magnitude larger than modern-day conditional transfer programs.

Our paper is an investigation into how individual households responded to the incentives created by this program.

We hypothesize that individual allottees responded to the allotment policy’s incentive structure by signalling cultural assimilation to the BIA agents in order to be able to obtain full property rights over their allotments. First evidence of this comes from an annual panel of reservation- aggregate data from the BIA’s annual reports from 1911 to 1934. In addition to schooling, these data include very direct measures of assimilation or assimilation-signalling, namely the number of “church-going Indians” and of those “wearing civilized dress.” We combine these data with the universe of Indian allotments, which the Bureau of Land Management (BLM) has digitized with geo-location and issuance year. In a within-reservation over-time comparison, we ﬁnd that school-attendance, the number of church-going Indians and the number of those wearing civilized dress increased in lock-step with the expansion of allotment, even after controlling for potential changes in school and clerical infrastructure.

To provide more ﬁne-grained evidence, we work with the historical Full Count Census data, which enumerated around 300,000 Native Americans in each decade from 1900–1940.2 In these data, we focus on individual farming and schooling choices as the best measures of assimilation or assimilation-signalling. The main challenge working with these data is the fact that the census reports only Native American race, but neither tribe nor reservation.3 This matters because every historical policy aimed at Native Americans varied considerably across reservations, in both extent and phase-in. We address this problem by constructing a novel reservation-to-individual crosswalk, using spatial matching techniques based on a combination of Census enumeration districts and towns.4 With these, we assign reservations to around ninety percent of Native Amer- icans in the 1930 and 1940 census waves. The unassigned are essentially all ‘urban Indians,’ so that we effectively assign one-hundred percent of the Native Americans living on reservations at

2 The vast majority of Native Americans were not enumerated in the census before then. 3 Until 1980, tribe was only once enumerated on the regular Population Census form, in 1930. But even in 1930 tribe is not among the variables digitized in the Full Count census. Today’s recognized tribes are, with a few exceptions, synonymous with reservations. We can usually clearly trace them to a single cultural/ancestral tribe (e.g., White Earth is an Ojibwe reservation, Rosebud is a Sioux reservation). 4 We generate a stable crosswalk from the Census’ historical person-id to reservation/tribe, which will be shared with other researchers.

2 the time. This allows us to investigate the response to allotment in the universe of on-reservation

Native Americans. [We are now building the crosswalk for 1900–1920.]

Figure 1: Treatment Arms of Indian Allotment

Reservation 1 Reservation 2

Fee Unalloted

Trust

Had the policy been allowed to run its course, all reservations would have been allotted eventually, and all allottees would have eventually received full legal title to their land. However, the policy was abruptly ended by the 1934 IRA, and as a result, roughly half of all reservations were never allotted, and roughly half of all allotments were never converted into fee simple. Figure1 illustrates the resulting variation: on the left is an allotted reservation, on the right an unallotted one; within the allotted reservation there are households whose allotments are in the original trust and those whose allotments have been converted to fee simple.

Allotment is measured as a reservation-aggregate intent-to-treat effect because the census includes no allotment information. This intent-to-treat includes a start-year (almost everyone on a reservation received their allotments within a short period of time), and a gradually increasing share of allotments converted into fee simple. Once we have the 1900–1920 crosswalks fully constructed, the core of our empirical analysis will be a pseudo-panel that investigates the effect of allotment over essentially the entire allotment period. Working with the 1930 and 1940 data for now, we observe only a shorter two-period pseudo-panel at the tail end of the allotment policy.

We ﬁnd that allotted households had higher school attendance and were farming at a higher rate, but that this difference compressed signiﬁcantly from 1930 to 1940, i.e. after the IRA ended the incentive to signal assimilation.

Fortunately, 1940 is the ﬁrst year to include educational attainment data, which allows us to perform a cohort analysis in which we utilize the differential phase-in of allotment timing across reservations to ask whether school-attendance increased with allotment, comparing the same birth

3 cohorts across reservations. The cohort analysis also allows us to test whether school-attendance receded with the conversion to fee simple. This is critical because it allows us to distinguish assimilation signalling from assimilation.

To get individual household-level allotment information for at least a subset of the data, we record-link the Native American population in the census to a completely separate BIA-run census called the Indian Census Rolls (ICR). For the linked sample, we can assign actual allotment information to households, including the date of their allotment and when (if ever) it was converted into fee simple. [Discussion of results to be added here.]

Lastly, we investigate whether the effect of allotment depended on tribes’ ancestral norms of having individual (rather than collective) property rights, a variable we code from the Ethnographic

Atlas (EA). [Discussion of results to be added here.]

Our paper speaks to a literature on the effects of cultural assimilation policies. Most such policies are designed to achieve cultural assimilation through education (Alesina, Giuliano, and Reich

2013; Sakalli 2017; Bandiera, Mohnen, Rasul, and Viarengo 2018; Fouka 2019).5 In contrast, we study a policy that was designed to achieve cultural assimilation through the allocation of private property. Allotment was a conditional transfer program and the program’s conditional arm (full legal title) was worth almost 20 times annual per capita incomes in our data, orders of magnitude larger than modern-day conditional cash transfer programs (Schultz, 2004; Barrera-Osorio,

Bertrand, Linden, and Perez-Calle, 2011; Behrman, Parker, and Todd, 2011; Parker and Todd, 2017;

Barrera-Osorio, Linden, and Saavedra, 2019). We ﬁnd that this program design provided power- ful incentives for individuals to signal their assimilation in order to obtain full legal title to their allotments.

Our paper also contributes to a growing literature emphasizing the importance of cultural norms as drivers of economic outcomes and decision making (Algan and Cahuc, 2010; Fernandez´ ,

2011; Nunn, 2012; Costa-Font, Giuliano, and Ozcan, 2018; Dohmen, Enke, Falk, Huffman, and

Sunde, 2018; Enke, 2019). To the best of our knowledge, our paper is the ﬁrst that documents the

5 Assimilation through education was also prominent on reservations, albeit mostly before the allotment era. Gregg (2018) shows that reservations that historically sent a higher share of children to boarding schools have more assimilated populations today. At a more micro level, Feir(2016b) shows that Native students of (Canadian) boarding schools were more likely in the long run to live off-reserve and not speak an indigenous language. Feir(2016a) ﬁnds evidence for a cultural “backlash” to boarding schools that is similar to ﬁndings in Fouka(2019).

4 importance of cultural traditions of private property in an economic setting.6

2 A Note on the Welfare Effects of Indian Allotment

Before proceeding, we wish to acknowledge that Indian allotment is rightly viewed critically today. The most obvious reason is that it was a colonial policy aimed at erasing Native American culture, and at undermining tribes as polities.

Another reason is that the allotments that were transferred to individual tribal members were taken from the tribe. In short, Indian allotment ‘robbed Peter to pay Paul’.

A third reason is that allotments led to a dramatic decrease in the land base of Native American tribes, primarily through the sale of ‘surplus land’ that was “left over” after all households had been allotted. See Appendix-Figure A2 for a sales ad for such surplus land. Prior to the Dawes

Act, Indians controlled over 138 million acres of lands within their reservations. By 1934, Native land holdings had fallen to 52 million acres (Ofﬁce of Indian Affairs, 1935). Moreover—regardless of the lawmakers’ intentions—there was clearly corruption in the implementation of allotment, and land speculators (possibly in cahoots with Indian agents) almost surely tried to inﬂuence the allotment process for personal gain.

A fourth reason is that in-trust allotted land was kept in trust in perpetuity after 1934. Under the ill-conceived inheritance rules determined by the BIA, this in-trust land had to be bequeathed to all of an allottee’s heirs in multiple, undivided interests. Over time, this led to a steady increase in claimants on these lands, which in turn has created large inefﬁciencies in land use for today’s heirs of the original allottees whose land remained in trust (Leonard and Parker, 2020; Dippel,

Frye, and Leonard, 2020).

None of these considerations has a direct bearing on our analysis, and none of our results should be interpreted as taking away from any of them.

6 A related question that we do not explore is whether some tribes’ (or even households’) norms in relation to private property may have changed in response Indian allotment. Di Tella, Galiant, and Schargrodsky(2007) provide evidence for such a mechanism, and the seminal economics paper on ‘property rights’ uses Native American tribes as an illustration of how such rights may be endogenized (Demsetz, 1967).

5 3 Allotment and Assimilation

Allotment: Following the establishment of the reservation system, American Indian reformers considered land allotment as a requisite element in the assimilation of American Indians (Otis,

2014).7 Early allotment efforts were negotiated into treaties, but legislation implementing a general allotment act stalled for many years before Henry Dawes introduced a modiﬁed allotment bill to the Senate in 1886. The bill quickly passed the Senate and the House, and on February 8,

1887, President Grover Cleveland signed the Dawes General Allotment Act into law. Heads of household received 160 acres, single persons over 18 received 80 acres, orphans under 18 received

80 acres. Allotments were mandatory and anyone not selecting an allotment within the ﬁrst four years, would be assigned a parcel by the Indian Agent. Unallotted reservation land was desig- nated as surplus and made available for outside settlement; see e.g. Appendix-Figure A2. Imple- menting the Dawes Act on an individual reservation was a complicated process. First, the BIA agent in charge of the reservation was tasked with determining the list of eligible tribal members entitled to an allotment and the household structure for every household within the reservation

(Banner, 2009). These agents were also tasked with managing the surveying of the reservation and its division into parcels. Allottees could select a parcel, but usually did not, in which case the local

BIA agent determined the assignment of allotments (Banner, 2009; Otis, 2014; Carlson, 1981). The

Meriam Report characterized the process as follows: “The original allotments of land to the Indians were generally made more or less mechanically. Some Indians exercise their privilege of making their own selections; others failing to exercise this right where assigned land. Often Indians who exercise the privilege made selections on the basis of the utility of the land as a means of continuing their primitive mode of ex- istence. Nearness to the customary domestic water supply, availability of firewood, or the presence of some native wild food were common motives. Few were sufficiently far sighted to select land on the basis of its productivity when used as the white man used it”(Meriam, 1928, p470).8 Each allotment was given an allotment number and a patent was filed with the Government Land Office (GLO) upon approval by the President. These patents—subsequently digitized by the Bureau of Land Management (BLM)— specified the trustee, the specific plot location, the date, and the unique allotment number. From

7 The two most prominent reform organizations were the Indian Rights Association, founded in 1882, and the National Indian Defense Association, formed in 1885. 8 Unsurprisingly, there are also accounts of outside settlers inﬂuencing agents to set aside the highest quality land for surplus (Otis, 2014, p145).

6 1887–1906, constraints over personnel and resources slowed surveying and the allocation of allotments. This changed with the introduction of the Burke Act, which dramatically increased the pace of allotting. This is clearly visible in the ﬂow of allotments over time in Appendix-Figure A3, which we constructed from the BLM GLO data. Once selected, allotments were approved by the

Secretary of Interior and each household or individual was issued a trust patent. This ‘trust patent’ held the allotted land in trust for a trust period, during which the allottee or their heirs were the beneﬁciary of the allotment. Land held in trust could not be alienated or sold (and therefore nor collateralized) and was not subject to state or local taxes.

Assimilation: Initially, land was to be held in trust for a ﬁxed period , after which the allotment would be transferred to the owner in sfee simple. With the 1906 Burke Act, local BIA agents were given dramatically increased authority over the process of converting land from trust status to fee simple. The key determinant in this decision was whether the allottees had proven themselves

“competent” in the eyes of the agent. Cultural assimilation was meant to be the key consider- ation in agents’ determination of competency, because the aim of allotment was “the enlightened self interest of the individual Indian. Freed from the binds of tribal customs and authoritarian chiefs, the individual would soon want to accumulate wealth and property and, as he progressed economically, would acquire the habits and customs of Christian society. The key was to be private property”(Carlson, 1981, p80). The aim of cultural assimilation was reﬂected even in the ritual of converting allotments to fee simple. McDonnell(1980, p26) describes how this process played out at the Yankton Reserva- tion in 1916: “Each Indian who was to receive a patent [i.e. in fee simple] stepped from a tepee and shot an arrow to signify that he was leaving behind his Indian way of life. He placed his hands on a plow to show that he had chosen to live the farming life of a white man, with sweat and hard work. The secretary of the interior then handed the Indian a purse as a reminder that he must save what he earned.”

The activities that BIA agents looked on favorably in their decision to convert allottees’ land to fee simple were (i) farming, (ii) sending children to school, and (iii) going to church (Carlson 1981, ch4, Golenko 2010). From the allottee’s point of view, an allotment could be valuable even when held in trust. According to the 1920 Annual Report of the Commissioner of Indian Affairs, annual leasing from a typical 160-acre allotment would have generated 150 dollars of proceeds, around

50 percent of average household income. Yet, obtaining the land in fee-simple was signiﬁcantly more valuable because the average sales value of a 160-acre allotment in the same data was almost

7 Figure 2: Allotments, Church-Going and Civilized Dress

Church-Going Indians & #Allotments (dashed) #Indians Who Wear Citizens' Dress & #Allotments (dashed) 90000 160000 70000 70000

80000 150000

60000 60000

70000 140000

50000 50000 (sum) Churchgoing

60000 (sum) cum_trustpatents 130000 (sum) cum_trustpatents (sum) NumberIndianswhoWearcitizens 50000 40000 120000 40000 1912 1914 1916 1918 1920 1912 1914 1916 1918 1920 year year

Notes: The left panel of this ﬁgure shows two time-series: the evolution of the aggregate number of “Church-going Indians,” reported on the BIA’s annual reports for 1911–1920, and the evolution of the aggregate number of allotments recorded in the BLM GLO database on the same set of reservations. The right panel of this ﬁgure shows the aggregated number of “Indians Wearing Civilized Dress”, also reported on the BIA’s annual reports for 1911–1920, and

20 times that at 2,800 dollars.

These numbers lead us to investigate how allotted American Indian households responded to this conditional cash program. One hypothesis is that households may have rationally tried to signal their cultural assimilation in order to improve their chances of being declared “competent” by the local BIA agent. In the next sections, we will study micro-data of Native American households in the Full Count Census. Unfortunately, the most direct measures of cultural assimilation, namely church-going and the wearing of “civilized dress,” are not enumerated in the Census. Fortunately, those measures were reported in BIA annual reports from 1911 to 1920. Figure2 shows how the totals of these measures co-evolved with allotment over that period. Both measures were reported by reservation, so that we have the panel structure needed to investigate the relationship between allotment and assimilation measures in a sharper way that parses out cross-sectional differences in a generalized difference-in-difference relationship.

