This essay originally appeared in University of Michigan Journal of Law Reform, vol. 20, no. 2 (Winter 1987) pp. 409-490. Copyright © 1987, Peter Suber.

Population Changes and Constitutional Amendments:
Federalism Versus Democracy
Peter Suber, Philosophy Department, Earlham College

I. The Problem

Paul Bunyan once told his camp that if ten loggers shouting together could be heard ten times farther away than one logger shouting alone, then ten loggers listening together should be able to hear something ten times farther away than one logger listening alone. This is a political fable.

In general, when things are done by vote (even by an indirect vote of representatives), we believe that our voice is increased when people vote with us. Like Paul Bunyan, we believe that there is strength in numbers. A close look at the process by which we amend the federal constitution, however, reveals a counterexample that tries our faith. We find an exception where there should be none, unlike Paul Bunyan who had the sturdy faith to stand by his rule when there should be an exception.

To amend the federal Constitution, we need the assent of two-thirds of each house of Congress and three-fourths of the states.[Note 1] This Article focuses on the three-fourths requirement for the states. This threshold is particularly high, and it suggests that constitutional amendment is very difficult. In fact, amendment is difficult in different degrees for different constituencies, depending not on their numbers but on where they live.

The three-fourths requirement does not dictate the size of body that may amend the Constitution, only its distribution. If a majority of Americans is squeezed into fewer than three-fourths of the states, it will be powerless to amend the Constitution. In contrast, if a tiny minority of Americans is spread throughout more than three-fourths of the states, it will be able to adopt an amendment against the will of the remaining "supermajority."[Note 2]

Suppose that three-fourths of the population of the United States migrated to California, which then renamed itself "New California." Using the present methods of constitutional amendment, the other forty-nine states could adopt an amendment over New California's dissent, and could block or veto an amendment over New California's assent. New California's citizens could support their state's assent or dissent unanimously and intensely without increasing its voice in the slightest. New California would be just as powerless in the amending process and could just as easily be outvoted and vetoed if it contained 99.99% of the national population. The reason is simply that article V asks for the assent of three-fourths of the states without regard to population. Article V requires the assent of a supermajority of states, not that of a supermajority of people.[Note 3]

Except in details, the case of New California is not imaginary and the loss of voice for a majority of Americans is not hypothetical; mathematically and politically, a New California has existed in the United States since early in the nineteenth century.[Note 4]

In 1915, Joseph R. Long observed that the twelve most populous states (apparently under the 1910 census) comprised over half the national population.[Note 5] At that time, there were forty-eight states. Thirty-six states thus sufficed to ratify an amendment and, as Long observed, the least populous thirty-six states comprised a minority of the national population. Hence, in 1910, even the stringent three-fourths requirement would allow a minority of the nation to amend the Constitution. Fifty years later, Charles L. Black, Jr., noted that the federal Constitution could be amended by a forty percent minority of the American people (using 1960 census data).[Note 6]

Despite these tantalizing reports, no systematic inquiry has been made into the data from the twenty census decades since the first in 1790 to determine when this loss of majority rule first became possible, how often it has been possible, or how small a minority may amend the Constitution. In addition, no systematic inquiry has been made into the victims and beneficiaries of this otherwise innocent shift of the national population into a few large states. Although Long, Black, and others have drawn a picture of the shift of political power by geography, or state of residence, no one has drawn the picture of the shift of power by political interest, economic class, or race. I have looked at the past 200 years of census data in order to draw this picture. This Article reviews the disturbing results of these inquiries, which include the following:

  1. The loss of majority rule in the amending process first appeared in the census of 1820. It continued in 1830, disappeared in 1840, and has remained with us constantly since 1850. It has existed in sixteen of the twenty census decades since the adoption of the Constitution.

  2. The minority of the national population that can amend the Constitution has shrunk in every census for more than three-fourths of the nation's history. In 1820, when a minority first became able to amend the Constitution, it constituted 49.6% of the national population. Until the twentieth century, the figure stayed between 46% and 49%. but in the twentieth century, the figure dropped quickly; in 1970, it dwindled to 39.8%.

  3. The minority that can veto an amendment by refusing to ratify it has always been small and, recently, has shrunk to a mere sliver. The smallest minority with veto power is the population of the least populous one-fourth of the states, rounded upwards if there is a remainder, or plus one state if there is not.[Note 7] In 1790, states representing only 13.32% of the national population could veto an amendment. The percentage has not climbed so high since then, and has not even reached double digits since 1820. In 1980, the census closest to the expiration of voting on the Equal Rights Amendment (ERA), the percentage that could veto an amendment was mere 4.3% of the national population.

