Chapter 3 — The Northwest Ordinance Template

Ohio to Florida, 1803–1848

“This isn’t just government land; it’s trust land. And just like any other trust, the trustee — which is the state — has a fiduciary duty to act in the best interest of the beneficiary.” — Margaret Bird, August 2025 video transcript

What this chapter does. I want to walk you through the chapter where the federal trust template moves from a sentence on a page to ten state legislatures and forty-five years of practice. The Confederation Congress wrote the architecture in 1785 and 1787; Ohio in 1803 was the first state asked to actually live inside it. By Florida in 1845 the country had run the experiment ten times, and the pattern was visible. The architecture worked. The architecture was not enough.

I. A blueprint older than the Constitution

In the spring of 1787 the Confederation Congress was, by most contemporary measures, a failing institution. It could not pay its soldiers. It could not raise reliable revenue. It could not, on most weeks, get nine states in the same room on the same Tuesday. And yet, in the months before the delegates in Philadelphia sat down to draft a Constitution that would replace it, that same broken Congress passed two of the most consequential statutes in American history. Together they laid down a continental-scale infrastructure for public education before there was a continent to put it on.

The first was the Land Ordinance of May 20, 1785. It instructed federal surveyors to lay out the public domain west of the Appalachians in a regular grid — six-mile-square townships, each subdivided into thirty-six numbered sections of one square mile — and reserved one of those thirty-six, section 16, near the geometric center of every township, “for the maintenance of public schools within the said township.”¹ The second was the Northwest Ordinance of July 13, 1787, governing the territory north of the Ohio River. Its Article III stated the principle the surveyors’ grid was meant to serve, in language that has never been bettered:

Religion, morality, and knowledge being necessary to good government and the happiness of mankind, schools and the means of education shall forever be encouraged.²

Two months after that sentence was passed, in a hotter room a few hundred miles east, a different group of delegates was still arguing about whether the new federal government should be allowed to tax. The Northwest Ordinance pre-dates the Constitution. It is older bedrock. And the schools clause of Article III, paired with the section-16 reservation of the 1785 Land Ordinance, represented something no nation had ever attempted: a continental-scale infrastructure for public education, financed not by taxation, not by appropriation, not by charity, but by land set aside before the settlers arrived.

I want to slow down and let the audacity register. A republic with almost no settled population beyond the Appalachians, no civil service, no public schools to speak of, no theory of universal literacy that had been tested at scale — and Congress was reserving one thirty-sixth of every future township from the Ohio River to whatever lay beyond, in trust for schools that did not yet exist, in towns that had not yet been named, for children whose grandparents had not yet been born. The army got guns. The schools got land. Each carries on, generation after generation, by a different kind of momentum. Guns get appropriated every two years. Land, once set aside, stays set aside — unless someone, somewhere, decides otherwise.

This chapter is the story of what happened to that blueprint when it ran into ten state legislatures over forty-five years. The ten states are: Ohio (1803), Louisiana (1812), Indiana (1816), Mississippi (1817), Illinois (1818), Alabama (1819), Missouri (1821), Arkansas (1836), Michigan (1837), and Florida (1845). They are an architecturally coherent set — every one of them admitted under the same lean federal template, every one of them endowed with section 16 of every township for the support of common schools. By the end of the nineteenth century, in most of them, the corpus that should have come from those grants was substantially gone.

I want to put down the chapter’s thesis at the front, because it organizes everything that follows. The architecture worked. The architecture was not enough. And the gap between those two sentences is what this chapter is for.

II. Ohio, 1803: the template’s first instantiation

Ohio is admission number seventeen, the first state carved out of the Northwest Territory, and the first test of whether the 1785–1787 architecture would survive contact with statehood. The 1802 Enabling Act drafted by Congress for Ohio is the first operative federal grant of school lands to a state. Section 7 of the Act offered:

the section number sixteen, in every township sold, or directed to be sold, by the United States … shall be granted to the inhabitants of such township, for the use of schools.³

The corresponding state-acceptance language, embedded in the Ohio enabling-act compact, vested those sections “in the legislature of the said state, in trust for the use aforesaid, and for no other use, intent or purpose whatever.” That formulation matters. In trust for the use aforesaid, and for no other use, intent or purpose whatever is, on its face, the strongest restriction-on-use clause in any of the early enabling acts — stronger than the section-7 grant language itself. It says, in plain legislative English, that the legislature is taking these sections as trustee, that the use is fixed, and that no other use is permitted. I belabor the point because the same formula travels forward, almost word for word, into one admission act after another for the next four decades. It is the seed of every “in-trust” clause in the federal-grant tradition.