This is done in Table1: we ﬁnd that the measured expansion of allotment in a reservation over time strongly correlated with these two measures of cultural assimilation. This remains true after conditioning on the time-varying supply of churches and missionaries on reservations over time.9

We also ﬁnd that school attendance increased over time with allotment. The results in Table1 are important because the aggregate data is consistent with a “signalling of cultural assimilation”

9 Appendix-Figure A4 shows the evolution of churches and the number of missionaries over time.

8 Table 1: Allotments and Measures of Assimilation

(1) (2) (3) (4) (5) (6) (7) (8) (9) Panel A, outcome: # Church-going Indians, 1911-1920 Total Alloted Acreage 0.101*** 0.084*** 0.080*** 0.099*** 0.082*** 0.078*** 0.080*** 0.065*** 0.061** / 160 [0.000] [0.002] [0.005] [0.000] [0.000] [0.001] [0.001] [0.010] [0.019] year 22.031*** 21.427*** 20.281*** [0.000] [0.000] [0.000] Missionaries -6.128 -5.718 -5.684 -6.494 -6.089 -6.048 [0.246] [0.303] [0.305] [0.173] [0.229] [0.233] Churches 21.346* 20.352* 20.713* [0.066] [0.084] [0.069] R-squared 0.963 0.964 0.964 0.963 0.964 0.964 0.964 0.965 0.965 Panel B, outcome: # Indians Wearing Civilized Dress, 1911-1920 Total Alloted Acreage 0.145*** 0.057 0.052 0.148*** 0.059 0.055 0.105** 0.033 0.029 / 160 [0.003] [0.110] [0.127] [0.005] [0.155] [0.173] [0.017] [0.269] [0.315] year 40.850*** 41.169*** 37.788*** [0.006] [0.006] [0.009] Missionaries 8.706 9.299 8.931 8.002 8.690 8.383 [0.247] [0.179] [0.191] [0.267] [0.196] [0.208] Churches 39.430** 31.380** 30.740** [0.022] [0.038] [0.044] R-squared 0.935 0.939 0.940 0.936 0.940 0.940 0.938 0.941 0.941 Year FE  Observations 818 818 818 818 818 818 818 818 818 # Years 999999999 # Reservations 106 106 106 106 106 106 106 106 106 Panel C, outcome: School-Attendance / Capacity School-Attendance / Enrollment Total Alloted Acreage 0.172 0.291** 0.345*** 0.294*** 0.193* 0.325** 0.377*** 0.312*** / 160 [0.176] [0.033] [0.004] [0.001] [0.084] [0.011] [0.001] [0.001] year -0.656** -0.633** [0.037] [0.021] # schools -3.136*** -3.034*** -3.036*** -2.904*** [0.001] [0.000] [0.002] [0.001] $ government spending -0.000 -0.000 [0.537] [0.407] R-squared 0.472 0.510 0.575 0.576 0.572 0.628 0.697 0.698 Year FE   Observations 1,088 1,088 1,088 1,088 1,088 1,088 1,088 1,088 # Years 27 27 27 27 27 27 27 27 # Agencies 44 44 44 44 44 44 44 44

Notes: The assimilation measures in Panels A–B were only reported in 1912–1920, whereas education in Panel C was reported from 1887, but by agency because many reservations attended schools under a shared agency umbrella. In Panels A–B, standard errors are clustered at the reservation-level, In Panels C at the agency level. p-values are reported in square brackets. *** p<0.01, ** p<0.05, * p<0.1.

9 story, even after controlling for institutions that could have separately coerced or incentivized assimilation on reservations. In the following sections, we will investigate less direct measures of assimilation, but do so in the universe of on-reservation Native American households.

4 Native Americans in the Full Count Census

The 1930 and 1940 Full-Count US Censuses (FCC) each include over 300,000 individual Native

Americans living in roughly 80,000 households (with some intermarriage). The data provides us with measures of individuals’ and households’ incomes, occupational rank, property ownership, and a range of other outcomes. Unfortunately, the micro data do not contain any information on the reservation an individual Native American belonged to. We address this problem through the construction of a stable reservation-to-individual crosswalk. This crosswalk is based on a combination of spatial matching techniques using Census-enumerated spatial variables. The most readily available spatial information in the historical Census is county. Unfortunately, reservations tend to be relatively tightly clustered in parts of the Paciﬁc Northwest, the Southwest, the

Plains, and around the border area of Minnesota and Michigan. These within-county clusters pro- hibit accurate matching of individuals to reservations.10 Instead of county, we therefore relied on the city (variable STDMCD) and enumeration district variables (variable ENUMDIST) . There is considerable variation in the usefulness of these variables across census waves. Speciﬁcally, the spatial information in the 1930 Census is better than in the 1940 census on both variables. There- fore, our approach is to create a complete ENUMDIST-to-reservations crosswalk for 1930 (which incorporates the STDMCD information), and then to apply this to 1940 through a ENUMDIST-1930- to-ENUMDIST-1940 crosswalk, which we web-scraped from Morse and Weintraub(2019).

The city variable STDMCD: In 1940, city is often available only for large metropolitan areas, rendering them unhelpful for determining reservation assignment. The 1930 Census is unique in that it has city or township information that can be used to assign reservations to enumeration districts.11 Because STDMCD in 1930 includes the many small towns one ﬁnds on or near reser-

10 An additional concern is that county boundaries were also in ﬂux over this period. Between 1900 and 1940 there were nearly 750 county boundary adjustments, with many of these occurring in western states with American Indian reservations (Siczewicz, 2011). 11 STDMCD is not available on the IPUMS/NBER version of the Full Count, but it is available on the Center for Popu- lation Research (CCPR) server at UCLA version.

10 vations, we identiﬁed cities associated with almost the entire universe of reservations, which we could then link to the enumeration districts that contained these towns.

We focus our search on the 5,154 townships that included a Native American resident in 1930.

We geolocated this set of townships using both the BING Maps and Google Maps APIs. Both

APIs deliver speciﬁc coordinates for each township with the level of accuracy or precision of each match. We spatially join each geolocated township to modern reservation boundaries to assign each township to a reservation. To identify adjacent townships, we calculate the distance from each township to the closest reservation boundary. A small set of townships are identiﬁed by their location in the Public Land Surveillance System (PLSS). For this subset, we assign reservation location using the spatial overlap of reservations to the PLSS.

Enumeration Districts ENUMDIST: ENUMDIST in 1930 is a numeric value that, when com- bined with state and county, uniquely identifies an enumeration district in 1930. In 1940, enumeration districts are identified using state, supervisor district, and the enumeration district.12 These enumeration districts have not been geo-referenced, and we therefore rely on the text of their official descriptions to make use of them. We web-scraped these descriptions from Morse and

Weintraub(2019). These descriptions were much richer in 1930 than in 1940: the ofﬁcial descriptions contained explicit references to being on a reservation for 513 enumeration districts covering

100 reservations.

Assigning Reservation in 1930: Our final location assignment is based reservation information from five sources. First, we manually searched the enumeration district descriptions that we web-scraped from Morse and Weintraub(2019) and assigned a specific reservation if it was listed in the location description. These descriptions identified 513 enumeration districts that connected to 100 unique reservations. Second, we used the spatial overlap from the PLSS. Third, we applied the spatial overlap and adjacency from the APIs. We restricted ourselves to assignments with a high degree of precision. Fourth, for cases where there was only one reservation present in the county, we assigned every resident of the county to the reservation. After sequentially applying the first four steps, we were left with 309 unassigned townships, which we manually matched to reservations or specified as outside of a reservation boundary.

12 There were again differences in data availability between different server versions. ENUMDIST is available on the IPUMS/NBER version in 1940, but not in 1930. Fortunately, ENUMDIST was available for 1930 on the Center for Population Research (CCPR) server at UCLA version.

11 The process resulted in a reservation assignment for nearly 330,000 of the 402,000 individuals living in households with a Native American. Table2 shows the distribution of reservation assignment by source. The reservation population comes from nearly 330 different reservations. Over

2/3 of the off-reservation population is comprised of non-Natives. Among the Native American population, over 90 percent were still living on or adjacent to reservations in 1930.

Table 2: Reservation Assignment by Source and Decade Source 1930 1940 Enumeration District Descrip 85,563 97,046 PLSS Overlap 20,071 22,212 Township API Overlap 116,822 78,843 Adjacent Township API 72,083 53,295 Single County 16,092 12,141 Hand Link 16,356 13,112 Total Reservation Population 326,987 276,649 Off-Reservation Population 75,225 83,178 Total Population 402,212 359,827

Assigning Reservation in 1940: The 1940 Census does not contain the township information nor the reservation speciﬁc descriptions we used to assign reservations in 1930. However, we constructed a crosswalk between enumeration districts in 1930 and 1940 with content web-scraped from Morse and Weintraub(2019). This ENUMDIST-1930-to-ENUMDIST-1940 crosswalk allows us to map the 1930 reservation assignment at the enumeration district level to enumeration districts in 1940. Table2 summarizes the results of the enumeration district crosswalk assignment.

Appendix C lists the number of households we mapped to each reservation in the 1940 census.

5 Results

Indian allotment was a combination of an unconditional and a conditional transfer program: Al- lottees received their allotments in trust unconditionally, and had the right to use the land for their own purposes, as well as the right to leasing rents. The associated income stream was comparable in magnitude to average per capita incomes at this time. The second arm—obtaining the land in

‘fee simple’— was easily an order of magnitude more valuable, but to obtain it, the allottee had to be declared “competent” by the local BIA agent. We hypothesize that households rationally tried

12 to signal their cultural assimilation in order to improve their chances of being declared “competent” by the local BIA agent. The most obviously assimilation-signalling activities — church-going and wearing “civilized dress” — were enumerated in the aggregate in the BIA Annual Reports from 1911–1920, but no in the population census. In the micro Census data, we will instead observe farming and school attendance as the two primary outcomes available in both 1930 and

1940. In 1940, we also have a range of income measures, as well as educational attainment. In

Section 5.1, we analyze outcomes in the 1930 and 1940 Full Count cross-sections. In Section 5.2, we investigate the Census variable educational attainment to conduct a cohort analysis that uti- lizes the differential phase-in of allotment across reservations to ask whether school-attendance increased with allotment, comparing the same birth cohorts across reservations.

5.1 Cross-Sectional Results

We begin with a pseudo-panel exercise, in which we estimate the difference-in-difference speciﬁ- cation

Y Y 0 0 Yit(re) = αA · Allottedi + αA,1940 · µ1940 · Allottedi + β Xi + γ Xr + µBIA + µ1940 + it(r), (1)

where Yit(re) denotes individual outcomes, i indexes a household, t the year, r indexes reservation and subscript BIA indexes the BIA region. The Full Count Census contains no allotment information so that Allottedi is purely a reservation-level measure in equation (??), although we will be able to improve on this for a subset of the data in Section6. 13 Across-the-board changes in farming

Y and schooling from 1930 to 1940 are captured in µ1940, and αA,1940 estimates whether farming and schooling dropped off (or expanded less) on reservations where the 1934 IRA ended allotment

Y and thereby the incentive to signal assimilation through these activities; αA · Allottedi estimates the effect of being allotted, comparing allotted households on allotted reservations to un-allotted households on un-allotted reservations in the same administrative BIA region.

The BIA ﬁxed effect µBIA washes out regional differences in the administration of allotment as well as in the implementation of other policies may affect our farming and schooling outcomes.

Even conditional on regional BIA ﬁxed effects, however, allotment was probably rolled out in a

13 Allotted and unallotted reservations are identiﬁed in Ofﬁce of Indian Affairs(1935), cross-checked against BIA annual reports.

13 non-random order across reservations, and the characteristics that made a reservation more likely to be allotted are unlikely to have been econometrically exogenous. Fortunately, the detailed analysis of Indian allotment in Leonard, Parker, and Anderson(2020) provides us with clear guidance on the determinants of allotment and the control variables Xr that are needed to control for selection into being allotted.

Table 3: Pseudo-Panel Results for Farming and Schooling

(1) (2) (3) (4) (5) (6) (7) (8) Panel A, outcome: D (Industry = Agriculture) D (Live on Farm & Industry ≠ Agriculture) D(Alloted) 0.096** 0.109*** 0.114*** 0.106*** -0.011 -0.004 -0.015 -0.012 [0.034] [0.006] [0.004] [0.006] [0.655] [0.859] [0.517] [0.618] D(Alloted) × Year=1940 -0.088 -0.078 -0.079 -0.079 0.039 0.040 0.041 0.041 [0.164] [0.206] [0.200] [0.198] [0.364] [0.345] [0.336] [0.331] year=1940 -0.109*** -0.126*** -0.124*** -0.125*** 0.101*** 0.098*** 0.097*** 0.097*** [0.002] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] N = # Households 128,053 127,000 127,000 127,000 128,053 127,000 127,000 127,000 R-squared 0.168 0.220 0.223 0.225 0.066 0.072 0.080 0.081 Panel B, outcome: D (Go to School = Yes) D(Alloted) -0.015 -0.009 -0.010 0.005 -0.015 -0.009 -0.010 0.005 [0.429] [0.671] [0.587] [0.723] [0.426] [0.666] [0.583] [0.731] D(Alloted) × Year=1940 -0.057*** -0.055*** -0.054*** -0.052*** -0.057*** -0.054*** -0.054*** -0.052*** [0.006] [0.006] [0.005] [0.006] [0.006] [0.006] [0.005] [0.006] year=1940 0.066*** 0.066*** 0.065*** 0.064*** [0.000] [0.000] [0.000] [0.000] year=1940 × [5

Notes: Panel A shows the farming results, Panel B shows the schooling results. Everything is presented in blocs of 4 columns, where we introduce increasingly more stringent controls. These controls are listed in Appendix-Table A1

Table3 shows the results of estimating equation (1) for two farming outcomes and for children’s schooling decisions. Our hypothesis of assimilation-signalling naturally gives rise to the prediction that there was less of it after the 1934 IRA removed the incentive to signal. Columns 1–4 of Panel A show that households of allotted reservations were more likely to be farmers. Columns

14 1–4 of Panel A also show that this difference shrunk by two-thirds after the IRA, although this re- duction is imprecisely estimated. Similarly, columns 5–8 of Panel A show the share of households who lived on a farm, but did not themselves work the farm (a proxy for leasing out the allotted land) increased considerably after the 1934, although this increase is again imprecisely estimated.

Panel B shows the results for children’s schooling choices, considering children aged 6-20.

Columns 1–4 show that the likelihood of going to school dropped off on allotted reservations after the 1934 IRA. Columns 5–8 show this is robust to allowing the 1940 time-trend to vary by age- group. (Age-group is also always included among the demographic controls in all columns.) This robustness check is important because of potential state-level changes in compulsory schooling.

5.2 Evidence from Cohort Analysis

Figure 3: Distributions of Allotments Issued around the Main Issue Year

60000

40000

20000 # Patents Issued

0 -20 -10 0 10 20 time Notes: The figure depicts the distribution of allotments in event-time, where time= 0 is defined on each reservation as the year with the biggest number of allotment issuances. The figure shows that allotments were issued in a very concentrated time-window, which allows us to define the start-year of allotments on reservations, as in Figure4.

In this section, we use a cohort analysis to support our hypothesis that allottees sent their kids to school in part to signal their assimilation. One way to approach this is to utilize the differential phase-in of allotment timing across reservations to ask whether school-attendance increased with

15 Figure 4: Cohort-Analysis of Incentive to ‘Signal Assimilation’

Allotments Issued 1934: IRA ends 1907-09 Allotment

Reservation 1

Allotments Issued 1919-21

Reservation 2

Notes: The ﬁgure depicts two reservations. On reservation 1, allotment happened from 1907 to 1909, on reservation 2, allotment happened from 1919 to 1921.

allotment, comparing the same birth cohorts across reservations. Consider Figure4: it compares two reservations, one of which was allotted in 1907–09, the other in 1919–21. This implies the assimilation-signalling incentive in reservation 1 began with cohorts that were ten years younger than in reservation 2.14 To test this, we run the following cohort analysis:

Y ITT 0 Yi(re) = αS · Incentive-to-Signali + β Xi + µr + i(r), (2)

where Incentive-to-Signali is an indicator variable that takes value 1 for individuals who were of school age (or of a more narrowly deﬁned age window) after allotment started on a reservation.

Incentive-to-Signali takes value 0 for all individuals who were of school age before this. Speciﬁca- tion (2) includes either a reservation ﬁxed effect µr — comparing individuals of all ages (above age 5 in 1940) and both genders within a reservation. — or a much more conservative household

ﬁxed effect, comparing the educational attainment of individuals across cohorts within the same households.

Table4 presents results of estimating equation (2). Across Panels A–D, the table considers four

ITT different measures of Incentive-to-Signali . Panels A and B consider continuous measures that vary from age 0 to 7, or from age 0 to 13 (for the number of years in the prime school-age years from 6–12 or from 6–18 when a child would have had an incentive to signal assimilation under

14 On each reservation, we know the date on which allotment started in earnest from the BIA reports as well as from the BLM data described in Appendix-Figure A3.