  4. Although a 4% minority can veto amendments if scattered throughout enough states, an absolute majority remains powerless to veto amendments if concentrated in too few (fewer than one-fourth) of the states. Presently, a veto of an amendment requires thirteen states. Under 1980 census data, the most populous thirteen states comprise an absolute majority of the population —as do the most populous 12, 11, 10, and 9 states. Although these states harbor a majority of Americans, they are powerless to adopt or even to veto amendments. In contrast, a differently distributed 39% of the nation can adopt amendments and a differently distributed 4% can veto them.

  5. Not only has majority rule vanished from the amending process, but the resulting disregard of the one-person, one-vote principle has made the impact fall unevenly on different groups. The minority that can adopt amendments is the population of the least populous three-fourths (thirty-eight) of the states. Neither the gainers of power who live in those states, nor the losers of power who live in the most populous one-fourth (thirteen) of the states, are representative of the Nation.[Note 8] Most importantly, disproportionate numbers of American blacks, hispanics, members of labor unions, recipients of Aid to Families with Dependent Children (AFDC), and persons with income below the poverty line remain concentrated in the states that have lost the most power in the amending process.

I conclude that the amending process under article V serves federalism more than it serves democracy. When a small number of states contains the bulk of the population, then the amendment process serves federalism at the expense of democracy. By "federalism" I mean the principle to treat the Nation as a federation, or as a collection of units (states) that might be of unequal size, rather than as a collection of people that might be distributed unevenly into units. I need only a minimal theory of democracy for the purposes of this Article, and it is the principle to treat people as equally significant voices in public affairs, and hence to weigh their voices equally in voting and other procedures of change.[Note 9]

As states of unequal size joined the Unions and the population shifted from the farmlands to the industrial areas, the United States population became concentrated in a comparatively few large states. The units of the federation became more and more imbalanced in population. A large state population is counted only once in a vote on an amendment proposal, and thus counts as much as the population of a much smaller state. As a result, the growing majority of citizens in the crowded states paid a price unforeseen by the framers of the Constitution:  shrinking power in the federal amending process. Citizens in the more sparsely populated states received the corresponding boon of increased power. This is undemocratic, even antidemocratic, although produced by innocent migration and expansion. The appendix documents the shifts in population that brought about this condition.

The Article evaluates the census data from the 200 years of our history under article V, interprets its political significance, and submits a brief for legal change. Part II reviews some political history at the time of the framing of article V, describes the federalist and republican principles underlying the dangers discerned in article V, and offers an overview of our demographic history as it affects the amending process. Part III examines whether an amendment has been adopted by states representing a minority of the national population, and discusses some of the methodological problems of interpreting the data. Part IV describes the discriminatory impact of the population shifts, and shows some of the groups that have been hurt most. In light of who has gained and lost power in the amending process, Part IV also estimates the prospects for various current amendment proposals. Part V discusses constitutional amendments that could remedy the problems disclosed in the Article and restore democracy to the amending process. The Article concludes that the loss of democracy and majority rule in the federal amending process should be remedied, and can be remedied, without jeopardizing federalist principles in other departments of our constitutional law and without departing from the intentions of the framers or current principles of fairness.

II. Background

A. Some Political History, Pre-1790

Article V embodies the federalist rationales of deference to state sovereignty and the institutionalization of state equality without regard to population. The framers clearly intended at least some of the resulting federalist incursions on democracy. The framers deliberately provided that the adoption of amendments requires the assent of the states as units, not the majority of popular votes. The strong federalist incentives at the time —the need to knit together a federation of sovereigns and to secure approval of the Constitution being drafted— may explain the framers' failure to take state population into account in the amendment process.

We do not know how many of the framers foresaw that the addition of new states and the internal migration of the national population could give the amending power to a minority of Americans. Alexander Hamilton, a delegate to the convention, foresaw the general problem, if not its application to the amending process, when, in Federalist No. 22, he wrote, "It may happen that this majority of States is a small minority of the people of America."[Note 10] Hamilton did not apparently raise this issue at the convention, and it was not debated.

The disregard of popular will in article V also supports a republican rationale. By calling for state assent rather than popular votes, article V requires that the people's representatives, not the people themselves, decide the fundamental questions raised by amendment proposals. A belief that the representatives would make wiser voices than the people themselves would not, of course, have been unique to the law of constitutional amendment.

In the drafting of article V, however, the framers evidently relied upon the federalist rationale more than the republican rationale. Article V provides two methods of ratifying amendments:  by state legislatures and by state conventions. The convention method bypasses the layer of representatives in an important way while preserving the federalist principle of disregarding population. State ratifying conventions are composed of delegates elected by the people. Although the delegates remain "representatives of the people," rather than "the people themselves," the election of delegates is a single issue election. The convention method allows the people to choose delegates for the single issue of ratification, instead of deferring to the legislators already sitting. The election results can therefore approach a popular referendum very closely. This focus of popular will represents the only significant difference between the legislative and convention methods of ratification.