Three other things about that 1802 sentence will matter for the next two centuries.

First, the federal text is short. It reserves the land, names the beneficiary, names the purpose, and stops. There is no express enforcement clause. There is no restoration mechanism if the lands are diverted. There is no provision authorizing the federal Attorney General to sue if the state breaches. There is no use of the phrase “breach of trust” anywhere in the federal text. The architecture is lean — load-bearing in its essentials, decorative in nothing. By the standards Congress would adopt a century later for New Mexico and Arizona, the 1802 Act is a blueprint with the framing nailed but no sheathing on the walls.

Second, the operative grant runs to the inhabitants of such township — not to the State, not to a state-level board, not to a permanent fund. The architectural decision in 1802 was that schools would be administered locally, by the people who lived nearest the schoolhouse. This was a Jeffersonian commitment, and on its own terms it was admirable: distributed authority, local accountability, no distant capital deciding whose schoolchildren got what.

I want to be fair to the Jeffersonian instinct, because as someone whose career has been spent watching trust administrations succeed and fail, I think it was honorably motivated and I think it was, in retrospect, the structural weakness that doomed most of the corpus. Township trustees with a few hundred neighbors and a handful of section-16 acres do not have the patience, the institutional memory, the audit capacity, or the political insulation of a state-level fiduciary. When a buyer offered cash for a square mile of unimproved forest, the trustees took the cash. The state had not asked them not to. The federal text had not asked them not to. Cooper v. Roberts would arrive, half a century later, to declare the trust character — but it would arrive too late to organize the trustees, and it could not, by itself, supply the discipline the architecture had assumed.

Third — and this is the deepest current — the 1802 Act establishes the compact form that every subsequent admission would follow. Ohio accepted the granting propositions; Congress accepted Ohio. Statehood and the school grant arrived on the same handshake, an irrevocable bargain between the new state and the United States. That structure carried doctrinal weight half a century later, when the Supreme Court in Cooper v. Roberts (1855) construed the same template language and held that the federal grant of section 16 created a trust on the public faith of the state — a “sacred obligation,” in the Court’s phrase, enforceable against state encroachment.⁴ Cooper arose on the Michigan record, not the Ohio record, but it was construing the architecture Ohio had inaugurated. The U.S. Supreme Court in Papasan v. Allain (1986) would later trace the entire national pattern back to “the practice begun with the admission of Ohio to the Union” — Ohio is the federal-grant prototype, not merely another section-16 state.⁵

I want to underline what Cooper did and what it could not do. Cooper supplied, judicially, the trust character that the federal text had not bothered to supply expressly. From 1855 forward, every section-16 state was on notice that its school lands were not state property in the ordinary sense. They were a trust. The trustee was the state. The beneficiary was the township and, beyond it, the children. As Margaret Bird put it on a Salt Lake City porch in 2025: this isn’t just government land; it’s trust land — and just like any other trust, the trustee (which is the state) has a fiduciary duty to act in the best interest of the beneficiary. Cooper is the case that locked that distinction into federal-public-lands doctrine.

What Cooper could not do was reach back through the half-century of disposals already on the record and undo them. The doctrine was prospective. The corpus, by 1855, was already substantially diminished in Ohio.

Ohio’s 1851 Constitution, in Article VI § 1, governs the residue. After a 1968 amendment, that section provides that the principal of school-purpose lands and funds “shall be used or disposed of in such manner as the General Assembly shall prescribe by law” — broad legislative discretion where later states would have inviolability.⁶ The Ohio Supreme Court has consistently treated section-16 lands as held in trust for the townships, but has also upheld the use of trust rentals for drainage and other “improvements” benefiting the lands themselves — see Trustees of Greene Township v. Campbell (1864) and the Upper Scioto Drainage Conservancy line (1932).⁷ Trustee status was real. Trustee discipline was elastic.