16 Table 4: Cohort Analysis of Education Attainment in 1940 Census

(1) (2) (3) (4) (5) (6) (7) (8) D (Dropped Out of D (Attended Middle Years of Education D (No Schooling) Primary School) School) Panel A # of Years Age 6--12 under 0.034* 0.041** -0.008*** -0.009*** 0.007*** 0.008*** -0.001 -0.000 Allotment [0.079] [0.045] [0.002] [0.001] [0.006] [0.005] [0.623] [0.865] R-squared 0.538 0.735 0.179 0.400 0.452 0.638 0.519 0.679 Panel B # of Years Age 6--18 under 0.025 0.030 -0.005*** -0.006*** 0.005** 0.005** -0.001 -0.000 Allotment [0.133] [0.101] [0.010] [0.004] [0.031] [0.033] [0.679] [0.812] R-squared 0.538 0.735 0.178 0.399 0.452 0.637 0.519 0.679 Panel C D (Years Age 6--12 under 0.285 0.330* -0.092*** -0.100*** 0.081*** 0.089*** -0.011 -0.011 Allotment > 0) [0.109] [0.083] [0.001] [0.000] [0.002] [0.003] [0.370] [0.401] R-squared 0.537 0.735 0.181 0.401 0.453 0.638 0.519 0.679 Panel D D (Years Age 6--18 under 0.306 0.322 -0.099*** -0.107*** 0.084** 0.089** -0.015 -0.019 Allotment > 0) [0.186] [0.212] [0.000] [0.000] [0.013] [0.020] [0.194] [0.170] R-squared 0.538 0.735 0.181 0.402 0.453 0.638 0.519 0.679 Observations 245,264 236,668 238,549 230,002 238,549 230,002 238,549 230,002 # Fixed Effects 335 49,600 334 49,173 334 49,173 334 49,173 Controls Demog.  Fixed Effects Rez Household Rez Household Rez Household Rez Household

Notes: This table presents results of estimating equation (2). Each panel considers 4 different outcomes, and every other column switches from reservation to household ﬁxed effects. Across Panels A–D, the table considers four different mea- ITT sures of Incentive-to-Signali . The number of observations varies between even-and odd-numbered columns because of unitary households being absorbed by ﬁxed effects.

17 allotment). Panels C and D consider indicator measures that are equal to 1 for persons who had any incentive to signal assimilation under allotment during their school-age years. Each panel

ITT considers 4 different outcomes. Across panels, columns 1–2 suggest an effect Incentive-to-Signali on years of schooling but this effect is not very precisely estimated and averages over the range of possible years of schooling, including high-school and college. Columns 3–4 and 5–6 focus on

ITT more essential cutoffs, showing that Incentive-to-Signali made it considerably less likely that a child dropped out of primary school, and considerably more likely that a child not only ﬁnished their primary education but went on to middle school. From the point of view of statistical identi-

fication, every other column switches from reservation to household fixed effects, and thus offers a much more narrow comparison group. Strikingly, the within-household cross-cohort comparisons in even-numbered columns barely diverge or lose in statistical significance when compared to the the within-reservation cross-cohort comparisons in odd-numbered columns. Column 7–8 show that these effects are not explained by the extensive margin; this is because almost the entirety of the population in the data (except very old people) had some schooling.

6 Household-Level Allotment Variation

In Section5, we only used aggregate BLM data to determine on which reservations allotment happened, and when. In this section, we assign allotment data to individuals in the Census to do a more ﬁne-grained analysis. To do this, we record-link the Native American population in the 1940 Census to a newly digitized separate BIA census of American Indians called the Indian

Census Rolls (ICR). The ICR contain households’ allotment numbers, which we can then link to the BLM database on the universe of Indian allotments. In Section 6.1, we describe how we can link household-level allotment information to the micro Census data. In Section 6.2, we undertake statistical analysis of these linked data.

6.1 Measuring Treatment: The Indian Census Rolls

The ICR were special censuses that were unrelated to the population census and were collected on reservations into the late 1930s. The ICR did not record any economic outcome data; instead they are simply complete enumerations of all Native American households. Appendix-Figure A6

18 depicts a sample page from the ICR. They include only individuals’ names, their ages, and their relations within the household (e.g. spouse or son). Critically, they also report whether individuals received land allotments, including a complete listing of unique allotment numbers by individuals. We digitized a complete cross-section of the ICR for all reservations in the mid-1930s.

Linking the ICR to the population census is difﬁcult because the main linkage variables (names and birth years) are recorded with higher-than-average noise in all Native American records at this time.15 To address this problem, we propose a new record linkage algorithm that emphasizes the similarity of households, adjusting individual record similarity scores up or down to account for the similarity of individuals’ households across data sets.

In the following, we brieﬂy describe our approach to record linkage. We discuss the process in more detail in Appendix B.

1) We ﬁrst use standard individual record linkage methods to establish for each individual in

a master data-set (say, the 1940 FCC), a full set of potential individual links in a using data-

set (say, the 1930 FCC). Given the noise in Native American names, particularly traditional

names, we first apply bigram-indexation to first and last name to define the set of individuals

in the using data with similar names.16 We are ‘generous’ in this set, in the sense that we set

the string-similarity cutoff low enough that each person in the master data has an average of

8 links in the using data, to accommodate the high name variation in our sample. We then

trim that set to include only potential matches that deviate by at most 4 years in birth-year

for master records born after 1910, and by at most 6 years for master records born before

1910.

2) Next, we adjust the link similarity of these individual links by considering the similarity

of adjacent household members. At this step, each individual in the master data and each

individual in the using data is associated with a unique household id. Consider a pair of

households 1940-HH A in master and 1930-HH B in using. Our approach is to calculate

15 For one, census enumerators made frequent errors in spelling Indian names. For another, Native Americans were frequently anglicizing their names or re-arranging ﬁrst and last names in this period. (For example, one individual called ‘Little Running Bear’ in 1930 had become ‘Jim Bear’ in 1940.) Stated birth years also varied considerably from data source to data source, especially for older people. 16 We use the stringsim() function in R’s RecordLinkage, and block on gender and on ‘meta-state’. (A ‘meta- state’ is a set of two states whose boundaries are straddled by reservations. This occurs, e.g. in New Mexico and Arizona. We formed 40 meta-states. In principle, you could apply any of the commonly used approaches discussed in Bailey, Cole, Henderson, and Massey(2017) and Abramitzky, Boustan, Eriksson, Feigenbaum, and P erez´ (2019).

19 the number of individuals in 1940-HH A who are linked to 1930-HH B in step 1. This cal-

culates a metric of household similarity. We can then upward-adjust the similarity-scores

of individual links between individuals in 1940-HH A and 1930-HH B by a percentage for

each marginal increase in our metric of household similarity. We also re-generate adjusted

similarity-scores between all non-linked individuals in 1940-HH A and 1930-HH B to poten-

tially ﬁnd new links. The idea is that these new links did not meet the match-threshold in

the individual-level linkage in step 1, but do meet the threshold with the upward-adjusting.

Household similarity therefore brings new individual links into the fold.

3) From this list of potential matches, we apply a “best-off” gridsearch on the adjusted similar-

ity scores created in step 2 to identify the best unique match. The grid-search maximizes the

sum of similarity-scores of all links, with the constraint that each individual record in the

master data is linked to at most one record in the using data, and each record in the using

data is linked to at most one record in the master data.

4) Next, we apply a series of consistency checks to identify potentially problematic links for

manual veriﬁcation. Household structure introduces potential criteria to further evaluate

the match quality and generate ﬂags to identify potentially problematic links. For example,

if a husband and wife pair in 1940 were linked to a husband and wife belonging to differ-

ent households in 1930. From this, we establish an anchor household and evaluate every

individual connected to the anchor household across every linked wave of the census.

5) We then manually evaluate every ﬂagged link and, hard-code ‘true-positive’ and ‘true-

negative’ manual links to every individual associated with the anchor household across the

census waves.

6) To complete the process, we apply these the hard-coded links back onto our original data-

sets and re-run steps 1–3, as well as repeating step 4 which is expected to show the share of

ﬂagged households having been reduced by the manual work.

[Discussion of linked dataset here]

20 6.2 The Second Treatment Arm

In Section5, we focused only on the effect of the ﬁrst treatment arm of being allotted. Now that we can distinguish whether a household’s allotment was transferred into fee simple or not, we can consider the following two estimating equations

Y Y 0 0 Yi(re) = βF · Fee-Simplei + βT · In-Trusti +γ Xr + β Xi + µBIA + i(r) (3)

Y 0 Yi(re) = αF · Fee-Simplei +β Xi + µr + i(r), (4)

where equation (3) includes BIA-region ﬁxed effects µe, and equation (4) includes reservation ﬁxed effects µr. Assuming that the reservation controls Xr mostly control for unobserved determinants

Y Y of allotment, the difference in the estimated coefficients βcF − βcT in equation (3) should be approx- Y imately equal the estimated coefficient αcF in equation (4). By estimating specifications (3) and (4), we can decompose the allotment effect between households that obtained their land in fee simple and households whose allotments stayed in trust. We recognize that obtaining the land in fee simple is a function of a household being deemed “competent”, and as such the results of estimating equations (3) and (4) should only be interpreted as suggestive, not causal.

[Discussion of linked dataset here]

6.3 Evidence from Cohort Analysis

To support our hypothesis that individuals used educational attainment to signal readiness for a fee-simple patent is through a direct comparison of the incentive to signal. Figure5 illustrates a case where two households receive trust patents at the similar time, marking the period where they now have an incentive to pursued the local agent they are ready to have their patent converted to fee-simple. For Household 1, their incentive to signal stops once the fee patent is issued.

Household 2 has a longer period with the incentive to signal, which stops in 1934 with the In- dian Reorganization Act. If the incentive to signal is responsible for the changes in educational attainment we observed, we would expect households to accumulate more education during the periods they have the incentive to signal.

21 Figure 5: Cohort-Analysis of Incentive to ‘Signal Assimilation’

1907: Allotment 1921: Transfer to Issued Fee Simple Household 1: Received Fee Simple

1934: IRA ends Allotment Household 2: Remained In Trust

Notes: The ﬁgure depicts two households that receive allotments in the same year. At this point, the incentive to signal assimilation begins. Household 1 is declared competent and receives their allotment in fee simple in 1921. At this point, the incentives to signal assimilation ends. Household 2’s allotment remains in trust, and the incentive to signal assimilation ends only in 1934, when the IRA ends transfers into Fee Simple.

To test this, we run a cohort analysis

Y 0 Yi(re) = αS · Incentive-to-Signali +αF · Fee-Simplei + β Xi + µr + i(r) (5)

Y 0 Yi(re) = αS · Incentive-to-Signali +β Xi + µHH + i(r), (6)

that is similar to specification (2) except that Incentive-to-Signali has an end-date (i.e. min{date(fee), 1934}), and that we are able to include household fixed effects (in specification (6)): Incentive-to-Signali is an indicator variable that takes value 1 for individuals who were of school age (or of a more narrowly defined age window) at the time their household had been allotted but the allotment not yet transferred to fee simple. Incentive-to-Signali takes value 0 for all individuals in unallotted households, and all individuals who were of school age before their household was allotted, or after their household’s allotment went into fee simple, or after 1934 when the IRA ended allotment.

Speciﬁcation (5) includes a reservation ﬁxed effect µr, and thus compares individuals of all ages

(above age 5 in 1940) and both genders within a reservation. Incentive-to-Signali varies with individual i’s age at the time of allotment. Speciﬁcation (6) replace µr with a household ﬁxed effect

µHH , thus comparing only different cohorts within a household.

[Discussion of results to be added here.]

22 7 Ancestral Traditions of Private Property

The data on ancestral cultural norms comes from the Murdock (1967) Ethnographic Atlas (EA).

Among many other measures, the EA has tribal information on property rights norms. In fact, the

EA has separate measures for two different types of property, movable and real property, i.e. land.

We will be interacting the effects of land allotment with the types of property rights separately.

To the best of our knowledge, there is currently no paper in the ‘cultural economics’ literature that has focused on the effects of property rights traditions. To rule out spurious correlations of property rights norms with other tribal characteristics, we will also look at two other dimen- sions of traditional norms that have received attention in the cultural economics literature, namely norms of having nuclear family and norms of having a bride price, i.e. of the groom’s family pay- ing a material transfer to the bride’s parents.

The results in Section 6.2 provide suggestive (non-causal) evidence that obtaining fee simple rights to one’s allotment may have had positive effects on farming, home ownership, labor market outcomes and education. We recognize that these partial correlations may be explained by selection on unobservable characteristics that made households attain more education, more likely to earn higher wages as well as making them more likely to be deemed competent. In contrast, they may also be biased by selection on outcomes: households who were more likely to beneﬁt from fee simple rights to their lands were also more likely to obtain these rights, in part because they signalled their cultural assimilation to a larger extent. Another way of saying this is that obtaining fee simple rights had heterogenous effects on the treated households, and that the treatment effect on the treated (ToT), is potentially far off the population average treatment effect (ATE) because treated households were disproportionately those for whom full legal title had the biggest effect.

We address this possibility by looking for evidence of heterogenous treatment effects along the measurable and well-studied dimension of ancestral cultural norms. These data come from the Murdock (1967) Ethnographic Atlas (EA). Among many other measures, the EA has tribal information on property rights norms, which we can interact with our baseline allotment treatment.

If the effect of fee simple property rights was heterogenous, then it seems likely that the effect of obtaining fee simple rights (on the second treatment arm, i.e. relative to households with in-trust allotments on the same allotted reservation) would be most pronounced among populations with

23 established ancestral traditions of private property. If the effect of obtaining full fee simple title is indeed more pronounced in tribes with traditions of private property, then allotted households in those tribes should also have had stronger incentives to signal their competency in order to obtain legal title. This corollary hypothesis cannot be tested in the same within-reservation comparison, however, because all households in this comparison may have had identical incentives to for signalling assimilation. Instead, one can test the corollary hypothesis of a stronger incentive to signal by comparing allotted households who did not obtain their land in fee simple to un-allotted households in the same tribe, taking advantage of the many-to-one mapping from reservations to ethnographic tribes.

Figure 6: Slicings of the Data

"Wealth Transfer Effect" "Assimilation Signalling"

Rez 1 Rez 2 Rez 1 Rez 2

Fee Fee Unalloted Unalloted

Trust Trust

Notes: With reservation fixed effects, the within-reservation comparison of obtaining allotments in fee simple isolates the wealth transfer effect (left panel of figure). With tribe fixed effects, omitting households that obtained their allotments in fee simple isolates a lower bounds on assimilation signaling effect (right panel of figure)

The ﬁrst hypothesis amounts to testing whether the difference between the two shaded groups in the left panel of Figure6 is more pronounced within reservations belonging to tribes with traditions of private property. The second hypothesis amounts to testing whether the difference between the two shaded groups in the right panel of Figure6 is more pronounced within reservations belonging to tribes with traditions of private property. This comparison is useful because it shuts off any potential effects of the fee simple transfer itself.17

17 This second comparison likely estimates a lower bound on the effect of traditions of private property no the incentive to assimilate, because the treated group is the “negatively selected” group of households who were not declared “competent” by the BIA agent.

24 The following estimating equation tests the ﬁrst hypothesis

EA 0 Yi(re) = αF · Fee-Simplei + αF · Fee-Simplei × EAe +β Xi + µr + i(r), (7)

EA where EAe is an indicator for whether tribe e had ancestral traditions of private property., and αF estimates the interaction between obtaining full title to the land and having ancestral traditions of private property in the EA. The next estimating equation tests the second hypothesis

EA 0 Yi(re) = αT · In-Trusti + αT · In-Trusti × EAe +β Xi + µe + i(r), (8)

where EAe does not enter as a regressor, because tribe-characteristics are absorbed by the tribe

ﬁxed effect µe.

8 Conclusion

TBA

25 References

Abramitzky, R., L. Boustan, and K. Eriksson (2016). To the new world and back again: Return migrants in the age of mass migration. ILR Review, 0019793917726981.

Abramitzky, R., L. P. Boustan, and K. Eriksson (2012). Europe’s tired, poor, huddled masses: Self- selection and economic outcomes in the age of mass migration. American Economic Review 102(5), 1832–56.

Abramitzky, R., L. P. Boustan, K. Eriksson, J. J. Feigenbaum, and S. Perez´ (2019). Automated linking of historical data. Technical report, National Bureau of Economic Research.

Alesina, A., P. Giuliano, and B. Reich (2013). Nation-building and education. Technical report, National Bureau of Economic Research.

Algan, Y. and P. Cahuc (2010). Inherited trust and growth. American Economic Review 100(5), 2060– 92.

Bailey, M., C. Cole, M. Henderson, and C. Massey (2017). How well do automated linking methods perform in historical samples? evidence from new ground truth. Unpublished manuscript.

Bandiera, O., M. Mohnen, I. Rasul, and M. Viarengo (2018). Nation-building through compulsory schooling during the age of mass migration. The Economic Journal 129(617), 62–109.