The convention method therefore provides a more democratic than republican method of ratifying amendments to the Constitution. Both methods of ratifying amendments —by conventions and by legislatures— conform to the federalist principle because both provide that the assent of three-fourths of the states, regardless of population, suffices to ratify an amendment. We may assume, then, that the framers did not rest the provisions of article V on principles of republican government so much as on principles of federalism.[Note 11] We cannot assume, however, that they intended or foresaw all of the consequences of adhering to the federalist principle throughout radical changes in the size of the federation and the distribution of the population within it.

B. Federalist and Republican Principles

Let us say that a state legislature or convention "perfectly" represents its citizens only when a vote of ratification reflects unanimous popular assent to ratification and a vote against ratification reflects unanimous popular dissent. Needless to say, representation does not occur that perfectly, nor, under most republican and even democratic theories, need it be perfect. In its extremity, however, perfect representation is a useful idea. If the conditions of perfect representation exist and a minority can still amend the Constitution, then we know that the problem does not stem from the betrayal of popular will by elected officials. And in fact this is so. The concentration of the national population over the last 200 years has become so severe that a minority can amend the Constitution even if each state legislature represents its citizens "perfectly."

Let us call the adoption of a constitutional amendment by states representing less than half the national population "minority amendment." It follows that minority amendments are adopted over the dissent (or indifference) of a majority. At the federal level, minority amendment can occur in only two ways:  (1) the state legislatures or conventions represent their citizens very imperfectly, or (2) the least populous three-fourths of the states comprise less than half the national population. The first may be called the republican type of minority amendment, because it is made possible by the republican principle to let representatives speak for the people even when representation may be imperfect. The second type of minority amendment may be called the federalist type, because it is made possible by the federalist principle to seek state assent without regard to population.

Proponents of the Equal Rights Amendment may suspect that the amendment expired unratified because the legislatures in a few key states did not fairly represent their citizens. The framers designed the convention method of ratifying amendments to satisfy doubts of that kind, but this method has only been used once —for the twenty-first amendment (the repeal of prohibition). For other amendments, such as prohibition itself, doubts may linger on that the state legislators exceeded their popular mandate in voting to ratify. For defeated amendment proposals, such as the ERA, doubts may linger on that state legislators violated popular will in voting not to ratify.

I mention the problem of imperfect representation here only to emphasize that the federalist type of minority amendment constitutes a very different problem from the republican type. The republican form of minority amendment represents a risk inherent in any republican form of government in which representatives may speak for the people and may therefore speak inaccurately or against their interests. The framers clearly chose this risk, and it seems tolerable to most Americans.[Note 12]

The federalist form of minority amendment remains distinct and independent from the republican form. Neither provides the basis for the other. The federalist type may thus be remedied without challenging the republican form of government. Because even perfect representation in the state legislatures would not cure the problem of New California, the federalism of article V constitutes the more fundamental flaw. It now provides that three-fourths of the national population may be powerless to adopt amendments and that even a majority may be powerless to veto amendments. It does so either by insufficient foresight or by eighteenth century incentives to make a federation out of the several states. While federalism is far from obsolete, the grounds of many of its application, including those of article V, were exhausted of urgency 200 years ago. But whether the problem arose from lack of foresight or federalist needs long since met, the problem may be remedied today without sacrificing the republican form of government or any contemporary applications of the federalist principle outside the process of constitutional amendment.[Note 13]

C. Some Demographic History, 1790-1980

When the Constitution was written, debated, and adopted, the requirement for three-fourths of the states invariably translated into a requirement for a significant supermajority of the population.[Note 14] Thirty years and nine new states after the adoption of the Constitution, the assent of three-fourths of the states could be gathered without mustering even a simple majority of the national population. Today, the twenty most populous American states form a "New California" containing three-fourths of the national population. Yet we still operate under an amendment clause written when three-fourths of the states represented an absolute majority of the national population, and when the population disparity between the largest and smallest states was significantly smaller.

In the first census in 1790, three years after article V was written, the Nation's most populous state (Virginia) contained nearly twelve times the population of the least populous state (Delaware). In the most recent census (1980), the most populous state (California) has nearly sixty times the population of the least populous state (Alaska). The disparity between the most and least populous states has increased fivefold. In 1790, the five most populous states had slightly more than three times the population of the five least populous states; in 1980, this disparity has increased tenfold as the five most populous states contain nearly thirty times the population of the five least populous states.

In 1790, 31% (four states) represented the fewest states needed to comprise half the national population, and 54% (seven states) represented the fewest needed to comprise three-fourths of the population. In 1980, these "ruling" sets have shrunk considerably:  half the population now lives in 18% of the states (nine states) and three-fourths of the nation lives in only 40% of the states (twenty states).

These figures show the contours of a major change in population density. The national population has become much less evenly distributed among the states and much more concentrated in a few very large states. To count state votes regardless of population now greatly dilutes the voice of most Americans in the sovereign power of the American legal system.