By 1900 Ohio had sold most of its 724,000 acres of section-16 grant lands. The proceeds, where they were preserved, were folded into a residual Common School Fund that today is a minor line item against the dominant general-fund and lottery-funded sources of Ohio K-12 financing. Ohio’s school lands are not where Ohio’s schools come from. The template’s first instantiation became the template’s first cautionary tale.

III. The pattern repeats, ten times

What happened in Ohio happened, with regional accents, in nine more states. I want to walk the pattern through each, because the regional accents are themselves part of the diagnosis. The same lean federal template produced different decades of failure in Free states and Slave states, in well-watered Mississippi-valley country and in the lower-rainfall plains, in centralized Michigan and in distributed Indiana. The template was the constant. The states were the variable. The corpus, in most of them, ended up in the same place.

Louisiana (1812) was the first state admitted from the Louisiana Purchase, and the first test of whether the section-16 template would carry into territory that had been French and Spanish before it was American. The 1812 Admission Act applied the template; the geography fought back. Across the lower Mississippi alluvial plain, large stretches of section 16 turned out to be already in private hands under confirmed French and Spanish titles. Louisiana spent decades selecting in-lieu lands — a different square mile, picked from the available federal domain, substituted for the section-16 the state could not have because somebody’s grandfather had a Spanish patent on it. The footprint of Louisiana’s school lands was therefore fragmented before management ever began. By the time the bookkeeping caught up, the disposition pattern was already underway. Louisiana’s modern Permanent School Fund is a constitutional artifact of modest dollar magnitude; the larger education-purpose vehicle in the state today, the Louisiana Education Quality Trust Fund, derives from a 1986 federal Outer Continental Shelf section 8(g) settlement memorialized in Article VII § 10.1 of the Louisiana Constitution — not from the 1812 grant.⁸ The OCS money is real, and large. But it is a twentieth-century federal-energy-revenue settlement wearing the constitutional clothing of a school trust. The 1812 corpus, in the place where it was supposed to be, is largely gone.

Indiana (1816) is the structural proof that strong state-constitutional text alone is not sufficient — and the cleanest counter-example to the assumption that the depletion pattern was about textual weakness. Indiana’s 1851 Constitution Article 8 § 3 contains language as strong as any irreducibility clause in any state constitution: “The principal of the Common School fund shall remain a perpetual fund, which may be increased, but shall never be diminished; and the income thereof shall be inviolably appropriated to the support of Common Schools, and to no other purpose whatever.”⁹ On paper, Indiana has near-Oregon-grade architecture. In practice, Indiana’s section-16 lands were largely sold off through the nineteenth century. The Common School Fund operates today predominantly as a loan portfolio to local school corporations rather than a productive corpus, and Indiana K-12 education is funded predominantly from non-trust sources. The Indiana Supreme Court has occasionally enforced Article 8 § 3 — notably in Horner v. Curry (2019) on civil forfeitures — and the Indiana Attorney General in 2003 (Official Opinion No. 2003-5) blocked a $25 million legislative sweep from Common School Fund principal back to the General Fund.¹⁰ But the larger story is that Indiana never built a constitutional ex-officio trustee board comparable to Oregon’s State Land Board; trust-fund administration is statutory, distributed across the State Board of Finance and the Treasurer of State.

I want to slow down at Indiana, because Indiana is the case that disposes of the textual-strength theory of trust durability all by itself. If words on the page were sufficient, Indiana’s corpus would still be intact. It is not. Strong text without a constitutionally-anchored trustee yields the same drift, only slower. The corpus was already small by the time the 2003 Attorney General opinion arrived to defend it. The discipline that protects a trust over two centuries is supplied by an institution and a constituency, not by a sentence — however strongly the sentence is drafted.

Illinois (1818) never developed comparable constitutional architecture and ran the depletion to its logical end. The most striking modern episode is the 1989 enactment of Public Act 86-970, codified in 105 ILCS 5/15-26, which authorized — and in many cases required — the liquidation and distribution of township permanent school funds. Where a township fund’s average annual income fell below $2,500 for three years, liquidation became automatic. Where a township petitioned, liquidation was permitted on streamlined procedure.¹¹ This was not drift. It was statutory dissolution of trust corpora that had survived 170 years of mismanagement. Illinois had a small remnant of the 1818 grant still on the books, and the General Assembly authorized its conversion to flow.