Banner, S. (2009). How the Indians lost their land: Law and power on the frontier. Harvard University Press.

Barrera-Osorio, F., M. Bertrand, L. L. Linden, and F. Perez-Calle (2011). Improving the design of conditional transfer programs: Evidence from a randomized education experiment in colombia. American Economic Journal: Applied Economics 3(2), 167–95.

Barrera-Osorio, F., L. L. Linden, and J. E. Saavedra (2019). Medium-and long-term educational consequences of alternative conditional cash transfer designs: Experimental evidence from colombia. American Economic Journal: Applied Economics 11(3), 54–91.

Behrman, J. R., S. W. Parker, and P. E. Todd (2011). Do conditional cash transfers for schooling generate lasting beneﬁts? a ﬁve-year followup of progresa/oportunidades. Journal of Human Resources 46(1), 93–122.

Carlson, L. A. (1981). Indians, bureaucrats, and land: the Dawes Act and the decline of Indian farming. Number 36. Praeger Pub Text.

Costa-Font, J., P. Giuliano, and B. Ozcan (2018). The cultural origin of saving behavior. PloS one 13(9), e0202290.

Demsetz, H. (1967). Toward a theory of property rights. American Economic Review 62, 347–359.

Di Tella, R., S. Galiant, and E. Schargrodsky (2007). The formation of beliefs: evidence from the allocation of land titles to squatters. The Quarterly Journal of Economics 122(1), 209–241.

Dippel, C., D. Frye, and B. Leonard (2020). Property Rights without Transfer Rights: A Study of Indian Land Allotment. NBER working paper #27479.

26 Dohmen, T., B. Enke, A. Falk, D. Huffman, and U. Sunde (2018). Patience and comparative development. Technical report, Working Paper.

Enke, B. (2019). Kinship, cooperation, and the evolution of moral systems. The Quarterly Journal of Economics 134(2), 953–1019.

Federal Deposit Insurance Corporation (1992). Federal deposit insurance corporation data on banks in the united states, 1920-1936.

Feigenbaum, J. J. (2016). A machine learning approach to census record linking. Technical report, Working Paper.

Feir, D. L. (2016a). The intergenerational effects of residential schools on childrens educational experiences in ontario and canadas western provinces. The International Indigenous Policy Jour- nal 7(3), 5.

Feir, D. L. (2016b). The long-term effects of forcible assimilation policy: The case of indian boarding schools. Canadian Journal of Economics 49(2), 433–480.

Fernandez,´ R. (2011). Does culture matter? In Handbook of social economics, Volume 1, pp. 481–510. Elsevier.

Ferrie, J. P. (1996). A new sample of males linked from the public use microdata sample of the 1850 us federal census of population to the 1860 us federal census manuscript schedules. Historical Methods: A Journal of Quantitative and Interdisciplinary History 29(4), 141–156.

Fouka, V. (2019). Backlash: the unintended effects of language prohibition in us schools after world war i. Technical report.

Golenko, A. (2010). Did allotment contribute to the diminishment of american indian incomes? a study of allotment policies and american indian incomes during 1912-1920.

Gregg, M. T. (2018). The long-term effects of american indian boarding schools. Journal of Develop- ment Economics 130, 17–32.

Haines, M. R. (2010). Historical, Demographic, Economic, and Social Data: The United States, 1790- 2002. Inter-University Consortium for Political and Social Research.

Leonard, B. and D. Parker (2020). Fragmented ownership and natural rerssource use: Evidence from the bakken. Economic Journal Forthcoming.

Leonard, B., D. Parker, and T. Anderson (2020). Poverty from Incomplete Property Rights: Evi- dence from American Indian Reservations. Journal of Development Economics Forthcoming.

McDonnell, J. (1980). Competency commissions and indian land policy, 1913-1920. South Dakota History 11(1), 21–34.

Meriam, L. (1928). The problem of Indian administration: report of a survey made at the request of honorable Hubert Work, secretary of the interior, and submitted to him, February 21, 1928. Number 17. Johns Hopkins Press.

Morse, S. P. and J. D. Weintraub (2019). The Uniﬁed Census Enumeration District Finder 1880– 1940. https://stevemorse.org/census/unified.html. Accessed: 2019-11-08.

27 Murdock, G. P. (1967). Ethnographic atlas.

Nunn, N. (2012). Culture and the historical process. Economic History of Developing Re- gions 27(sup1), S108–S126.

Ofﬁce of Indian Affairs (1920). Annual Report of the Commissioner of Indian Affairs for the Fiscal Year Ended 1920. Department of Interior.

Ofﬁce of Indian Affairs (1935). Indian land tenure, economic status, and population trends. In Part X of the Report on Land Planning. Washington: Department of Interior.

Otis, D. S. (2014). The Dawes Act and the allotment of Indian lands, Volume 123. University of Oklahoma Press.

Parker, S. W. and P. E. Todd (2017). Conditional cash transfers: The case of progresa/oportunidades. Journal of Economic Literature 55(3), 866–915.

Price, J., K. Buckles, I. Riley, and J. Van Leeuwen (2019, September). Combining family history and machine learning to link historical records. Technical report, NBER Working Paper No. 26227.

Sakalli, S. O. (2017). Secularization and religious backlash: evidence from turkey. Technical report, mimeo.

Schultz, T. P. (2004). School subsidies for the poor: evaluating the mexican progresa poverty program. Journal of development Economics 74(1), 199–250.

Siczewicz, P. (2011). U.S. Historical Counties. Available online from https://publications. newberry.org/ahcbp/.

28 Figure A1: Cultural Assimilation “Before/After Picture”

Appendix A Online Data Appendix

Figure A1 shows one of the most common “before/after pictures” one ﬁnds on the internet in association with the Dawes Act. (We could not pin down the exact origin of the picture.)

29 Figure A2: 1910 Advertisement for Reservation Lands Left from Allotment

Notes:

Figure A2 shows an advertisement for the sale of surplus land, discussed in Section3.

30 Figure A3: Flow of Allotments and Transfers into Fee Simple

Annual Acres Allotted and Fee-Simple 2500 1500 2000 1000 1500 1000 500 500 Allotment Allotment Acres (1,000s) Fee-Simple Fee-Simple Acres (1,000s) 0 0 1887 1906 1924 1934 1940 Year of Transfer 1887: Dawes Act; 1906: Burke Act; 1924: Citizenship; 1934: IRA Dashed:Allotments; Solid: Fee-Simple

Notes: This figure tracks the flow of total acres that were allotted and the flow of acres subsequently transferred into fee simple in the BLM data.

Following approval from the President, each patent issued on the reservation was ﬁled with the

Government Land Office (GLO). These patents—subsequently digitized by the Bureau of Land Man- agement (BLM)—record the transfer of land titles from the federal government to individuals. Each patent contains information regarding the patentee’s name, the specific location of the parcel(s), the official signature date, total acreage, and the type of patent issued. These patents include cash sales, all homestead entries, and the universe of Indian allotments. An important feature of the

GLO data is that we can see the exact date on which each allotment was issued and the date on which it was converted into fee-simple, if ever. This ability to follow the individual allotments and when they were converted to fee-simple allows us to identify them as either allotted-trust or fee-simple lands today. Appendix-Figure A3 depicts the aggregate process of the issuance of allotted-trust land and its conversion into fee simple.

31 Figure A4: Churches and Missions

Missionaries & Churches (dashed) 600

550

500

450

400

1910 1912 1914 1916 1918 1920 year

Notes: This ﬁgure shows the aggregated number of missionaries and churches reported in the BIA’s annual reports for 1911–1920, for the same set of reservations as in Figure2.

32 Aggregate Controls We compile additional county-level controls from two sources. First, we use data from the U.S. Decennial Census in 1940 constructed by Haines(2010). This series includes measures across four categories: population, agriculture, wealth and expenditures, and manufacturing. Our selected population measures include total population, urban population, native born white population, foreign born population, and population density per square mile.

Within agriculture we include the total number of farms, the number of white farmers, and the total farm value. Our third category covers measures of durable good ownership, represented by the percent of the county that own radios and refrigerators, expenditures, represented by the total value of retail sales, and the wealth in housing, represented by the average value of owner occu- pied dwellings. The ﬁnal category includes ﬁve measures of the manufacturing sector. We include controls for the number of establishments, the average number of wage earners, total wages paid, the cost of materials used in the production process, and the value of total output.

Our second county-level data series incorporates banking data from FDIC reports from 1936

(Federal Deposit Insurance Corporation, 1992). We include two measures of banking activity, the total number of deposits and the total number of banks in the county.

33 Table A1: Controls used in the micro-Census Regressions

Demographic Controls County Controls continud.. Sex=2 Number of Manuf. Establishments Agebin 1 by Sex Avg. Number of Manuf. Wage Earners Agebin 2 by Sex Total Manuf. Wages Agebin 3 by Sex Cost of Materials in Manufacturing race=black Value of Manufacturing Output race=white (Native=omitted) Total Deposits (1936) Female HH Head Total Banks (1936) Number in Nuclear Family Historic Controls Number Additional Family Members Total Wars Associated with reservation County Controls Number Settler Trail Kilometers Total Population # Competency Commission Allotments Total Urban Population # Surplus Settler Allotments Total Native White Pop. Leonard, Parker, Anderson Controls Total Foreign Born Pop. Ruggedness Population Density (per sq mi) Stream-Density Total Number of Farms % Timber Number of White Farmers Value of Minerals Total Farm Value Reservation Longitude Total Retail Sales Reservation Latitude Pct. HHs with Refrigerators Distance Fort 1880 Pcts. HHs with Radios Pct_Prime Avg. Value of Owner Occ. Dwelling Railway within 10 miles

Notes: These are the controls used in Table3.

34 Figure A5: Figure 1 in Gregg(2018)

M.T. Gregg Journal of Development Economics 130 (2018) 17–32 school attendance for women but not for men. Second, there may be a negative relationship between educational outcomes and exposure to boarding schools. In a study of children on First Nation reserves, Bombay et al. (2014) found that children who had a parent who attended a residential school were eight percentage points more likely to report learning difficulties in school, and 12 percentage points more likely to repeat a grade. Huffman (2013) surveyed 21 elementary and secondary educators who work at reservation schools and found evidence of a “reverse racism mentality,” which led some American Indian students to dismiss western education as for white students. One teacher, in particular, linked this attitude to factors that “stem from the early education of Native people and the boarding school system” (Huffman, 2013, 35). Using interview data from 31 American Indian grandparents who raised their grandchildren, Mooradian et al. (2007) found evidence, albeit limited, that past boarding school attendance increased elder's reluctance to trust government and educational systems. Third, sending children to boarding schools removed Indian parents as traditional role models, and research has linked contemporary family Fig. 1. Trends in School Enrollment of American Indian Children, 1880–1930. Notes: The violence and perverse child-rearing practices to early twentieth-century black circles represent the total enrollment of school-age Indian children in all schools (i.e., off-reservation boarding schools, boarding schools, day schools, mission-run schools boarding schools. Kawamoto (2001) examined family health problems and public schools). The blue diamonds represent the enrollment in off-reservation among the Confederated Tribes of Coos, Lower Umpqua, and Siuslaw boarding schools. For data availability reasons, the shaded data points reflect the years Indians and traced the use of shaming and negative child disciplining that are the focus of this paper. The calculations come from data in the Office of Indian techniques to their experiences at boarding schools. Irwin and Roll Affairs' Annual Reports, 1880–1930. (For interpretation of the references to colour in this fi (1995) documented cases of child abuse at boarding schools and found gure legend, the reader is referred to the web version of this article.) that a link between past child abuse and abusive behavior as an adult.Notes: The authors speculate that the size and intimacy of Indian reservations might perpetuate this cycle of abuse. were enrolled in a single boarding school. As more boarding schools were On the other hand, however, there are at least two mechanismsFigure established,A5 shows Figure enrollment 1 in Gregg grew(2018 from). 1000 children (nine percent of all through which boarding school students and future generations may Indians participating in schooling at the time) in 1884 to 10,000 children have benefited. First, off-reservation boarding schools provided an (or 18 percent) in 1913. During the second and third decades of the alternative to neighboring, mostly non-Indian public schools. There are twentieth century, as discussed, the federal government slowly changed few positive reports of the treatment of Indian students at public high course and advocated on reservation day schools and neighboring off- schools, and racism appears to have been omnipresent. Child (2000) reservation public schools. Student enrollment increased rapidly at this summarizes the calculus of attending a school that offered upper grades: time, with total enrollment nearing 83,000 in 1932. Despite this policy “When students felt unwelcome in nearby public institutions due to reform, total boarding-school enrollment still hovered around 10,000 racism against Indians, government boarding schools offered a less students per year from 1910 to 1930. threatening environment.” Thus, for some students, a boarding school From 1911 to 1932, Indian agents reported annually the number of was the lesser of two evils. For example, Hirshberg et al. (2005) inter- Indian children enrolled at the government-run off-reservation boarding viewed 61 Alaska Natives and found that roughly 60 percent spoke schools, agency boarding and day schools, mission-run boarding and day positively about their boarding school experience. For these reasons, schools, and neighboring public schools. These data are measured at the boarding school students may have achieved a higher level of education level of the Indian agency which, on35 average, held jurisdiction over 1.5 than non-boarding school students. reservations. Since contemporary Indian data are measured at the Second, boarding schools allowed students to escape the volatile reservation level, I link all historical data to individual reservations by weighting the agency school data by the reservation's share of the total environment on reservations. For example, when Congress established 12 the Chilocco School, the goal was to educate children from “Indian tribes agency population. Therefore, my boarding school variable is located in the Indian Territory who are least provided for under existing computed as the proportion of students from 1911 to 1932 who attended treaties or laws” (Kappler, 1975, 198). Alumni from this school often off-reservation boarding schools. claimed that homeless children benefited the most from the boarding Two characteristics of this variable are worth remarking. First, the school experience (Lomawaima, 1994, 38). Thus, as historian Brenda J. average reservation in the sample contains 19 (out of a possible of 22) Child explains, “In times of family crisis or economic hardship, Indians years of schooling data. Thus, this variable is quite representative of this could turn to boarding schools for help” (Child, 2016, 24). entire era in Indian education. Second, from 1911 to 1932, this variable is highly stationary. To illustrate, using a balanced panel of 55 reserva- 3. Data tions from 1911 to 1932, a Levin et al. (2002) unit root yields a bias-adjusted t-statistic of 3.308 (p-value<0.001), which easily rejects 3.1. Historical data

To measure the legacy of boarding schools on Indian reservations, I 12 There are three notable exceptions to this matching strategy. First, the Navajo Nation use historical data on school enrollment from the Annual Report of the belonged to four Indian agencies over these years: The Eastern Navajo, Northern Navajo, 11 San Juan and Pueblo Bonito Indian agencies. For this reservation, the agency data needed Commissioner of Indian Affairs. Fig. 1 illustrates trends in off- to be aggregated together rather than disaggregated. Second, some states (New York, reservation boarding school enrollment and total enrollment in the Michigan, Florida and Maine) contained only one Indian agency over the entire data period from 1880 to 1930. At the beginning of this period, 153 children window. In these cases, the variation in the proportion of students enrolled in boarding schools would be identical to the variation in reservation size. For this reason, I did not include any reservations from these states in the sample. Third, the Albuquerque Indian Agency and Santa Fe Indian Agency did not report population ﬁgures by reservation; 11 These reservation-level data have been adopted by others, most notably Carlson therefore, population-weighted boarding school proportions could not be calculated for (1981) who estimated the impact of allotment on American Indian farming. the reservations under these agencies.