Perhaps it is well at this point to name names. The special 40% minority that can adopt an amendment will overrule the 60% squeezed into the most populous twelve states:  California, New York, Texas, Pennsylvania, Illinois, Ohio, Florida, Michigan, New Jersey, North Carolina, Massachusetts, and Indiana.[Note 15] The special 4% minority that can veto an amendment, against the wishes of the 96% concentrated in the most populous three-fourths of the states, is composed of the citizens of Maine, Hawaii, Rhode Island, Idaho, New Hampshire, Nevada, Montana, South Dakota, North Dakota, Delaware, Vermont, Wyoming, and Alaska.[Note 16]

How many Californians does it take to make one Alaskan? Answer:  for power in the federal amending process, almost 60. The concentration of the population that has occurred over the past 200 years has injured the citizens of the most populous states the most severely. To measure the extent of the injury, we can set the voice of a citizen of California —the most populous state— at one unit, and ask how many units of voice are possessed by citizens of the least populous state, Alaska.[Note 17]

The following table[Note 18] summarizes the differentiation of amending power by state:

Rank in
Population (1980)
State Units of Voice
California 1.00
Illinois 2.07
North Carolina 4.00
Missouri 4.82
Washington 5.74
Connecticut 7.63
Oregon 9.00
Nebraska 15.09
Rhode Island 24.99
South Dakota 34.34
Alaska 59.24

These figures reveal the results of an amendment clause that weighs state assent to the exclusion of popular assent. The federalist rationale of the clause, which values the voices of states more than the voices of their citizens, was strong in 1787 and may still be strong today. Yet, in 1787, the federalist principle was not antidemocratic in its effects. The states have always been of unequal size, giving the citizens of the smaller states greater voice in the amending process than citizens of the larger states. Courts have never applied the one-person, one-vote principle to article V.[Note 19] In 1787, however, and up to the census of 1820, the differently weighted voices in the federation did not destroy majority rule. The addition of states of unequal size has enlarged the number of small states against the larger, and the internal migration of the population to cities has increased the disparity between the rural areas and the urban population centers. Hence, in 1820, and from 1850 onwards, the federalist principle in the amending clause has allowed a few citizens in many states to overrule the many citizens in a few states.

As these demographic changes accumulated, the gradual and quantitative shifts of power became qualitative changes. The voice of a certain minority has grown by degrees, and the voice of a certain majority has dwindled. Two important thresholds were surpassed sometime between 1820 and 1850. First, a rural minority became able to adopt an amendment on its own, destroying majority rule in the amendment process to this day. Second, the urban majority lost the power even to veto amendments if it had to face the minimum sufficient rural minority (the least populous three-fourths of the states). The combined effect of these demographic changes has done more than merely give certain minorities power over certain majorities; it has given them virtual sovereignty —the federal amending power represents the supreme power in the American legal system, able to change any law and limit every power in the system. Hence, even those with the strongest loyalty to the federalist principle must question its continuing wisdom in a way that the framers did not have to question its original wisdom. Should our amending clause continue to put state assent ahead of popular assent now that strongly antidemocratic results are possible?

These demographic changes have produced two antidemocratic effects, not foreseen 200 years ago, that need remedies today. The first is the loss of majority rule in the amending process. The remedy may, but need not, incorporate the one-person, one-vote principle. The second is the discriminatory impact on racial minorities and the poor. Americans have not lost their voices in the amending process in equal degrees. Neither the winners nor the losers in this shift of power represent a cross-section of Americans. The citizens of the sparsely populated states have gained voice, while those in the heavily populated states have lost. The groups concentrated in the heavily populated states who have lost the most amending power include blacks, hispanics, members of labor unions, recipients of AFDC, and persons with income below the poverty level.

Hence, the minority that has been given the federal amending power at the expense of the majority is not one of the "discrete and insular" minorities that we normally want to protect from majoritarian oppression.[Note 20] On the contrary, the minority that has gained power was already in power in the legislatures of the sparsely populated states, and the majority that has lost power includes disproportionate fractions of the Nation's discrete and insular minorities, disadvantaged, and powerless. The winning minority may not be wealthy or well-educated, but it is far from deserving special constitutional solicitude. The shrinking minority that can amend, then, and the shrinking superminority that can veto, entrench an already powerful class and aggravate the estrangement from power of other minorities.

III. To What Extent Have the Possible Dangers Become Actual?

The data show so far only a mathematical possibility that a minority of Americans may adopt a constitutional amendment and that a very tiny minority may veto one by refusing to ratify it. One may well ask whether either of these antidemocratic scenarios has occurred. One may ask about more subtle harms beyond these dramatic possibilities, such as the loss of majority rule and the rejection of the one-person, one-vote- principle. Are there any existing harms to which we may point, or are we left to reckon the danger of a mathematical possibility? Throughout this section I will focus on the possibilities of federalist, not republican, minority amendment.