Mississippi and Alabama show the depletion pattern in its extreme antebellum-South form, and they show it on different mechanisms.

Mississippi’s 1.07 million acres of “Sixteenth Section” lands were never sold outright in bulk — the 1890 Constitution, Article 8 § 211, forbids sale of school trust lands and preserves the township-as-beneficiary structure, with administration distributed across roughly 140 individual school districts. But the 1944 amendment to Section 211 authorized leases of sixteenth-section lands within municipalities for terms of up to ninety-nine years for a gross sum, with provision for extension of existing municipal leases.¹² What followed in the town of Forest, Mississippi, became one of the era’s signature scandals: at least 441 ninety-nine-year leases were issued at nominal sums, including one representative lease for $7.50 — seven dollars and fifty cents — for ninety-nine years’ use of school trust land. The federal Papasan v. Allain litigation in the 1980s — the “Five Counties” case — exposed the related disparity for the Chickasaw Cession in northern Mississippi, where the original 1817 grant had failed because the Cession lands were not federally surveyed at admission, and where the substitute lieu lands had been lost during the Civil War. By 1984 the Cession districts were receiving $0.63 per pupil from school-trust appropriations against $75.34 per pupil in the rest of the state. The U.S. Supreme Court held the equal-protection claim could proceed past the Eleventh Amendment but ultimately rejected it on the merits.¹³ The corpus was not restored.

Alabama’s depletion ran through bank failure. After the 1827 Congressional act authorizing sales, Alabama sold large quantities of its sixteenth-section lands and deposited the proceeds in the State Bank of Alabama, which used them as lending capital. The bank failed in 1843. Later judicial findings in Knight v. Alabama (2004) report $1.3 million in sixteenth-section land proceeds and over $300,000 in university-land proceeds lost to the failure — a complete extinction of the corpus from a single banking collapse.¹⁴ What followed was a “certificate-and-interest system” — Alabama booked a notional school-fund principal and obligated itself to pay 6% interest in perpetuity, on a corpus that did not exist. The state is still paying interest on a fund the State Bank dissolved in 1843. Long v. Brown (Ala. 1843) treated the lands as held in trust; the trust had no money in it.¹⁵

Missouri (1821) and Arkansas (1836) received the same lean grant and disposed of the bulk of the corpus through township-level mechanisms over the nineteenth century. Missouri rebuilt its constitutional architecture across four constitutions (1820, 1865, 1875, 1945), but the corpus was already substantially gone before the 1945 Constitution arrived to protect it. Arkansas’s modern adequacy fight in Lake View School District No. 25 v. Huckabee (Ark. 2002) is litigated entirely outside the trust frame, on Article 14 § 1’s “general, suitable and efficient system” clause, because there is no meaningful trust corpus left to fight over.¹⁶

Florida (1845) received the same single-section grant on the same lean template and disposed of most of it in the same way. The modern Florida architecture is unusual in that the federal section-16 trust corpus and the State School Fund are minor structural pieces; the dominant designated education-revenue stream is the Lottery Education Fund, established by the 1986 amendment that added Article X § 15 to the Florida Constitution and operationalized when the lottery launched in 1988.¹⁷ The Lottery Education Fund is not a permanent corpus. It is annual designated revenue passing through the general fund. Florida looks, on a balance sheet, like it has a large education endowment. It does not. It has a stream of lottery dollars, vulnerable to redirection by the legislature in any given session, dressed in the constitutional clothing of a fund. The 1845 grant is a footnote. The old Article XII §§ 4–5 sacred-and-inviolate principal language survived in Hampton v. State Board of Education (1925), but the corpus it protected was already small.¹⁸

[BOXED-INSERT placeholder: chapter 3 — see Volume_I_Boxed_Inserts_2026-05-10.md]

IV. Michigan, 1837: the centralization pivot

The most consequential structural innovation in this period happens in Michigan, the twenty-sixth state, admitted January 26, 1837. Michigan inherits the Ohio template like everyone else — section 16 of every township, lean federal text, no express federal trust language, the standard package.¹⁹ But Michigan’s 1835 Constitution does something none of its predecessors did: it places school lands under a state-level Superintendent of Public Instruction, with operational administration through the State Treasurer. The grant still arrives “to the inhabitants of such township.” The administration of the grant moves to the state capitol.