20 Figure A6: Sample Page of the Indian Census Rolls 36

Notes: Appendix B Record Linkage

The BLM data and the ICR data are straightforwardly linked by allotment number. Once treatment and instrument are attached, however, the ICR needs to be linked to the Full Count Census (FCC) data for the second-stage analysis, and this record-linkage is difﬁcult. The main individual linkage variables at our disposal in both ICR and FCC are ﬁrst and last name, birth-year, location, and gender. Fortunately for us, spatial mobility among Native Americans in this period of time was very low, with the vast majority of Native Americans living on reservations or former reservation lands in both 1930 and 1940.18 We can therefore block the record linkage on location. In our case, because some reservations in the Southwest straddle state borders, we create 40 meta-states so that each reservation is uniquely contained in one, and then block on these. Secondly, we also block on gender.19

Within this blocking, individuals are linked by name and birth-year. The challenge in our data is that both of these linkage variables are recorded with a lot of noise. For names, this is partly driven by the fact that Census enumerators made more frequent errors in recording Native

American names because they were unfamiliar with them. It is further driven by the fact that there was a strong trend of anglicization of names in this period so that names changed over time. For birth-years, noise is introduced into the records by the fact that most Native Americans at that time did not have birth certiﬁcates, and were thus more likely than other populations to revise their stated birth-year as time passed. These challenges make it impossible to obtain a large number of unambiguous individual record links in these data based on only individuals’ information. To make progress in this direction, and ultimately enable us to perform the second stage analysis, we had to move beyond existing individual-based record linkage methods and develop a linkage scheme that incorporates the similarity of households’ structure across data sets.

We begin with a quick discussion of the commonly used linkage methods. Older record linkage methods used a smaller number of variables for matching, and often used name only, often focusing on samples of people with unusual names in order to reduce false positive matches , e.g.

Ferrie(1996). Matching on names is almost always fuzzy matching, i.e. a matching algorithm that

18 Substantial off-reservation mobility really took hold in the 1950s during the ‘Termination Era’, in which the federal government attempted to dissolve reservations. 19 The FCC does not report reservation or tribe, but county of residence, which is sufﬁcient to uniquely place most individuals in a reservation.

37 allows for typos and mis-spellings. A common approach involves splitting ﬁrst and last names into substrings (‘bigrams’), and to construct a similarity index over all bigrams. A commonly used similarity index is to calculate the ’Jaro Winkler index’ between two names. Newer iterations of fuzzy matching have increased the ﬂexibility to include matching on a set of numeric as well as string variables, including distance calipers on numeric variables, e.g. giving a higher match prob- ability when two records’ birth-years are one year rather than three years apart. See Abramitzky,

Boustan, and Eriksson(2012, 2016) for more recent applications. More recently, the emergence of machine learning algorithm has given a real boost to the precision of record linkage methods, as it allows for training an algorithm. See for example the method outlined in Feigenbaum(2016).

There is an active and ongoing debate on the choice of methods. Bailey et al.(2017) review several of these methods and show that all algorithm may produce samples that are not fully representative of the underlying population. This includes linking records by hand, although this method is favored by Bailey et al.(2017). By contrast, Abramitzky et al.(2019) show that a range of automated linkage methods on a range of standard linkage sample perform as well as manual linkage can be expected to.20 Methods of historical record linkage are fast evolving, driven by advances in the access to historical individual level data, computational power needed for linkage algorithms, the ability to scale up manual linkage through online job platforms, and machine learning capa- bilities. However, existing linkage methods are still almost entirely focused on individual linkage.

The one exception that uses household structure for record linkage that we are aware of is Price,

Buckles, Riley, and Van Leeuwen(2019).

Table A2: Household-Based Linkage Example ICR [master dataset] FCC [using dataset] birthy birthy hhid r namelast namefrst relate links score hhid r namelast namefrst relate

35545638 1908 SESSPOOCH WAUN Head 2 6.3 79055 1907 CESSPOOCH JUAN Head 35545638 1901 SESSPOOCH ELLEN Wife 2 11.9 79055 1902 CESSPOOCH ELLEN Wife 35545638 1934 SESSPOOCH LOUIS Son 5 11.2 79055 1933 CESSPOOCH LEWIS Son 35545638 1937 SESSPOOCH DEBOIA Daughter 3 6.3 79055 1936 CESSPOOCH DOVELIA Daughter

Notes: This table shows how household-information can increase the conﬁdence of individual level record linkages. No individual link looks very compelling as not a single person is a perfect match on age and name. Yet, viewed in combination, these two households are clearly the same.

20 For another summary, Ran Abramitzky’s website at https://people.stanford.edu/ranabr/ matching-codes.

38 Table A2 provides an example for the potential power of using household information in data like these. The table shows two four-person families (one in the ICR, and one in the 1940 FCC). The noteworthy feature of the table is that not a single person is a perfect match on age and name. The shared last name is different in the two data-sets, all birth-years are off by one year, and only one of four ﬁrst names is an exact match. Nonetheless, viewing the four records in combination gives a high degree of conﬁdence in this being the same household in the two data sets. This example illustrates the potential usefulness of household structure in determining individual links, when individual linkage variables are measured with a lot of noise. In the following we describe our step-by-step approach to record-linkage:

1) We ﬁrst use standard individual record linkage methods to establish for each individual in a

master data-set (say, the ICR), the set of all potentially linked individuals in a using data-set

(say, the 1930 FCC). This can be any of the commonly used approaches discussed in Bailey

et al.(2017), Abramitzky et al.(2019), and ??. We pursue a two-step approach: we ﬁrst

apply bigram-indexation to ﬁrst and last name to deﬁne the set of individuals in the using

data with similar names.21 We are ‘generous’ in this set, in the sense that we set the string-

similarity cutoff low enough that each person in the master data has an average of 8 links in

the using data.

We then trim that set to include only potential matches that deviate by at most 4 years in

birth-year for master records born after 1910, and by at most 6 years for master records born

before 1910.

2) Each individual in the master data and each individual in the using data is associated with

a unique household id. Consider a pair of households ICR-HH A in master and FCC-HH

B in using. Our approach is to calculate the number of individuals in ICR-HH A who are

linked to FCC-HH B in stage 1. This gives a metric of household similarity. We can then

upward-adjust the similarity-scores of individual links between individuals in ICR-HH A

and FCC-HH B by a percentage for each marginal increase in our metric of household sim-

ilarity. We also re-generate adjusted similarity-scores between all non-linked individuals in

21 We use the stringsim() function in R’s RecordLinkage, and block on gender and on ‘meta-state’ and gender. (A ‘meta-state’ is a set of two states whose boundaries are straddled by reservations. This occurs, e.g. in New Mexico and Arizona. We formed 40 meta-states.

39 ICR-HH A and FCC-HH B to potentially ﬁnd new links. The idea is that these new links

did not meet the match-threshold in the individual-level linkage in stage 1, but do meet the

threshold with the upward-adjusting. Household similarity therefore brings new individual

links into the fold.

The researcher has three levers to control the adjustment process in stage 2: One, household

similarity can be calculated in different ways. For example, we use the absolute number of

stage 1 individual links as our metric. This inherently creates more adjustments in larger

households than in smaller ones. One could alternatively use the share of household mem-

bers who are linked. As another alternative, one could also create a series of household-pair-

speciﬁc dummies for binary measures such as whether there is both a male and female head

of household, whether a household is a nuclear or an extended family, etc.

Two, the gradient of how individual similarity is adjusted to household similarity can be

controlled. With more aggressive upward-adjusting, the researcher puts more weight on

household-similarity. With less aggressive upward-adjusting, the researcher puts more weight

on individual similarity.

Three, one could deﬁne a series of explicit rules based on household types in a pair. For

example, if the only individual that is linked across ICR-HH A and FCC-HH B, is a ‘son’ in

the ICR, and a ‘head’ in the FCC, then two households are likely to be otherwise completely

disjoint sets, and the researcher might want to prevent the one exiting individual link’s simi-

larity score to be downward-adjusted (in relative terms) for the lack of household-similarity.

To what extent these levers are used will be dictated by the structure of the speciﬁc dataset

as well as by computational concerns.

3) Then a “best-off” gridsearch is applied on the adjusted similarity scores created in stage 2.

The grid-search maximizes the sum of similarity-scores of all links, with the constraint that

each individual record in the master data is linked to at most one record in the using data,

and each record in the using data is linked to at most one record in the master data.

4) Relative to individual linkage methods, steps 2 and 3 above have the potential to create

additional links to the extent that household similarity can lead individual similarity scores

to be scaled up. This is evident in Table A2 where no individual link looks particularly

40 compelling when viewed in isolation, but the links look very compelling when viewed in

combination.

More importantly, the household structure also suggests a range of criteria to generate ﬂags

for potentially problematic links, based on reported family relations. For example, while it

is reasonable for the husband and the wife in a household in the (later) master data to be

linked to records in the (earlier) using data where they are recorded as a son and a daughter,

they should then not share a household in the using data, lest they be brother and sister.22

The ﬁnal linked data-set will invariably be improved by adding a fourth stage of manually

checking links that that throw up ﬂags for family relations that are either logically consistent

or unlikely.

Manual linkage after stage 4 is greatly aided by the fact that stages 2–3 have identified the most likely individual link for each member of a given “anchor household” in the master data, so that one can view all members in the anchor household together with all person records belonging to households with any links to that anchor household. In other words, manual links can at this stage be confirmed or changed with all relevant household relations in view. Table A3 shows two examples of how our research assistants manually check flagged households in practice. The

ﬂagged household is the ”anchor household,” and associated with it is “network” of direct and indirect links to person-records in other datasets. In the top-panel, that anchor household has id 15426032 in the 1930 Full Count Census. In the bottom-panel, that anchor household has id

10 in the ICR. A network consists of all individuals that belong to any household linked to the anchor household. For instance, in the top-network, 1930-household 15426032 has one link to an individual in 1940-household 19002788, so that all ﬁve members of that household are included in the bloc. Household id, the number of people in a household, and individual id’s are reported by data-source; columns 1–3 report on the 1940 FCC, columns 4–6 report on the 1930 FCC, and

22 Of course, if master and using data are collected in the same year, we should expect family relations to be exactly identical in the two datasets. Any time gaps between the years in which master and using data were recorded will imply some differences in household structure between the data-sets. For example, the ICR was collected between 1930 and 1937 and is therefore likely to contain more households whose nucleus is unchanged. By contrast, when linking two decadal Census waves like the 1930 and 1940 FCC, a sizeable share of 1930 children will have formed new households in 1940, and become heads or spouses. As well, more 1930 household heads and spouses will have passed away, and two-parent households will thus have turned into single-head households.

41 columns 7–9 report on the ICR. One observations is one person-record. If a person in a 1930- household is individually linked to persons in the 1940 Census and the ICR after stage 2, then there are entries in columns 1–3, 4–6, and 7–9 of that person record.

Columns 16–33 report on six variables that represent the relevant person- and household- information, again separately for each of three data sources. These variables are a person’s ﬁrst and last name, birth-year, gender, their marital status, and their relation to the household head.

Based on this information, we can assess whether existing links are correct, and make manual adjustments in columns 10–15, which then get hardcoded into the data before re-running stages 2–3 of the program above. Columns 10-12 are ‘true positive’ links. A ‘true positive’ link can conﬁrm an existing link, or it can over-ride an existing link. Columns 13-15 are ‘true negative’ un-links, which declare that an existing link is wrong. To avoid clutter for illustrative purposes, Table A3 omits entries in column 10–15 that merely conﬁrm existing links, and only includes entries for records where at least one link gets manually over-ridden. In the top-record for example, columns

11-12 conﬁrm the existing link between the 1930 FCC and the ICR, but column 13 unlinks these from their link to a person-record in 1940. In the fourth row, columns 11-12 establish a new link between a 1930-record called Josephine Rose and an ICR-record called Josephine Rose who dif- fers by 11 years in birth-year but shares an ICR household with three others who are linked to

1930-records in Josephine Rose’s 1930 household. Column 15 in the same row lists the unlinked previous link.23 In the bottom-bloc, columns 11-12 link four previously unlinked individuals in the 1930 census to four previously unlinked individuals in the ICR. The records were not linked by the algorithm because of unstable last names (‘Big’ in the ICR and ‘Big Knife’ in the 1930 FCC) and ﬁrst names (‘Dorothy Ann’ in the ICR, and ‘Pretty Woman’ in the 1930 FCC).

5) Manual changes made in stage 4 are the ‘true positives’ and ‘true negatives’ in columns 10–

15 of Table A3.

Hard-coded ‘true positives pairs’ mean that any other individual links with either record

in a pair get deleted. Hard-coded ‘true negative pairs’ mean that this particular link gets

deleted from the stage 1 output.24 These become hard-coded and are fed into the data-output

23 This particular column 15 entry is redundant because columns 11-12 already hard-code a unique ‘true positive’ link between these records, thus precluding any other possible links. 24 These need to get deleted a second time at the end of stage 2 in case that any are re-introduced during stage 2 by the similarity-score adjustment.

42 Table A3: Household-Based Linkage Example

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33) 1940 1930 ICR true positives true negatives 1940 1930 ICR

Birth Mar Birth Mar Birth Mar hhid # pid hhid # pid hhid # pid pid 1940 pid 1930 pid ICR pid 1940 pid 1930 pid ICR LN FN Y Sex ried Rel LN FN Y Sex ried Rel LN FN Y Sex ried Rel Anchor-HH : hhid=15426032 in 1930 19002788 4 19002788-1 15426032 4 15426032-1 68934 5 9-806-11 15426032-1 9-806-11 19002788-1 RACE WILLIAM 1891 M M Head ROSE WILLIAM 1891 M M Head ROSE WILLIAM 1890 M M Head 19051844 5 19051844-2 15426032 4 15426032-4 68934 5 9-807-2 ROCE FOREST 1925 M gr-ch ROSE FORREST 1925 M child ROSELIAM FORREST 1924 M child 19051844 5 19051844-3 15426032 4 15426032-3 68934 5 9-807-1 ROCE JAUNITA 1922 F gr-ch ROSE WANETA 1923 F child ROSE JUANITA 1922 F child 15426032 4 15426032-2 67409 3 86-10-2 15426032-2 9-806-12 86-10-2 ROSEJOSEPHINE 1890 F M Sp SHANE JOSEPHINE 1891 F M Sp 19051844 5 19051844-5 15463552 8 15463552-7 68452 16 9-674-17 REDFOX CYNTHA 1922 F gr-ch RED FOX CYNTHIA 1922 F step-ch RED FOX CYNTHIA 1921 F step-ch 19051844 5 19051844-1 15463565 2 15463565-2 68933 2 9-806-10 ROCE AGNES 1861 F W Head ROSE AGNES 1862 F Sp ROSE AGNES 1860 F M step-ch 19066701 9 19066701-5 15477934 7 15477934-6 67409 3 86-10-3 BEAS CUB ARCHIE 1929 M child BEARCUB ARCHIE 1928 M childEARGROUND ARCHIBALD 1930 M Sp 19002788 4 19002788-4 68258 12 9-627-12 BEARDEAU MELVIN 1928 M step-ch BURDEAU MELVIN JOE 1927 M child 19002788 4 19002788-3 68258 12 9-627-11 BEARDEAU JAMES 1926 M step-ch BURDEAUAMES ADAMS 1924 M child 19002788 4 19002788-2 RACE MAGGIE 1890 F M Sp 19062368 11 19062368-5 68934 5 9-807-3 LA ROGUE DOROTHY 1930 F child ROSE DOROTHY 1930 F child 67409 3 86-10-1 ES TO OTHER GROUND 1870 M M Head 68934 5 9-806-12 ROSEOSEPHINE N S 1901 F M Sp

Anchor-HH : hhid=10 in ICR OLD OLD 19009442 3 19009442-1 15429377 8 15429377-6 10 9 426-617-4 BIG DOY WOMAN 1890 F W Head BIG DOG WOMAN 1868 F W Parent BIG DOG OLD WOMAN 1868 F W Head RUTHY PRETTY PRETTY 19009442 3 19009442-2 15429377 8 15429377-7 10 9 426-617-5 BIG DOY LODGE 1910 F child BIG DOG LODGE 1909 F Sibling BIG DOG LODGE 1908 F child 15429377 8 15429377-8 10 9 426-617-6 BIG DOG ALBERT 1912 M adop-ch BIG DOG ALBERT 1913 M neph 43 10 9 426-617-10 BIG KNIFE JOHN 1928 M child DOROTHY 10 9 426-617-9 BIG KNIFE ANN 1926 F child 10 9 426-617-11 BIG KNIFE JOSEPH 1933 M child 10 9 426-617-8 BIG KNIFE FLORA 1907 F M Sp 10 9 426-617-7 BIG KNIFE 1900 M M Head 10 9 426-617-12 BIG KNIFE DOROTHY 1935 F child

COLORE 19009442 3 19009442-3 FOUR GAESTOTHE 1887 F W child 15429377 8 15429377-1 15429377-1 426-617-7 BIG KINFE 1901 M M Head LEVEL 15429377 8 15429377-4 BIG KNIFE WALING 1928 F child PRETTY 15429377 8 15429377-3 15429377-3 426-617-9 BIG WOMAN 1926 F child 15429377 8 15429377-2 15429377-2 426-617-8 BIG FLORA S 1908 F M Sp HAD LIGHTENI 15429377 8 15429377-5 15429377-5 426-617-10 BIG KNIFE NG 1929 M child

Notes:(a) Columns 20, 26, 31: M=married, W=widowed. Columns 21, 27, 32: SP=spouse, step-ch=step-child, adop-child=adopted child, gr-child=grand-child. (b) This table reports on two “household networks” Each household network has one ”anchor household.” In the top-panel, that anchor household has id 15426032 in the 1930 Full Count Census. In the bottom-panel, that anchor household has id 10 in the ICR. Household id, the number of people in a household, and individual id’s are reported by data-source; columns 1–3 report on the 1940 FCC, columns 4–6 report on the 1930 FCC, and columns 7–9 report on the ICR. One observations is one person-record. If a person in a 1930-household is individually linked to persons in the 1940 Census and the ICR after stage 2, then there are entries in columns 1–3, 4–6, and 7–9 of that person record. Columns 10-12 are ‘true positive’ links. A ‘true positive’ link can conﬁrm an existing link, or it can over-ride an existing link. Columns 13-15 are ‘true negative’ un-links, which declare that an existing link is wrong. To avoid clutter for illustrative purposes, Table A3 omits entries in column 10–15 that merely conﬁrm existing links, and only includes entries for records where at least one link gets manually over-ridden. generated at the end of stage 1.