The first danger is that of minority amendment itself. So far no amendment has been adopted by states representing less than half the national population. The eighteenth amendment (prohibition) presented the closest case. At the time of its ratification in 1919, a constitutional amendment required thirty-six states for adoption; the first thirty-six to ratify the eighteenth amendment comprised 62.51% of the national population.[Note 21] Forty-five states, however, comprising 90.48% of the national population, eventually ratified the eighteenth amendment.[Note 22] The twenty-fifth amendment (presidential disability and succession) presented the next closest case. By 1967, when this amendment was ratified, the nation had grown to its present size. Hence, an amendment required thirty-eight states for ratification (three-fourths of fifty, rounded upwards); the first thirty-eight states to ratify the twenty-fifth amendment comprised 70.33% of the national population.[Note 23] The eighteenth and twenty-fifth amendments marked the only amendments whose "first three-fourths" comprised less than 75% of the population. The average for all amendments adopted after the Bill or Rights is 79.56% (first three-fourths) or 89.12% (all states eventually ratifying).

The federalist type of minority amendment has not occurred. Every adopted amendment has been ratified by states containing more than half of the national population, and most have been ratified by supermajorities exceeding 75% of the national population. Whether we look at the "first three-fourths" of the states to ratify, or especially if we look at all the states eventually ratifying, we must conclude that the adopted amendments have been very popular with the people or at least with their representatives.

Let us call the veto of an amendment by states representing less than 25% of the national population a "superminority veto." The same demographic conditions that make minority amendment possible make superminority veto possible. One-fourth of the states plus one may always veto an amendment by refusing to ratify it. When a majority of the national population is concentrated in fewer than one-fourth of the states, then the remaining three-fourths may adopt an amendment against the wishes of a majority of the people (federalist minority amendment), and the least populous one-fourth may veto an amendment against the wishes of very large supermajorities of the national population (federalist superminority veto).

Although we know that federalist minority amendment has not occurred for any ratified amendment, it is more difficult to say whether federalist superminority veto has occurred for any defeated amendment. We eliminate at the start the amendment proposals defeated in Congress and never sent to the states. Only six proposals have passed both houses of Congress and failed to attain the assent of three-fourths of the states.[Note 24] When a defeated proposal contained no fixed expiration date —which was the case for all proposals prior to the eighteenth amendment— then it is impossible to say which states "vetoed" them unless nearly every state voted one way or another. Otherwise, arguably, they are still pending, awaiting the requisite votes of ratification.[Note 25]

Despite these difficulties, it seems clear that none of the six proposals passed by Congress and rejected by the states was defeated by a superminority veto. The proposals not known to have enough rejecters to escape superminority veto are known to have too few ratifiers.[Note 26] Even if any of the latter proposals are "still pending," their failure to become law so far is not the work of superminority veto.

We do have good data in the case of ERA.[Note 27] The states that never ratified the ERA comprised either 29.87% or 27.95% of the national population, depending upon whether we use 1980 or 1970 data. If we count rescissions of ratification as effective, these figures are even higher.

These figures for the ERA clearly reveal that the federalist type of minority amendment, or veto in this case, did not cause the amendment's defeat:  the states refusing to ratify the ERA contained over one-fourth of the Nation's population. Yet the republican type of minority amendment or veto may well have undermined the ERA's success.[Note 28] We do not know precisely how many citizens across the country would have voted for the ERA in a popular referendum. If these citizens ever exceeded 75% of the national population, then only imperfect representation can account for the demise of the ERA.[Note 29] The set of ratifying states could have grown, however, to contain 75% of Americans, without reaching the threshold of thirty-eight states needed to ratify the amendment. For example, as the most populous of the nonratifying states, Illinois could have cast a ratifying vote and easily put the percentage of citizens in ratifying states over 75% without meeting the three-fourths requirement for state votes. If this had happened, then the federalist type of superminority veto would have appeared in addition to any possible republican superminority veto or very imperfect representation.

If one is prone to fear that mathematically possible dangers may become actual, then one has more to fear from the republican type of minority amendment than from the federalist type. The number of members of Congress and state legislatures needed to propose and ratify a federal amendment is exceedingly small as a percentage of the national population. Suppose it takes 2000 today; that is less than one-thousandth of one percent of the national population. In principle this group could adopt or prevent amendments against the will of every other single American. Past critics of the fairness of article V have focused on this potential,[Note 30] which in effect is an indictment of representative forms of democracy. Whether or not one is tempted to join this critique, one should note that the republican type of abuse is separate from the federalist type, not the basis for it. Perfect representation in state legislatures would leave the potential for federalist minority amendment intact.