This is the architectural pivot of the entire chapter, and I want to dwell on it. From Ohio in 1802 through Arkansas in 1836, school lands had been administered locally — by township trustees, county school commissioners, the people closest to the schoolhouse. Michigan in 1835 centralizes. The decision was not driven by abstract fiduciary theory; Michigan’s framers had watched what was happening in Ohio and Indiana and concluded that distributed administration was producing distributed depletion. Centralization was the proposed remedy. It was the model that Wisconsin would inherit in 1848, that Minnesota and Oregon would build on in 1858 and 1859, that the entire post-Civil-War West would adopt by default. Michigan also carried forward, verbatim, the Northwest Ordinance Article III moral-architecture clause into its Article VIII § 1 — “Religion, morality and knowledge being necessary to good government and the happiness of mankind, schools and the means of education shall forever be encouraged” — preserved across all four Michigan constitutions (1835, 1850, 1908, 1963).²⁰

It is therefore not a coincidence that Cooper v. Roberts (1855) — the foundational Supreme Court decision construing the section-16 grant as creating a trust on the public faith of the state — arose on the Michigan record. The doctrine emerged where the architecture had moved up a level. The Court held that a section-16 grant, once attached to surveyed land, gave rise to “trusts created by these compacts” carrying a “sacred obligation” enforceable against state encroachment. Cooper supplied, retroactively, the trust character that the federal text had never expressly named. The follow-on Supreme Court decision in Minnesota Co. v. National Co. (1865) reinforced Cooper as a rule of property for Michigan school-section lands.²¹ From 1855 forward, every section-16 state was a trustee whether its enabling act used the word or not.

But here is the surprise of the Michigan story, and it is the central surprise of this chapter: centralization did not by itself preserve the corpus. Michigan’s school lands were sold off through the nineteenth century at essentially the same pace, by essentially the same mechanisms, as Ohio’s and Indiana’s. The Superintendent of Public Instruction in Lansing was no more proof against political pressure to dispose of unimproved lands at suppressed prices than the township trustees in Cuyahoga County. And in 1875, the Michigan Legislature enacted Act 22, which the Michigan Supreme Court a century later in Milliken v. Green described as using educational-land sale proceeds to defray state-government expenses — expropriating the primary-school fund to the general fund.²² Attorney General v. Smith (1875) shows the Attorney General himself moving in state court to vacate a state land-office patent for unauthorized sale of primary-school lands in the Upper Peninsula mineral range.²³ Centralization improved discipline. It did not deliver fiduciary backbone.

I belabor this point because the Michigan lesson is exactly the lesson the post-1859 Western states would have to absorb the hard way: moving administration up the org chart is necessary, but it is not sufficient. What Michigan was missing — what every Northwest Ordinance state was missing — was the combination of three things: a state constitution with an irreducibility clause and an inviolable-appropriation clause; a fiduciary board defined in the constitution rather than created by statute; and an active beneficiary constituency that would notice when the legislature reached into the fund. Michigan had part of (1) in different forms across its four constitutions, did better than most on (2) by 1908, and never developed (3) at all. The 1963 Constitution simplified the structure further by omitting the older perpetual school-land fund clause in favor of the modern State School Aid Fund (Article IX § 11), which is funded by sales-tax dedication rather than by trust corpus.²⁴ The lands were sold. The fund stayed small. The schools, today, are funded by sales tax.

I want to be honest about what that means, because it is easy to slide into elegy here and the situation does not call for elegy. Michigan’s modern K-12 funding system works. Children get educated. The sales-tax dedication is real money. What Michigan no longer has is the corpus — the perpetual endowment the federal grant was designed to create — and what Michigan therefore no longer has is the protection that a corpus provides against the political weather of any given budget cycle. Sales-tax dollars can be redirected. A protected trust corpus, by definition, cannot. The corpus was the structural insulation. It was sold for the cash flow of the moment, and the cash flow of the moment is what Michigan’s schools now run on. That is a real loss, and it is the kind of loss the architecture of 1785 was designed to prevent.

V. Drift and directed seizure

I want to introduce a distinction now that will run through the rest of this volume, because without it the depletion record across these ten states becomes a single undifferentiated shrug of “things went badly,” and the diagnosis matters too much for a shrug.