Table A4: Possible Household Links, Grouped by Internal Consistency (1) (2) (3) (4) (5) 1940 Possible Links to ICR Households Acceptable Matches :

Nuclear H  H non-H non-H  non-H Family Sp  Sp non-Sp non-Sp non-Sp 

H + widowed  Single Head H  H  Sp + widowed

Disregard until Manually Checked:

Nuclear H  H non-H H  H Family Sp  non-Sp Sp Sp Sp 

Notes:(a) The side of the table content to the left of the arrow ⇒ depicts the 1940 households for which we can study outcomes. Across columns 1–5, the right side of the table content depicts possible ways in which a 1940 household can be linked to ICR households. (b) The top-two panels depict linkages that are logically consistent, ﬁrstly for nuclear households with a head and a spouse, and secondly for single-head households. The bottom panel depicts linkages that appear logically inconsistent and should be checked before using these observations. (c) In the top panel: Column 1 indicates that both head and spouse in a 1940 FCC household are linked to a head and spouse who are in the same household in the ICR. Column 2 indicates that both 1940 head and spouse are linked to persons in the same household, and that neither of these is head or spouse in the ICR. This captures the relatively common occurrence of young couples living with one spouse’s parents. Column 3 indicates that 1940 head and the spouse are linked to persons in the ICR who live in different households and are neither head nor spouse in the ICR. This captures the most common occurrence of new households being formed by young adults. Columns 4–5 capture the case where only one of 1940 head or spouse is linked to a person record in the ICR, and that record is not a head or spouse in the ICR. Column 5 is common because young wives in 1940 who were living with their parents in the ICR are less likely to be linked because of their different last names. The middle panel depicts single-head households in which the head is linked to person-record in the ICR. Such links are consistent if the ICR person-record was also a single head or was neither head nor spouse (e.g. an adolescent living with parents); they are also consistent if the ICR person-record was part of a nuclear head-spouse pair, but the 1940 person-record is widowed or divorced. In the bottom panel: Columns 1–2 indicate that both head and spouse in a 1940 FCC household are linked to person-records in the same ICR-household but that these person-records had a relation to each other that is inconsistent with their 1940 relation. Column 3 indicates that 1940 head and spouse are both linked to person-records of the same relation but in different households. Columns 4–5 indicate that only of head or spouse are linked to person-records in the ICR, but that they are already a head/spouse.

The researcher can choose the decision rules that determine which records to manually check and link by hand. In our case, we deﬁned a number of different ﬂags for households with inconsistent linkages between the 1940 FCC and the ICR. Possible linkages and their consistency are visualized in Table A4. The side of the table content to the left of the arrow ⇒ depicts the 1940 households for which we can study outcomes. Across columns 1–5, the right side of the table content depicts possible ways in which a 1940 household can be linked to ICR households. The

44 top-two panels depict linkages that are logically consistent, ﬁrstly for nuclear households with a head and a spouse, and secondly for single-head households. In the top panel, column 1 indicates that both head and spouse in a 1940 FCC household are linked to a head and spouse who are in the same household in the ICR. Column 2 indicates that both 1940 head and spouse are linked to persons in the same household, and that neither of these is head or spouse in the ICR. This captures the relatively common occurrence of young couples living with one spouse’s parents. Column

3 indicates that 1940 head and the spouse are linked to persons in the ICR who live in different households and are neither head nor spouse in the ICR. This captures the most common occurrence of new households being formed by young adults. Columns 4–5 capture the case where only one of 1940 head or spouse is linked to a person record in the ICR, and that record is not a head or spouse in the ICR. Column 5 is common because young wives in 1940 who were living with their parents in the ICR are less likely to be linked because of their different last names. The middle panel depicts single-head households in which the head is linked to person-record in the ICR.

Such links are consistent if the ICR person-record was also a single head or was neither head nor spouse (e.g. an adolescent living with parents); they are also consistent if the ICR person-record was part of a nuclear head-spouse pair, but the 1940 person-record is widowed or divorced. The bottom panel depicts linkages that appear logically inconsistent. In the bottom panel, columns

1–2 indicate that both head and spouse in a 1940 FCC household are linked to person-records in the same ICR-household but that these person-records had a relation to each other that is inconsistent with their 1940 relation. Column 3 indicates that 1940 head and spouse are both linked to person-records of the same relation but in different households. Columns 4–5 indicate that only of head or spouse are linked to person-records in the ICR, but that they are already a head/spouse.

Links depicted in the bottom panel are “ﬂagged” for potential logical inconsistency, and should be omitted from the statistical analysis until they have been manually checked and linked (or de- linked).

Previous to stages 4–5 described above, we have roughly 7,000 Native American households from the Census to the ICR that are “un-ﬂagged”, i.e. that belong to the top-two panels in Table A4.

45 Alabama-Coushatta Reservation and Off-Reservation Trust Land Blackfeet Reservation

Appendix C Mapping from Individuals to Reservations in Data

Here we list all reservations in our data, together with their state in (·), their allotment status according to Ofﬁce of Indian Affairs(1935), and the number of households in 1940 with at least one

Native American member, the total population in these households, and the total tribal population in these households, with the difference between the two coming from non-Native spouses.

Alabama-Coushatta Reservation and Off- N, Number of Households 1940: 53, Total Popula-

Reservation Trust Land (TX) Alloted: N, Num- tion 1940: 223, Native Population 1940: 215. • ber of Households 1940: 67, Total Population 1940: Bay Mills Reservation and Off-

323, Native Population 1940: 321. • Reservation Trust Land (MI) Alloted: N, Num-

Allen Canyon Pah-Ute (CO, UT) Alloted: Y, ber of Households 1940: 34, Total Population 1940: Number of Households 1940: 85, Total Population 147, Native Population 1940: 140. •

1940: 305, Native Population 1940: 304. • Bear River Rancheria (CA) Alloted: Y, Num-

Alturas Indian Rancheria (CA) Alloted: N, ber of Households 1940: 44, Total Population 1940: Number of Households 1940: 39, Total Population 168, Native Population 1940: 154. •

1940: 146, Native Population 1940: 140. • Berry Creek Rancheria and Off-

Auburn Rancheria and Off-Reservation Reservation Trust Land (CA) Alloted: N, Num-

Trust Land (CA) Alloted: N, Number of House- ber of Households 1940: 29, Total Population 1940: holds 1940: 122, Total Population 1940: 474, Native 109, Native Population 1940: 103. •

Population 1940: 440. • Big Pine Reservation (CA) Alloted: N, Num-

Augustine Reservation (CA) Alloted: Y, ber of Households 1940: 59, Total Population 1940: Number of Households 1940: 17, Total Population 200, Native Population 1940: 195. •

1940: 62, Native Population 1940: 61. • Big Valley Rancheria (CA) Alloted: N, Num-

Bad River Reservation (WI) Alloted: Y, ber of Households 1940: 26, Total Population 1940: Number of Households 1940: 183, Total Population 118, Native Population 1940: 116. •

1940: 867, Native Population 1940: 846. • Bishop (CA) Alloted: N, Number of House-

Barona Reservation (CA) Alloted: N, Num- holds 1940: 11, Total Population 1940: 40, Native ber of Households 1940: 85, Total Population 1940: Population 1940: 37. •

375, Native Population 1940: 364. • Blackfeet Reservation (MT) Alloted: Y,

Battle Mountain Reservation (NV) Alloted: Number of Households 1940: 888, Total Population

46 Blue Lake Rancheria Chickasaw OTSA

1940: 4042, Native Population 1940: 3873. • tion (NY) Alloted: N, Number of Households 1940:

Blue Lake Rancheria (CA) Alloted: Y, Num- 158, Total Population 1940: 654, Native Population ber of Households 1940: 31, Total Population 1940: 1940: 646. •

107, Native Population 1940: 88. • Cayuga Indians at Tonawanda Reservation

Bois Forte Reservation (MN) Alloted: Y, (NY) Alloted: N, Number of Households 1940: Number of Households 1940: 108, Total Population 119, Total Population 1940: 472, Native Population

1940: 478, Native Population 1940: 465. • 1940: 470. •

Bridgeport Reservation (CA) Alloted: N, Cedar City Band (UT) Alloted: Y, Number of Number of Households 1940: 13, Total Population Households 1940: 32, Total Population 1940: 133,

1940: 41, Native Population 1940: 40. • Native Population 1940: 133. •

Brighton Reservation (FL) Alloted: N, Num- Chehalis Reservation and Off-Reservation ber of Households 1940: 33, Total Population 1940: Trust Land (WA) Alloted: Y, Number of House-

143, Native Population 1940: 142. • holds 1940: 47, Total Population 1940: 186, Native

Cabazon Reservation (CA) Alloted: Y, Num- Population 1940: 174. • ber of Households 1940: 35, Total Population 1940: Cherokee OTSA (AR, OK, KS) Alloted: Y, 126, Native Population 1940: 120. • Number of Households 1940: 3460, Total Popula-

Caddo-Wichita-Delaware OTSA (OK) Al- tion 1940: 13786, Native Population 1940: 12250. loted: Y, Number of Households 1940: 133, Total •

Population 1940: 640, Native Population 1940: 622. Cheyenne River Reservation and Off-

• Reservation Trust Land (SD) Alloted: Y, Num-

Camp Verde (AZ) Alloted: N, Number of ber of Households 1940: 522, Total Population 1940: Households 1940: 58, Total Population 1940: 223, 2323, Native Population 1940: 2248. •

Native Population 1940: 220. • Cheyenne-Arapaho OTSA (OK) Alloted: Y,

Campo (CA) Alloted: N, Number of House- Number of Households 1940: 698, Total Population holds 1940: 32, Total Population 1940: 123, Native 1940: 3243, Native Population 1940: 3009. •

Population 1940: 118. • Chickahominy SDTSA (VA) Alloted: N,

Catawba Reservation (SC) Alloted: N, Num- Number of Households 1940: 17, Total Population ber of Households 1940: 35, Total Population 1940: 1940: 112, Native Population 1940: 110. •

182, Native Population 1940: 177. • Chickasaw OTSA (OK) Alloted: Y, Number Cayuga Indians at Cattaraugus Reserva- of Households 1940: 1109, Total Population 1940:

47 Chicken Ranch Rancheria and Off-Reservation Trust Land Crow Reservation

4363, Native Population 1940: 3786. • Native Population 1940: 138. •

Chicken Ranch Rancheria and Off- Colville Reservation and Off-Reservation

Reservation Trust Land (CA) Alloted: N, Num- Trust Land (WA) Alloted: Y, Number of House- ber of Households 1940: 29, Total Population 1940: holds 1940: 579, Total Population 1940: 2217, Na-

119, Native Population 1940: 111. • tive Population 1940: 2063. •

Choctaw OTSA (AR, TX, OK) Alloted: Y, ”Coos, Lower Umpqua, and Siuslaw Reser-

Number of Households 1940: 2385, Total Population vation and Off-Reservation Trust Land (OR) 1940: 9185, Native Population 1940: 7885. • Alloted: N, Number of Households 1940: 26, Total

Citizen Potawatomi Nation-Absentee Population 1940: 119, Native Population 1940: 109.

Shawnee OTSA (OK) Alloted: Y, Number of • ”

Households 1940: 1111, Total Population 1940: Coquille Reservation (OR) Alloted: N, 4943, Native Population 1940: 4533. • Number of Households 1940: 11, Total Population

Cochiti Pueblo (NM) Alloted: N, Number of 1940: 36, Native Population 1940: 30. •

Households 1940: 62, Total Population 1940: 310, Coushatta Reservation and Off-

Native Population 1940: 310. • Reservation Trust Land (LA) Alloted: N, Num-

Cocopah (AZ) Alloted: N, Number of House- ber of Households 1940: 26, Total Population 1940: holds 1940: 59, Total Population 1940: 252, Native 116, Native Population 1940: 116. •

Population 1940: 249. • Cow Creek Reservation and Off-

Coeur d’Alene Reservation (ID) Alloted: Y, Reservation Trust Land (OR) Alloted: N, Num- Number of Households 1940: 84, Total Population ber of Households 1940: 21, Total Population 1940:

1940: 331, Native Population 1940: 321. • 84, Native Population 1940: 66. •

Coharie SDTSA (NC) Alloted: N, Number of Crandon Sub-Agency (WI) Alloted: N, Households 1940: 110, Total Population 1940: 631, Number of Households 1940: 64, Total Population

Native Population 1940: 630. • 1940: 275, Native Population 1940: 253. •

Colorado River Indian Reservation (AZ, Creek OTSA (OK) Alloted: Y, Number of

CA) Alloted: Y, Number of Households 1940: 150, Households 1940: 1254, Total Population 1940: Total Population 1940: 734, Native Population 5111, Native Population 1940: 4631. •

1940: 719. • Crow Reservation (MT) Alloted: Y, Num-

Colusa Rancheria (CA) Alloted: N, Number ber of Households 1940: 255, Total Population 1940: of Households 1940: 38, Total Population 1940: 145, 1326, Native Population 1940: 1278. •

48 Dresslerville Colony Fort McDowell Yavapai Nation Reservation

Dresslerville Colony (NV) Alloted: N, Num- Reservation Trust Land (WI, MN) Alloted: Y, ber of Households 1940: 66, Total Population 1940: Number of Households 1940: 187, Total Population

215, Native Population 1940: 213. • 1940: 875, Native Population 1940: 821. •

Dry Creek Rancheria (CA) Alloted: N, Num- Fort Apache Reservation (AZ) Alloted: N, ber of Households 1940: 29, Total Population 1940: Number of Households 1940: 774, Total Population

122, Native Population 1940: 114. • 1940: 3043, Native Population 1940: 3030. •

Eastern Cherokee (NC) Alloted: N, Num- Fort Belknap Reservation and Off- ber of Households 1940: 406, Total Population 1940: Reservation Trust Land (MT) Alloted: Y, Num- 2079, Native Population 1940: 2009. • ber of Households 1940: 422, Total Population 1940:

Elk Valley Rancheria and Off-Reservation 2026, Native Population 1940: 1976. •

Trust Land (CA) Alloted: Y, Number of House- Fort Berthold Reservation (ND) Alloted: Y, holds 1940: 40, Total Population 1940: 145, Native Number of Households 1940: 309, Total Population