How realistic is it to fear the federalist type of minority amendment? The census data of the last 200 years prove that minority amendment and superminority veto are living opportunities for the appropriate populations to seize. But should we fear that this will happen, or simply observe that it is possible?

Although states representing a minority of the national population have not yet adopted any amendment, the antidemocratic possibilities of minority amendments and vetoes are not merely speculative. Rule by a minority, and veto by an extraordinary minority, are mathematically possible today, and have been for 130 continuous years. In this they resemble the dangers of runaway legislatures, conventions, and electoral colleges; grossly imperfect representation is mathematically possible and has been for our entire history under the Constitution. There are many informal political checks that prevent legislators from defying popular will in large numbers on important issues, especially when that popular will is unified, intense, and discernible. Although in principle these checks may fail, their gross failure is rarely a realistic possibility. Hence, critics who focus on the danger of mass legislative betrayal of popular will, or the republican type of minority amendment, are easily cast as cranks, frightened of a bare possibility and willing to sacrifice republican democracy to meet it.

Without joining these suspicions, or answering them, I simply observe that the possibility of the federalist type of minority amendment and veto vastly differs from the possibility of imperfect representation. Not only do the two types of abuse represent independent dangers, they also present actual (as opposed to potential) harms in different degrees. The concentrating American population has already placed the federal amending power in the hands of a shifting and shrinking minority. That minority is not subject to the formal and informal checks that restrain state legislators —preeminently the fear of defeat, hope of re-election, hope of future cooperation from other legislators, and a sense of professionalism.

Another difference between the republican and federalist dangers, more important to lawyers than to other citizens, is that the framers evidently foresaw and intentionally left us at the risk of the former, but almost certainly did not even foresee the latter. Hence, even if the two dangers were equally likely to materialize, action to prevent the federalist danger would be far more consonant with our constitutional design than action to prevent the possibility of imperfect representation.

For the republican type of minority amendment to succeed, representation must be so imperfect that most citizens would judge it unfair. In contrast, with the federalist type of minority amendment, the minority that holds the power to overrule the majority does not even putatively represent the majority. Neither a sense of fairness and obligation nor a sense of accountability, political prudence, or Machiavellian calculation present a deterrent to abuse here. When we recall that the federalist type of minority amendment and veto may exist even under the assumption of perfect representation in each state legislature, then we see very clearly that the potential for antidemocratic results does not depend upon whether the maldistributed amending power will be abused or used unfairly, but whether it will be used at all by certain groups.

In short, there are two reasons why the possibility of the federalist type of minority amendment and veto is already an actual harm different from the possibility of imperfect representation. First, the power of minority amendment and veto is already vested in the hands that can use it, and its use does not depend upon ill-will, bad faith, usurpation, antidemocratic sentiments, or the failure to informal but reliable checks. Furthermore, its use is entirely compatible with the strict conditions of perfect representation. Although public opinion, the electoral process, and some sense of duty in many representatives restrain republican abuses, nothing comparable exists to restrict the federalist abuse. Not only can the smaller states outvote the bulk of the population without misrepresenting their own citizens, but they may often find it in their own interests to do so.[Note 31]

Second, and most importantly, the federalist danger has materialized in different degrees already. First, although no amendment has been adopted by states comprising less than one-half of the national population, states comprising less than three-fourths of the national population have adopted two amendments. Second, the departure from the one-person, one-vote principle has never been a mere mathematical possibility. The principle has been violated in the ratification of every amendment in American history. Moreover, the nonuniform dilution of voice implied by the violation of the one-person, one-vote principle is already actual. For more than a century, the citizens of the heavily populated states have had less weight in the amendment process than other Americans. This violation of the one-person, one-vote principle disproportionately weakens the voice of city dwellers, racial minorities, labor union members, and the poor. Finally, we have long since lost supermajority rule, and even majority rule, in the federal amending process.

Article V contains the potential for federalist and republican violations of popular will. The republican type may have occurred often but in small degrees compatible with the constitutional design intended by the framers and, to a large extent, accepted by the people of the United States. The federalist type, however, has already generated some antidemocratic effects. Its potential to create more need not be suffered as the price of representative democracy or the intent of the framers. Those of us who would like a remedy do not confuse the federalist danger with legislative betrayal of popular will and do not fear a bare mathematical possibility. We regret the existing loss of majority rule in the amending process, and the violation of the one-person, one-vote principle, and we fear the chance that a superminority veto may block an amendment favored by a national supermajority, and the chance that an honest minority of Americans accidentally clustered in sparsely populated states will use the power they have inherited to amend the Constitution against the will of a national majority.

Methodological Demurrers

Two methodological problems complicate the task of interpreting the data in the appendix to determine whether potential harms are actual harms. First, the data in table 1 only show a potential for minority amendment under certain assumptions. The tidiest, but least plausible, assumption is that state legislatures (or state conventions) represent their citizens perfectly. If the least populous three-fourths of the states adopted an amendment, then we would not know, without more, whether a minority or a majority of Americans favored the amendment. The adopting states would contain a minority of Americans but, of course, the amendment could be favored by citizens in dissenting states and disfavored by citizens in assenting states.