The depletion pattern across these ten states was not a single mechanism. It was two mechanisms running in parallel, and the distinction matters because any account that names only one misses what was happening.

Drift was the slow attrition. Township trustees, with no fiduciary training and no professional staff, sold acreage at whatever the local market would bear, which in many places was less than the land was worth, and certainly less than the land would be worth twenty years later. Nineteenth-century land markets in newly settled territory ran at suppressed prices; section-16 sales went off at those prices, and the proceeds — when they were preserved at all, which was not always — were capitalized into funds that earned modest interest, were sometimes commingled with general district funds, and were subject to the cumulative erosions of inflation, mismanagement, and benign neglect across generations. No single bad actor caused this. The architecture was leaking, and the leak compounded.

I want the reader to register that drift is not a moral failure on the part of any individual trustee. It is a structural failure. A township trustee in 1830 selling a square mile of unimproved forest at the going rate is not, by his own lights, breaching a trust. He is doing his job by his neighbors’ standards. The fiduciary discipline that would have told him to wait twenty years for the rail line to arrive and triple the value of the land was not in his training, was not in his professional culture, and was not anywhere in the state’s institutional memory. Drift is what happens when an architecture assumes a standard of care that the surrounding institutions do not yet supply.

Directed seizure was the active component. It is concrete and dateable. Michigan, 1875: Act 22, transferring educational-land sale proceeds to general state expense.²² Mississippi, 1944: the Section 211 amendment authorizing ninety-nine-year municipal leases for a gross sum, followed by 441 nominal-rent leases in Forest alone, including the $7.50/99-year lease.¹² Alabama, 1828–1843: sixteenth-section sale proceeds deposited in the State Bank of Alabama, used as bank lending capital, lost in the 1843 bank failure — $1.3 million extinguished in a single banking collapse, with the state subsequently obligated to pay 6% interest on a notional principal that no longer existed.¹⁴ Illinois, 1989: Public Act 86-970, the statutory liquidation regime that authorized — and in low-income townships required — dissolution of township permanent school funds and distribution of the proceeds.¹¹ Indiana, 2003: House Enrolled Act 1001 § 115(b), directing a $25 million transfer from Common School Fund principal to the General Fund, blocked only because the Treasurer of State requested an Attorney General opinion and the Attorney General concluded it would violate Article 8 § 3.¹⁰ These are not failures of administration. These are statutory and executive acts that took trust corpus and put it somewhere else.

Both forces were operating. Both are still operating. The architecture wasn’t simply eroding; in places it was actively dismantled. The drift was real, the seizure was real, and the pattern across these ten states is the same: where the corpus was lost, it was lost partly to time and partly to deliberate transfer.

I belabor the drift-versus-seizure distinction because the policy responses are different. Drift is addressed by training, by professionalization, by stronger statutory standards of care, by audit cycles, by a constituency that pays attention. Seizure is addressed by constitutional discipline — irreducibility clauses that the legislature cannot evade, trustee boards that cannot be reorganized away, federal compact language that the state accepted as a condition of admission and cannot now repudiate, and a constituency that organizes against a directed seizure the moment it is proposed, not after it has cleared the second reading. Drift requires institution-building. Seizure requires constitutional armor. The Northwest Ordinance template attempted neither. That is the diagnostic point.

VI. The doctrinal floor: Cooper v. Roberts and what it could not do

Cooper v. Roberts (1855) is the case that makes this whole era doctrinally legible, and I want to put it on the page once more so the reader carries the language out of this chapter.

Justice John Campbell, writing for a unanimous Supreme Court, held that the section-16 grant in the Michigan admission act created a trust — that the State of Michigan held the lands “in the character of a trustee,” subject to a “sacred obligation” running to the township beneficiaries, on the basis of the “trusts created by these compacts” between Congress and the admitted state.⁴ The ruling supplied the trust character that the federal text had not. It applies, by its own logic, to every Northwest Ordinance template state. Papasan v. Allain (1986) confirmed the federal-prototype reading and traced the entire national pattern back to Ohio.⁵

What Cooper could not do was create a remedy strong enough to stop the depletion that was already underway. The case stands for the proposition that section-16 grants are trusts. It does not stand for any particular enforcement mechanism, any particular damages framework, any particular standing rule. Federal Attorney General enforcement was not provided in any of these states’ enabling acts. Express restoration mechanisms were absent. State courts were available, but the beneficiary constituency that would have brought such suits — children, future children, the diffuse public — was not organized to bring them and would not be for another century. Cooper declared the doctrine. The doctrine sat on a shelf while the lands were sold.