Population 1940: 134. • 1940: 1666, Native Population 1940: 1634. •

Elko Colony (NV) Alloted: N, Number of Fort Bidwell Reservation and Off-

Households 1940: 34, Total Population 1940: 123, Reservation Trust Land (OR, CA) Alloted: N, Native Population 1940: 114. • Number of Households 1940: 22, Total Population

Ely Reservation (NV) Alloted: N, Number of 1940: 85, Native Population 1940: 85. •

Households 1940: 17, Total Population 1940: 63, Fort Hall Reservation and Off-Reservation

Native Population 1940: 62. • Trust Land (ID) Alloted: Y, Number of House-

Fallon Paiute-Shoshone Reservation and holds 1940: 344, Total Population 1940: 1471, Na-

Off-Reservation Trust Land (NV) Alloted: Y, tive Population 1940: 1423. •

Number of Households 1940: 77, Total Population Fort Independence Reservation (CA) Al- 1940: 293, Native Population 1940: 290. • loted: Y, Number of Households 1940: 19, Total Pop-

Flandreau (SD, MN) Alloted: N, Number of ulation 1940: 70, Native Population 1940: 68. •

Households 1940: 61, Total Population 1940: 197, Fort McDermitt Indian Reservation (OR,

Native Population 1940: 190. • NV) Alloted: Y, Number of Households 1940: 58,

Flathead Reservation (MT) Alloted: Y, Num- Total Population 1940: 242, Native Population ber of Households 1940: 556, Total Population 1940: 1940: 241. •

2233, Native Population 1940: 2038. • Fort McDowell Yavapai Nation Reserva-

Fond du Lac Reservation and Off- tion (AZ) Alloted: N, Number of Households 1940:

49 Fort Mojave Hopland Rancheria and Off-Reservation Trust Land

60, Total Population 1940: 199, Native Population 374, Native Population 1940: 364. •

1940: 199. • Greenville Rancheria (CA) Alloted: Y, Num-

Fort Mojave (CA, NV, AZ) Alloted: N, Num- ber of Households 1940: 58, Total Population 1940: ber of Households 1940: 82, Total Population 1940: 235, Native Population 1940: 221. •

301, Native Population 1940: 298. • Grindstone Indian Rancheria (CA) Alloted:

Fort Peck Reservation (MT) Alloted: Y, N, Number of Households 1940: 10, Total Popula- Number of Households 1940: 562, Total Population tion 1940: 31, Native Population 1940: 31. •

1940: 2586, Native Population 1940: 2520. • Hannahville Indian Community and Off-

Fort Yuma Indian Reservation (CA, AZ) Al- Reservation Trust Land (MI) Alloted: N, Num- loted: Y, Number of Households 1940: 108, Total ber of Households 1940: 22, Total Population 1940:

Population 1940: 537, Native Population 1940: 534. 94, Native Population 1940: 94. •

• Havasupai Reservation (AZ) Alloted: N,

Gila River Indian Reservation (AZ) Al- Number of Households 1940: 65, Total Population loted: Y, Number of Households 1940: 727, Total 1940: 205, Native Population 1940: 204. •

Population 1940: 3276, Native Population 1940: Ho-Chunk Nation Reservation and Off-

3265. • Reservation Trust Land (WI) Alloted: Y, Num-

Goshute (UT, NV) Alloted: N, Number of ber of Households 1940: 153, Total Population 1940: Households 1940: 31, Total Population 1940: 130, 783, Native Population 1940: 769. •

Native Population 1940: 129. • Hoopa Valley Reservation (CA) Alloted: Y,

Grand Portage Reservation and Off- Number of Households 1940: 366, Total Population

Reservation Trust Land (MN) Alloted: Y, Num- 1940: 1383, Native Population 1940: 1328. • ber of Households 1940: 59, Total Population 1940: Hopi Reservation and Off-Reservation

245, Native Population 1940: 225. • Trust Land (AZ) Alloted: N, Number of House-

Grand Ronde Community and Off- holds 1940: 2068, Total Population 1940: 10428,

Reservation Trust Land (OR) Alloted: Y, Num- Native Population 1940: 10417. • ber of Households 1940: 51, Total Population 1940: Hopland Rancheria and Off-Reservation

202, Native Population 1940: 192. • Trust Land (CA) Alloted: N, Number of House-

Grand Traverse Reservation and Off- holds 1940: 28, Total Population 1940: 87, Native

Reservation Trust Land (MI) Alloted: N, Num- Population 1940: 86. • ber of Households 1940: 87, Total Population 1940: Houlton Maliseet Reservation and Off-

50 Houlton Maliseet Reservation and Off-Reservation Trust Land Kaw/Ponca joint-use OTSA

Reservation Trust Land (ME) Alloted: N, Num- Jamul Indian Village (CA) Alloted: N, Num- ber of Households 1940: 19, Total Population 1940: ber of Households 1940: 34, Total Population 1940:

82, Native Population 1940: 73. • 87, Native Population 1940: 71. •

Huron Potawatomi Reservation and Off- Jemez Pueblo (NM) Alloted: N, Number of

Reservation Trust Land (MI) Alloted: N, Num- Households 1940: 151, Total Population 1940: 732, ber of Households 1940: 10, Total Population 1940: Native Population 1940: 731. •

46, Native Population 1940: 43. • Jena Band of Choctaw Reservation (LA)

Ione Band of Miwok TDSA (CA) Alloted: Alloted: N, Number of Households 1940: 5, Total N, Number of Households 1940: 19, Total Popula- Population 1940: 30, Native Population 1940: 29. • tion 1940: 68, Native Population 1940: 60. • Jicarilla Apache Nation Reservation and

Iowa (KS-NE) Reservation and Off- Off-Reservation Trust Land (NM) Alloted: Y,

Reservation Trust Land (KS) Alloted: Y, Number Number of Households 1940: 246, Total Population of Households 1940: 34, Total Population 1940: 141, 1940: 1148, Native Population 1940: 1141. •

Native Population 1940: 126. • Kaibib (AZ) Alloted: N, Number of House-

Iowa OTSA (OK) Alloted: Y, Number of holds 1940: 20, Total Population 1940: 76, Native Households 1940: 64, Total Population 1940: 282, Population 1940: 76. •

Native Population 1940: 262. • Kalispel Reservation and Off-Reservation

Isabella Reservation (MI) Alloted: Y, Num- Trust Land (WA) Alloted: Y, Number of House- ber of Households 1940: 58, Total Population 1940: holds 1940: 29, Total Population 1940: 114, Native

245, Native Population 1940: 231. • Population 1940: 112. •

Isleta Pueblo (NM) Alloted: N, Number of Karuk Reservation and Off-Reservation

Households 1940: 206, Total Population 1940: 806, Trust Land (CA) Alloted: N, Number of House- Native Population 1940: 800. • holds 1940: 110, Total Population 1940: 403, Native

Jackson Rancheria (CA) Alloted: N, Number Population 1940: 375. • of Households 1940: 11, Total Population 1940: 30, Kaw OTSA (OK, KS) Alloted: Y, Number of Native Population 1940: 28. • Households 1940: 56, Total Population 1940: 202,

Jamestown S’Klallam Reservation and Native Population 1940: 168. •

Off-Reservation Trust Land (WA) Alloted: N, Kaw/Ponca joint-use OTSA (OK) Alloted: Number of Households 1940: 31, Total Population Y, Number of Households 1940: 29, Total Popula-

1940: 98, Native Population 1940: 90. • tion 1940: 92, Native Population 1940: 74. •

51 Kickapoo OTSA Lower Brule Reservation and Off-Reservation Trust Land

Kickapoo OTSA (OK) Alloted: Y, Number of Lac du Flambeau Reservation (WI) Alloted: Households 1940: 35, Total Population 1940: 150, Y, Number of Households 1940: 207, Total Popula-

Native Population 1940: 148. • tion 1940: 941, Native Population 1940: 917. •

Kiowa-Comanche-Apache-Fort Sill Lake Traverse Reservation and Off-

Apache OTSA (TX, OK) Alloted: Y, Number Reservation Trust Land (MN, ND, SD) Alloted: of Households 1940: 372, Total Population 1940: Y, Number of Households 1940: 348, Total Popu-

2030, Native Population 1940: 1974. • lation 1940: 1681, Native Population 1940: 1638.

Kiowa-Comanche-Apache-Ft Sill Apache/Caddo-•

Wichita-Delaware joint-use OTSA (OK) Al- Las Vegas Colony (NV) Alloted: N, Number loted: Y, Number of Households 1940: 182, Total of Households 1940: 14, Total Population 1940: 48,

Population 1940: 846, Native Population 1940: 805. Native Population 1940: 47. •

• Leech Lake Reservation and Off-

Klamath Reservation (OR) Alloted: Y, Num- Reservation Trust Land (MN) Alloted: Y, Num- ber of Households 1940: 343, Total Population 1940: ber of Households 1940: 579, Total Population 1940:

1236, Native Population 1940: 1163. • 2716, Native Population 1940: 2612. •

Kootenai Reservation and Off-Reservation Little River Reservation and Off-

Trust Land (ID) Alloted: Y, Number of Households Reservation Trust Land (MI) Alloted: N, Num- 1940: 25, Total Population 1940: 93, Native Popu- ber of Households 1940: 59, Total Population 1940: lation 1940: 92. • 235, Native Population 1940: 218. •

L’Anse Reservation and Off-Reservation Little Traverse Bay Reservation and Off-

Trust Land (MI) Alloted: Y, Number of House- Reservation Trust Land (MI) Alloted: N, Num- holds 1940: 123, Total Population 1940: 478, Native ber of Households 1940: 140, Total Population 1940:

Population 1940: 454. • 593, Native Population 1940: 568. •

Lac Courte Oreilles Reservation and Off- Lookout Rancheria (CA) Alloted: N, Num-

Reservation Trust Land (WI) Alloted: Y, Num- ber of Households 1940: 19, Total Population 1940: ber of Households 1940: 269, Total Population 1940: 69, Native Population 1940: 69. •

1237, Native Population 1940: 1209. • Lower Brule Reservation and Off-

Lac Vieux Desert Reservation (WI, MI) Al- Reservation Trust Land (SD) Alloted: Y, Num- loted: N, Number of Households 1940: 8, Total Pop- ber of Households 1940: 319, Total Population 1940: ulation 1940: 28, Native Population 1940: 24. • 1422, Native Population 1940: 1398. •

52 Lower Elwha Reservation and Off-Reservation Trust Land Namba

Lower Elwha Reservation and Off- Menominee Reservation and Off-

Reservation Trust Land (WA) Alloted: N, Num- Reservation Trust Land (WI) Alloted: N, Num- ber of Households 1940: 9, Total Population 1940: ber of Households 1940: 491, Total Population 1940:

45, Native Population 1940: 41. • 2373, Native Population 1940: 2284. •

Lower Sioux Indian Community (MN) Al- Mescalero Reservation (NM) Alloted: N, loted: N, Number of Households 1940: 36, Total Number of Households 1940: 206, Total Population

Population 1940: 186, Native Population 1940: 177. 1940: 841, Native Population 1940: 831. •

• Mille Lacs Reservation and Off-

Lumbee SDTSA (SC, NC) Alloted: N, Num- Reservation Trust Land (WI, MN) Alloted: Y, ber of Households 1940: 2657, Total Population Number of Households 1940: 118, Total Population

1940: 14964, Native Population 1940: 14872. • 1940: 497, Native Population 1940: 488. •

Lummi Reservation (WA) Alloted: Y, Num- Mississippi Choctaw (MS) Alloted: N, Num- ber of Households 1940: 126, Total Population 1940: ber of Households 1940: 293, Total Population 1940:

554, Native Population 1940: 540. • 1482, Native Population 1940: 1476. •

Lytton Rancheria (CA) Alloted: N, Number Moapa River Indian Reservation (NV) Al- of Households 1940: 61, Total Population 1940: 151, loted: N, Number of Households 1940: 34, Total

Native Population 1940: 119. • Population 1940: 138, Native Population 1940: 138.

Makah Indian Reservation (WA) Alloted: Y, •

Number of Households 1940: 103, Total Population Mono County (CA) Alloted: Y, Number of 1940: 451, Native Population 1940: 445. • Households 1940: 19, Total Population 1940: 58,

Match-e-be-nash-she-wish Band of Pot- Native Population 1940: 52. • tawatomi Reservation (MI) Alloted: N, Number Morongo Reservation and Off-Reservation of Households 1940: 17, Total Population 1940: 104, Trust Land (CA) Alloted: Y, Number of House- Native Population 1940: 101. • holds 1940: 29, Total Population 1940: 100, Native

Mdewakanton Sioux Indians (MN) Alloted: Population 1940: 91. • N, Number of Households 1940: 22, Total Popula- Muckleshoot Reservation and Off- tion 1940: 67, Native Population 1940: 59. • Reservation Trust Land (WA) Alloted: Y, Num-

Mechoopda TDSA (CA) Alloted: N, Number ber of Households 1940: 112, Total Population 1940: of Households 1940: 11, Total Population 1940: 49, 380, Native Population 1940: 349. •

Native Population 1940: 48. • Namba (NM) Alloted: N, Number of House-

53 Navajo Nation Reservation and Off-Reservation Trust Land Paskenta Rancheria holds 1940: 30, Total Population 1940: 121, Native ber of Households 1940: 210, Total Population 1940:

Population 1940: 120. • 1076, Native Population 1940: 1052. •

Navajo Nation Reservation and Off- Oneida (WI) Reservation and Off-

Reservation Trust Land (UT, AZ, NM) Alloted: Reservation Trust Land (WI) Alloted: Y, Num- Y, Number of Households 1940: 8187, Total Popula- ber of Households 1940: 324, Total Population 1940: tion 1940: 40226, Native Population 1940: 40142. 1433, Native Population 1940: 1392. •

• Oneidas at Onondaga (NY) Alloted: N,

Nez Perce Reservation (ID) Alloted: Y, Num- Number of Households 1940: 212, Total Population ber of Households 1940: 211, Total Population 1940: 1940: 922, Native Population 1940: 894. •

908, Native Population 1940: 887. • Onondagas at Allegany (NY, PA) Alloted:

Nisqually Reservation (WA) Alloted: Y, N, Number of Households 1940: 217, Total Popu- Number of Households 1940: 27, Total Population lation 1940: 843, Native Population 1940: 815. •

1940: 120, Native Population 1940: 116. • Onondagas at Tuscarora (NY) Alloted: N,

Nooksack Reservation and Off- Number of Households 1940: 76, Total Population

Reservation Trust Land (WA) Alloted: Y, Num- 1940: 291, Native Population 1940: 256. • ber of Households 1940: 29, Total Population 1940: Osage Reservation (KS, OK) Alloted: Y, 130, Native Population 1940: 128. • Number of Households 1940: 678, Total Population

North Fork Rancheria and Off-Reservation 1940: 2471, Native Population 1940: 2082. •

Trust Land (CA) Alloted: Y, Number of House- Otoe-Missouria OTSA (OK) Alloted: Y, holds 1940: 66, Total Population 1940: 277, Native Number of Households 1940: 63, Total Population

Population 1940: 265. • 1940: 329, Native Population 1940: 325. •

Northern Cheyenne Indian Reservation Ottawa OTSA (OK) Alloted: Y, Number of and Off-Reservation Trust Land (MT, SD) Al- Households 1940: 144, Total Population 1940: 473, loted: Y, Number of Households 1940: 365, Total Native Population 1940: 366. •

Population 1940: 1679, Native Population 1940: Pamunkey (state) Reservation (VA) Alloted: 1647. • N, Number of Households 1940: 24, Total Popula-

Ohkay Owingeh (NM) Alloted: N, Number tion 1940: 99, Native Population 1940: 99. • of Households 1940: 106, Total Population 1940: Paskenta Rancheria (CA) Alloted: N, Num- 555, Native Population 1940: 545. • ber of Households 1940: 15, Total Population 1940:

Omaha Reservation (NE) Alloted: Y, Num- 58, Native Population 1940: 52. •

54 Passamaquoddy Trust Land Quapaw OTSA

Passamaquoddy Trust Land (ME) Alloted: Trust Land (IN, MI) Alloted: N, Number of N, Number of Households 1940: 42, Total Popula- Households 1940: 24, Total Population 1940: 94, tion 1940: 181, Native Population 1940: 178. • Native Population 1940: 87. •