Adoption by the least populous three-fourths of the states would constitute "minority amendment" if each state legislature perfectly represented its citizens, and under a few other more realistic assumptions. Representation, of course, does not occur perfectly. Assenting states contain dissenters and dissenting states contain assenters. I need not (and do not) argue that, across the nation, these unrepresented preferences cancel each other out, as Livingston does.[Note 32] For if the unrepresented dissenters vastly outnumbered the unrepresented assenters, then the potential for minority amendment would be stronger than ever. The ratifying state legislatures could represent small populations to begin with, and disregard local majority sentiment against an amendment. Moreover, the unrepresented assenting citizens could outnumber the unrepresented dissenters nationwide without eliminating the potential for minority amendment. If the least populous three-fourths of the states ratified an amendment, then it would be mathematically possible for the outvoted large dissenting states to have more unrepresented assenters than the victorious small states have unrepresented dissenters.

Second, accuracy of census data raises an additional methodological problem. By comparing the data from different census decades, I am not assuming that each census is equally reliable. I am assuming, however, that each is roughly as reliable for its large states as for its small states.

IV. The Discriminatory Impact and Prospects For Future Amendments

Courts have never applied the one-person, one-vote principle to the federal amendment process. Two centuries of defiance of that principle under article V have allowed innocent shifts in the national population to weaken the voice of citizens in the heavily populated states. A citizen of New York or California has far less weight in the amending process than a citizen of Wyoming or Alaska. Consequently, as a general rule, the federalist type of minority amendment enhances the chances of amendments favored by the citizens of lightly populated states, and diminishes the chances of amendments favored by citizens of heavily populated states.

On many issues, citizens of lightly and heavily populated states have the same interests, or have no competing interests. Yet, on other issues, their interests divide them. Neither group represents a national consensus on all important issues of public policy. Of the groups that have lost power in the amending process, some have lost amending power overall, some have lost it less, and others more, than a fictitious "average American," and, paradoxically, some still retain control in the states that have gained power. Although difficult, it is possible to pick among the losing constituencies and to determine which groups have suffered a disproportionate loss of voice.

Two kinds of groups have lost voice in the amending process because of the concentration of the population. To articulate this distinction more clearly, let us call the states that have lost voice in the amending process the "losing states" and the rest the "winning states." The winners are the least populous three-fourths of the states. The losers are all the others —the most populous one-fourth. Now let us posit two groups of Americans, each having members in both the winning and losing states:  the "Evens" and the "Unevens." Suppose the Evens and Unevens have exactly the same number of members nationwide. The Evens dwell in the winning and losing states in the same proportion as Americans in general. If the losing states contain 60% of Americans, then they contain 60% of the Evens as well. The Unevens have concentrated in the losing states. If the losing states contain 60% of Americans and Evens, they might contain 70% or 80% of the Unevens. For example, women obviously represent a group like the Evens and big-city dwellers form a group like the Unevens.

The Evens and Unevens have both become net losers of voices in the amending process. Although each loses, as a class, because most of its members dwell in the losing states, the Unevens lose out of proportion to their numbers because a greater percentage lose voice than do Americans in general. The Evens, however, do not lose voice out of proportion to their numbers because the same percentage of Evens gain and lose voice as do Americans in general. Collectively, each group ends up a loser; distributively, a greater number and, therefore, a greater percentage of Unevens lose voice. As national constituencies, the Unevens as a class will have lost power relative to the Evens. Without the differential harm suffered by groups like the Unevens, we could speak only of detrimental impact or loss of voice. But because the detriment falls disproportionately on different groups, we may speak of discriminatory impact.

Now that we possess the distinction between the Evens and the Unevens, we observe that almost all Americans have emerged as losers of voice in the amending process, but that not all are uneven losers. Whites and blacks, men and women, rich and poor, employed and unemployed, as classes, have all lost voice in the amending process. Indeed, as a class, Americans have (evenly) lost voice in the amending process because more Americas live in the losing states than in the winning states. Table 3 of the appendix allows us to determine that the uneven losers include blacks, hispanics, members of labor unions, recipients of AFDC, and persons with income below the poverty line.[Note 33]

The worsening federalist danger possible under article V hinders the adoption of those amendments favored by the uneven losers. The resubmitted ERA and the proposal to give the District of Columbia most of the political rights of a state represent two such amendments. Any proposal (such as I will describe in Part V) to change the amendment process to correct this problem may present a third amendment of this type.