The modern restatement of the same doctrine, Lassen v. Arizona (1967), would arrive a hundred and twelve years later, on the much stronger 1910 New Mexico–Arizona Enabling Act.²⁵ By then the Northwest Ordinance corpus was already largely dispersed, and the Lassen strict-trust reasoning could be applied to it only by importing a textual rigor the original federal grants had never expressly carried. The modern enforcement record on Northwest Ordinance state grants — Ohio, Indiana, Illinois, Missouri, Arkansas — is thin, and the thinness is itself a finding. The trust was dissipated before any meaningful enforcement doctrine emerged at the state level. By the time the legal architecture was strong enough to defend the corpus, the corpus was small enough not to be worth the litigation.

I want the reader to register the asymmetry. A doctrine without a remedy is a doctrine the architecture cannot rely on. The Cooper trust character is not nothing — it shaped state-court jurisprudence, it constrained subsequent legislative behavior at the margins, it preserved at least the language of trusteeship through the second half of the nineteenth century. But a trustee who knows the doctrine and faces no enforceable remedy is a trustee operating on honor alone. Some trustees are honorable. Some are not. The architecture has to work for both.

VII. Where the modern fights live now

In most of these states, the modern school-funding fight has migrated from the trust frame to the adequacy frame. The dominant Ohio precedent is DeRolph v. State (1997, with continuing orders through 2002), holding the state’s K-12 funding system unconstitutional under Article VI § 2’s “thorough and efficient” clause — not a school-trust case, an adequacy case, fought over the totality of state K-12 funding rather than the federal-grant trust corpus.²⁶ Arkansas’s Lake View line (2002) does the same work under Article 14 § 1’s “general, suitable and efficient system” clause.¹⁶ Indiana’s Bonner v. Daniels (2009) forecloses adequacy litigation entirely in that state by holding the constitutional duty non-justiciable.²⁷ Florida’s modern litigation runs through Article IX § 1’s “uniform, efficient, safe, secure, and high quality” clause. Michigan’s Milliken v. Green and the post-Proposal-A Durant line address school-aid-fund disputes rather than land-grant trust corpus.²²

The migration is structurally telling, and I want to name what it tells us. Adequacy doctrine asks how much funding is enough, given the educational task, and asks the legislature to provide it from the general fund. Trust doctrine asks how much corpus has been preserved, given the original grant, and asks the trustee to account. The adequacy fight is the fight you have when the trust fight is no longer worth having, because the corpus is too small to argue over. In most Northwest Ordinance states, that is exactly where things stand. Mississippi and the antebellum-South sixteenth-section enforcement record (Papasan v. Allain; the Forest 99-year-lease litigation) is the partial exception — there, the trust frame survives because Mississippi’s 1890 no-sale rule preserved enough acreage to keep the fight alive on the original architecture.

That is one of the small mercies of the Mississippi case, and it is worth noticing. The state with the worst administrative record on antebellum-South sixteenth-section management is the state where the trust frame is still operative as a litigation tool, because the 1890 Constitution’s prohibition on sale preserved the land even where the income was thrown away in $7.50 leases. The acreage stayed in the trust. That is more than can be said for most of the rest of the cohort.

VIII. What the federal government learned, and what came next

The Northwest Ordinance template, by the time Florida came in in 1845, had been administered under for forty-two years across ten states. The pattern was visible. Land was being lost. Funds that should have been growing were not. The lean federal text was producing predictable depletion, regardless of regional context — the same mechanism in Ohio and Mississippi, in Michigan and Arkansas, in Free states and Slave states alike. This was not a Southern problem or a Northern problem. It was a template problem.