Pawnee OTSA (OK) Alloted: Y, Number of Ponca (NE) Trust Land (IA, SD, NE) Alloted: Households 1940: 99, Total Population 1940: 439, Y, Number of Households 1940: 47, Total Population

Native Population 1940: 418. • 1940: 185, Native Population 1940: 167. •

Pechanga Reservation (CA) Alloted: Y, Ponca OTSA (OK) Alloted: Y, Number of Number of Households 1940: 23, Total Population Households 1940: 114, Total Population 1940: 733,

1940: 76, Native Population 1940: 73. • Native Population 1940: 723. •

Penobscot Reservation and Off- Port Gamble Reservation (WA) Alloted: N,

Reservation Trust Land (ME) Alloted: N, Num- Number of Households 1940: 22, Total Population ber of Households 1940: 82, Total Population 1940: 1940: 97, Native Population 1940: 90. •

361, Native Population 1940: 353. • Port Madison Reservation (WA) Alloted: N,

Peoria OTSA (OK) Alloted: Y, Number of Number of Households 1940: 50, Total Population Households 1940: 43, Total Population 1940: 182, 1940: 191, Native Population 1940: 181. •

Native Population 1940: 164. • Prairie Band of Potawatomi Nation Reser-

Pine Ridge Reservation (NE, SD) Alloted: Y, vation (KS) Alloted: Y, Number of Households Number of Households 1940: 1641, Total Population 1940: 16, Total Population 1940: 55, Native Pop-

1940: 7703, Native Population 1940: 7531. • ulation 1940: 51. •

Pinoleville Rancheria (CA) Alloted: Y, Prairie Island Indian Community and Off-

Number of Households 1940: 72, Total Population Reservation Trust Land (MN) Alloted: N, Num- 1940: 325, Native Population 1940: 316. • ber of Households 1940: 21, Total Population 1940:

Pleasant Point Reservation (ME) Alloted: N, 93, Native Population 1940: 93. •

Number of Households 1940: 31, Total Population Puyallup (WA) Alloted: N, Number of House- 1940: 124, Native Population 1940: 124. • holds 1940: 100, Total Population 1940: 417, Native

Poarch Creek Reservation and Off- Population 1940: 396. •

Reservation Trust Land (FL, AL) Alloted: N, Pyramid Lake (NV) Alloted: N, Number of Number of Households 1940: 49, Total Population Households 1940: 117, Total Population 1940: 506,

1940: 244, Native Population 1940: 241. • Native Population 1940: 501. •

Pokagon Reservation and Off-Reservation Quapaw OTSA (MO, OK) Alloted: N, Num-

55 Quartz Valley Reservation and Off-Reservation Trust Land San Felipe Pueblo ber of Households 1940: 55, Total Population 1940: Rosebud Reservation (NE, SD) Alloted: Y, 190, Native Population 1940: 157. • Number of Households 1940: 1006, Total Population

Quartz Valley Reservation and Off- 1940: 4775, Native Population 1940: 4649. •

Reservation Trust Land (CA) Alloted: Y, Num- Round Valley Reservation and Off- ber of Households 1940: 39, Total Population 1940: Reservation Trust Land (CA) Alloted: N, Num- 136, Native Population 1940: 114. • ber of Households 1940: 133, Total Population 1940:

Quileute Reservation (WA) Alloted: Y, 551, Native Population 1940: 525. • Number of Households 1940: 30, Total Population Rumsey Indian Rancheria (CA) Alloted: N,

1940: 98, Native Population 1940: 97. • Number of Households 1940: 44, Total Population

Quinault Reservation (WA) Alloted: Y, 1940: 206, Native Population 1940: 195. • Number of Households 1940: 60, Total Population Sac and Fox of Iowa (IA) Alloted: N, Number

1940: 254, Native Population 1940: 245. • of Households 1940: 85, Total Population 1940: 419,

Red Cliff Reservation and Off-Reservation Native Population 1940: 417. •

Trust Land (WI) Alloted: Y, Number of House- Sac and Fox Nation Reservation and Off- holds 1940: 131, Total Population 1940: 655, Native Reservation Trust Land (KS, NE) Alloted: Y,

Population 1940: 630. • Number of Households 1940: 14, Total Population

Red Lake Reservation (MN) Alloted: N, 1940: 55, Native Population 1940: 50. • Number of Households 1940: 484, Total Population Sac and Fox OTSA (OK) Alloted: Y, Number

1940: 2370, Native Population 1940: 2328. • of Households 1940: 38, Total Population 1940: 186,

Redding Rancheria (CA) Alloted: N, Num- Native Population 1940: 175. • ber of Households 1940: 178, Total Population 1940: Salt River Reservation (AZ) Alloted: Y,

627, Native Population 1940: 580. • Number of Households 1940: 248, Total Population

Reno-Sparks Indian Colony (NV, CA) Al- 1940: 1121, Native Population 1940: 1096. • loted: N, Number of Households 1940: 76, Total Samish TDSA (WA) Alloted: N, Number of

Population 1940: 281, Native Population 1940: 272. Households 1940: 48, Total Population 1940: 131, • Native Population 1940: 103. •

Rocky Boy’s Reservation and Off- San Carlos (AZ) Alloted: N, Number of

Reservation Trust Land (MT) Alloted: N, Num- Households 1940: 703, Total Population 1940: 2791, ber of Households 1940: 179, Total Population 1940: Native Population 1940: 2784. •

847, Native Population 1940: 826. • San Felipe Pueblo (NM) Alloted: N, Num-

56 San Ildefonso Pueblo and Off-Reservation Trust LandSiletzReservation and Off-Reservation Trust Land ber of Households 1940: 112, Total Population 1940: Sappony SDTSA (NC) Alloted: N, Number 627, Native Population 1940: 627. • of Households 1940: 32, Total Population 1940: 211,

San Ildefonso Pueblo and Off-Reservation Native Population 1940: 211. •

Trust Land (NM) Alloted: N, Number of House- Sauk-Suiattle Reservation (WA) Alloted: N, holds 1940: 29, Total Population 1940: 132, Native Number of Households 1940: 20, Total Population

Population 1940: 132. • 1940: 72, Native Population 1940: 70. •

San Manuel (CA) Alloted: N, Number of Sault Sainte Marie Reservation and Off-

Households 1940: 52, Total Population 1940: 198, Reservation Trust Land (MI) Alloted: N, Num- Native Population 1940: 183. • ber of Households 1940: 78, Total Population 1940:

San Pasqual (CA) Alloted: N, Number of 326, Native Population 1940: 302. •

Households 1940: 60, Total Population 1940: 239, Seminole OTSA (OK) Alloted: Y, Number of Native Population 1940: 222. • Households 1940: 36, Total Population 1940: 195,

Sandia Pueblo (NM) Alloted: N, Number of Native Population 1940: 195. •

Households 1940: 93, Total Population 1940: 299, Sherwood Valley Rancheria and Off-

Native Population 1940: 274. • Reservation Trust Land (CA) Alloted: N, Num-

Santa Clara Pueblo (NM) Alloted: N, Num- ber of Households 1940: 11, Total Population 1940: ber of Households 1940: 87, Total Population 1940: 50, Native Population 1940: 46. •

429, Native Population 1940: 426. • Shingle Springs Rancheria (CA) Alloted: N,

Santa Rosa Rancheria (CA) Alloted: N, Number of Households 1940: 26, Total Population Number of Households 1940: 26, Total Population 1940: 78, Native Population 1940: 63. •

1940: 123, Native Population 1940: 118. • Shoalwater Bay Indian Reservation and

Santa Ynez (CA) Alloted: N, Number of Off-Reservation Trust Land (WA) Alloted: Y, Households 1940: 30, Total Population 1940: 100, Number of Households 1940: 32, Total Population

Native Population 1940: 89. • 1940: 102, Native Population 1940: 87. •

Santee Reservation (SD, NE) Alloted: Y, Sia Pueblo (NM) Alloted: N, Number of Number of Households 1940: 93, Total Population Households 1940: 45, Total Population 1940: 223,

1940: 355, Native Population 1940: 346. • Native Population 1940: 222. •

Santo Domingo Pueblo (NM) Alloted: N, Siletz Reservation and Off-Reservation

Number of Households 1940: 199, Total Population Trust Land (OR) Alloted: Y, Number of House- 1940: 961, Native Population 1940: 961. • holds 1940: 92, Total Population 1940: 350, Native

57 Skokomish Reservation Tesuque Pueblo and Off-Reservation Trust Land

Population 1940: 318. • holds 1940: 33, Total Population 1940: 136, Native

Skokomish Reservation (WA) Alloted: Y, Population 1940: 129. •

Number of Households 1940: 47, Total Population St. Regis Mohawk Reservation (NY) Al- 1940: 229, Native Population 1940: 225. • loted: N, Number of Households 1940: 337, To-

Smith River Rancheria and Off- tal Population 1940: 1411, Native Population 1940:

Reservation Trust Land (OR, CA) Alloted: Y, 1383. •

Number of Households 1940: 32, Total Population Standing Rock Reservation (ND, SD) Al- 1940: 110, Native Population 1940: 105. • loted: Y, Number of Households 1940: 737, Total

Soboba (CA) Alloted: N, Number of House- Population 1940: 3302, Native Population 1940: holds 1940: 92, Total Population 1940: 295, Native 3241. •

Population 1940: 288. • Stockbridge Munsee Community (WI) Al-

Sokaogon Chippewa Community and Off- loted: Y, Number of Households 1940: 167, Total

Reservation Trust Land (WI) Alloted: N, Num- Population 1940: 713, Native Population 1940: 668. ber of Households 1940: 24, Total Population 1940: •

104, Native Population 1940: 101. • Susanville Indian Rancheria and Off-

Southern Ute Reservation (CO) Alloted: Y, Reservation Trust Land (CA) Alloted: N, Num- Number of Households 1940: 69, Total Population ber of Households 1940: 41, Total Population 1940:

1940: 317, Native Population 1940: 311. • 146, Native Population 1940: 136. •

Spirit Lake Reservation (ND) Alloted: Y, Swinomish Reservation and Off- Number of Households 1940: 255, Total Population Reservation Trust Land (WA) Alloted: N, Num-

1940: 1165, Native Population 1940: 1133. • ber of Households 1940: 69, Total Population 1940:

Spokane Reservation and Off-Reservation 283, Native Population 1940: 282. •

Trust Land (WA) Alloted: Y, Number of House- Sycuan Reservation (CA) Alloted: Y, Num- holds 1940: 113, Total Population 1940: 535, Native ber of Households 1940: 26, Total Population 1940:

Population 1940: 513. • 98, Native Population 1940: 90. •

Squaxin Island (WA) Alloted: Y, Number of Table Bluff Reservation (CA) Alloted: Y, Households 1940: 24, Total Population 1940: 67, Number of Households 1940: 82, Total Population

Native Population 1940: 66. • 1940: 235, Native Population 1940: 193. •

St. Croix Reservation and Off-Reservation Tesuque Pueblo and Off-Reservation Trust

Trust Land (WI) Alloted: N, Number of House- Land (NM) Alloted: N, Number of Households

58 Timbi-Sha Shoshone Reservation and Off-Reservation Trust LandWampanoag-Aquinnah Trust Land

1940: 38, Total Population 1940: 177, Native Popu- ber of Households 1940: 235, Total Population 1940: lation 1940: 174. • 1032, Native Population 1940: 999. •

Timbi-Sha Shoshone Reservation and Off- Umatilla Reservation (OR) Alloted: Y, Num-

Reservation Trust Land (NV) Alloted: N, Num- ber of Households 1940: 144, Total Population 1940: ber of Households 1940: 9, Total Population 1940: 595, Native Population 1940: 566. •

30, Native Population 1940: 27. • United Houma Nation SDTSA (LA) Al- Tohono O’odham Nation Reservation and loted: N, Number of Households 1940: 213, To-

Off-Reservation Trust Land (AZ) Alloted: Y, tal Population 1940: 1045, Native Population 1940: Number of Households 1940: 1457, Total Population 1024. •

1940: 7401, Native Population 1940: 7356. • Upper Lake Rancheria (CA) Alloted: N,

Tonkawa OTSA (OK) Alloted: Y, Number of Number of Households 1940: 54, Total Population Households 1940: 158, Total Population 1940: 673, 1940: 194, Native Population 1940: 194. •

Native Population 1940: 617. • Upper Sioux Community and Off-

Trinidad Rancheria and Off-Reservation Reservation Trust Land (MN) Alloted: N, Num-

Trust Land (CA) Alloted: N, Number of House- ber of Households 1940: 22, Total Population 1940: holds 1940: 37, Total Population 1940: 138, Native 103, Native Population 1940: 102. •

Population 1940: 128. • Upper South Carolina Pee Dee SDTSA

Tulalip Reservation and Off-Reservation (SC) Alloted: N, Number of Households 1940:

Trust Land (WA) Alloted: Y, Number of House- 28, Total Population 1940: 166, Native Population holds 1940: 123, Total Population 1940: 494, Native 1940: 165. •

Population 1940: 472. • Waccamaw Siouan SDTSA (NC) Alloted: N,

TuleRiver (CA) Alloted: N, Number of House- Number of Households 1940: 63, Total Population holds 1940: 39, Total Population 1940: 163, Native 1940: 305, Native Population 1940: 305. •

Population 1940: 151. • Walapai (AZ) Alloted: N, Number of House- Turtle Mountain Reservation and Off- holds 1940: 92, Total Population 1940: 370, Native

Reservation Trust Land (MT, ND) Alloted: Y, Population 1940: 366. •

Number of Households 1940: 758, Total Population Wampanoag-Aquinnah Trust Land (MA) 1940: 3898, Native Population 1940: 3814. • Alloted: N, Number of Households 1940: 35, To-

Uintah and Ouray Reservation and Off- tal Population 1940: 113, Native Population 1940:

Reservation Trust Land (UT) Alloted: Y, Num- 106. •

59 Warm Springs Reservation and Off-Reservation Trust LandZuniReservation and Off-Reservation Trust Land

Warm Springs Reservation and Off- N, Number of Households 1940: 10, Total Popula-

Reservation Trust Land (WA, OR) Alloted: Y, tion 1940: 36, Native Population 1940: 33. •

Number of Households 1940: 199, Total Population Wyandotte OTSA (OK) Alloted: Y, Number 1940: 834, Native Population 1940: 796. • of Households 1940: 95, Total Population 1940: 390,

Wells Colony (NV) Alloted: N, Number of Native Population 1940: 339. • Households 1940: 28, Total Population 1940: 123, Yakama Nation Reservation and Off-

Native Population 1940: 122. • Reservation Trust Land (WA, OR) Alloted: Y,

Western Shoshone (ID, NV) Alloted: N, Number of Households 1940: 576, Total Population Number of Households 1940: 147, Total Population 1940: 2502, Native Population 1940: 2400. •

1940: 650, Native Population 1940: 640. • Yankton Reservation (NE, SD) Alloted: Y, White Earth Reservation and Off- Number of Households 1940: 292, Total Population

Reservation Trust Land (MN) Alloted: Y, Num- 1940: 1339, Native Population 1940: 1306. • ber of Households 1940: 637, Total Population 1940: Yavapai-Prescott Reservation (AZ) Alloted: 3186, Native Population 1940: 3038. • N, Number of Households 1940: 21, Total Popula-

Wind River Reservation and Off- tion 1940: 71, Native Population 1940: 69. •

Reservation Trust Land (WY) Alloted: Y, Num- Ysleta del Sur Pueblo and Off-Reservation ber of Households 1940: 455, Total Population 1940: Trust Land (TX) Alloted: N, Number of House- 2044, Native Population 1940: 1997. • holds 1940: 9, Total Population 1940: 31, Native

Winnebago Reservation and Off- Population 1940: 28. •

Reservation Trust Land (NE, IA) Alloted: Y, Zuni Reservation and Off-Reservation

Number of Households 1940: 276, Total Population Trust Land (NM) Alloted: N, Number of House- 1940: 1164, Native Population 1940: 1129. • holds 1940: 457, Total Population 1940: 2352, Na-

Winnemucca Indian Colony (NV) Alloted: tive Population 1940: 2350. •