Congress was paying attention. When Wisconsin came in in 1848 and especially when Minnesota came in in 1858, the federal text started to thicken. Minnesota got the first doubled grant — sections 16 and 36 of every township, not just section 16, on the theory that more land at the front end would survive more depletion at the back end. Oregon, in 1859, got the doubled grant and a much stronger compact-form structure. By the time Utah was admitted in 1896, the federal text included permanent-fund language, in-trust references, and the foundations of fiduciary discipline that the Northwest Ordinance template had simply omitted. By 1910, the New Mexico–Arizona Enabling Act would deliver the high-water mark of federal trust language: an express trust, enumerated permissible uses, restoration mechanisms, federal AG enforcement, and breach-of-trust naming. The quadrupled grant — sections 2, 16, 32, and 36 — recognized that arid Western land was less valuable per acre and required more acreage per township to fund the same level of schooling.

Each later template was, in part, a federal response to the failures of the Northwest Ordinance template in practice. The federal government had watched its first attempt at a continental school endowment leak, and each subsequent admission act tried to plug a leak the previous template had not anticipated. By 1910 the template was nearly watertight. By 1910, in many cases, the original Northwest Ordinance corpora were already gone.

I want the reader to hold both halves of that sentence at once. The federal government got better at writing school-trust architecture across the nineteenth century. It got better because it had data. The data came from the ten states this chapter has just walked through. The architecture of 1910 was paid for, in a real sense, by the corpus of the ten Northwest Ordinance states whose disposal records taught Congress what the lean template could not do. That is a heavy price for a lesson, and the children whose schooling the lean template was supposed to fund were the ones who paid it.

IX. The chapter’s lesson

The Northwest Ordinance template was, on the morning it was adopted, the most ambitious universal-public-education infrastructure project in human history. A section of every township in every new state, reserved before settlement, dedicated to schools, in perpetuity. The architecture worked. The architecture was elegant. The architecture was insufficient — not because it was wrongly designed, but because design alone does not sustain a trust across two centuries against legislatures with shorter time horizons and stronger constituencies than the children the trust was for.

Three things were missing. The federal text did not expressly say “in trust” in the granting clause itself, did not name a restoration mechanism, did not authorize federal enforcement — even though the corresponding state-acceptance language in the Ohio compact (and its successors) used the in-trust formulation in the strongest terms then available: in trust for the use aforesaid, and for no other use, intent or purpose whatever. The state constitutions, where they spoke at all, vested broad legislative discretion over the corpus — Ohio is the cleanest example, with Article VI § 1 directing that the funds “shall be used or disposed of in such manner as the General Assembly shall prescribe by law.” Indiana is the cleanest counter-example: strong textual irreducibility (Article 8 § 3), but no constitutional trustee board to enforce it, and the corpus was largely sold before the modern enforcement opinions of 2003 and 2019 had anything substantial left to defend. And the beneficiary class — common school children — had no organized voice. The army had guns. The fund had only the words on the page.

These ten states are the cautionary opening of the larger story. The Northwest Ordinance template gave the United States its first national experiment in public-land school endowment, and the experiment produced enough information to design something stronger. The states admitted next would inherit the lesson. Minnesota would double the grant. Oregon would harden the constitutional architecture. Utah would add permanent-fund language. New Mexico and Arizona would write the express trust. The blueprint was older than the Constitution; the lessons were older than the modern remedy. Chapter 4 picks up where this one ends — with Wisconsin in 1848 and the first deliberate federal effort to make the template stronger.

Before we turn to it, one observation I want to leave the reader with. The trust character of these grants was real from the beginning, even where the federal text did not name it expressly. It was real in the Ohio compact’s for no other use, intent or purpose whatever. It was real in the Northwest Ordinance’s “schools and the means of education shall forever be encouraged.” It was real in the Cooper v. Roberts declaration of “trusts created by these compacts” and a “sacred obligation” running from state to township beneficiary. The trust was on the page, and on the bench, and in the architecture of every admission. What was missing was not the trust. What was missing was the constituency to hold the trustee to it. That constituency, where it existed at all in the ten Northwest Ordinance states, was small, intermittent, and undermatched by the legislatures and land offices that had reasons to dispose of the corpus and minutes to do it in.

The work of the next five chapters is to trace what happened when, in some of the states admitted later, that constituency was built — and what happened, in others, when it was not. The architecture matters. The constituency that defends the architecture matters more.

— Margaret Bird, Salt Lake City, 2026.

Chapter 3 Footnotes