By contrast, citizens of the lightly populated states are disproportionately rural, western, white, and middle class (or nonindigent and nonunionist). Amendments favored by these constituencies are more likely to be adopted as the concentration of the population increases or as the federalist danger continues and worsens. Possible amendments of this type include proposals to require a balanced federal budget, to permit prayer in public schools, and to ban most abortions.

By "enhancing" and "diminishing" the chances of amendments, I mean the chances of an amendment's adoption today as compared to the chances of the same amendments (1) under a democratic rather than federalist procedure, and (2) under the demographic patterns of the past, rather than the present. Hence, if we only consider the effects of the increasing concentration of the population, then a resubmitted ERA is less likely to be adopted under the present amending procedure than under a democratic procedure, and less likely to be adopted today than in the past. Similarly, the proposal to ban most abortions is more likely to be adopted under the present methods than under democratic methods, and more likely to be adopted today than in the past.

Throughout American history, the density of the population in the rural areas has tended to decrease and the population of the urban and industrial areas has tended to increase. Under article V, that trend simply adds weight to the preferences of the rural citizens who stay in the country and subtracts weight from the preferences of city dwellers, further intensifying the discriminatory impact on the racial minorities and others unevenly concentrated in the large urban states. The trend may, however, be reversing. The 1980 census shows a decrease of population in the heavily populated states and an increase in certain lightly populated states, thus documenting the migration from the cities to the "sunbelt." Undoubtedly, such reverse shifts of population enhance the democracy of the present amending procedure but, so far, can only be considered one step toward democracy after 130 years of backward steps.

V. Remedies

Not all serious risks of abuse and even actual harms should be remedied, arguably, if they are inseparable from a constitutional design with overriding virtues. That is not the case here. The federalist type of minority amendment and veto represents an anomaly under the Constitution, not an inevitable or desirable byproduct of the framers' designs. Tolerating its risks is not a part of ordinary political trust of representatives or loyalty to the Constitution. The federalist type of minority amendment should be feared as a present and inadvertent vesting of supreme power in a shrinking and unrepresentative minority of Americans that has no reason to restrain itself, and some reason to use, its gift of power. For us, unlike the framers, the problem is not to foresee and avoid a danger, but to recognize and remedy an abuse that has already crept up on us.

How can the dangers latent in article V be remedied? The republican type of abuse could be remedied by eliminating the layer of representatives, or by specially electing representatives for the ratification vote and binding them to the publicly announced positions on which they campaigned. A popular referendum and the present convention method of ratification (modified to bind delegates to campaign positions) would each cure the republican type of abuse. The latter would eliminate the republican danger while preserving the federalist principle and its modern dangers.

The federalist type of abuse could be remedied only by taking population into account in the ratification process. This remedy would require repeal of the present methods of ratification, including the convention method. A popular referendum that transcended state boundaries, and a plan to weigh state votes by population, would each provide a remedy to the federalist type of abuse. Weighing state votes by population would eliminate the federalist danger but would preserve the republican principle and its dangers. Only a referendum would eliminate the federalist and republican dangers at once.

Americans should realize that if the past population shifts cannot be reversed, then a remedy must be sought in an amendment to article V.[Note 34] Naturally, any new method of amendment would have to be adopted under the present method, just as only whites could vote to give blacks the right to vote under the fifteenth amendment and only men could vote to give women the right to vote under the nineteenth amendment.[Note 35]

The referendum method of ratification presents a compelling solution to the problem. Without undue complexity, and in one stroke, it would eliminate both the federalist and republican risks to democratic process. A referendum would bypass representatives who might depart from popular will and surpass state boundaries that dilute the strength of most citizens' voices.

A few words on the referendum method of ratification may set the stage for informed debate. First, to prevent the republican and federalist dangers, the referendum method must become the exclusive, not supplementary, method of ratification. Or, at least, the present legislative and convention methods must be repealed. As long as the legislative method of ratification offers another option, both types of abuse will remain possible, and, as long as the convention method continues, the federalist type of abuse will remain possible.

Second, proposals to make a referendum exclusive are not new and have historically been offered by those wishing to make amendment even more difficult than it already is.[Note 36] Whether a referendum would make amendment more difficult is an empirical question on which the experience of several states will be relevant. Whether the amending process should be made easier or more difficult is a political question that should be answered anew by each generation. We should realize that even by leaving article V unchanged, we consent to changes, wrought by expansion and migration, that make amendment more difficult for certain majorities and less difficult for certain minorities.

If, in our best judgment, we conclude that an ordinary referendum would increase the difficulty, we could still use a modified referendum. For example, a referendum could require (say) 67% assent of those eligible to vote or 75% assent of those actually voting. Such a disjunctive requirement would lower the hurdle to amendment without diminishing minority rights.

Third, a properly designed referendum need not threaten the benefits of the federalist principle that still exist in the late twentieth century. Any amendment to article V that merely made a referendum exclusive would not affect federalism in the form of local control and experimentation. To prevent regional imbalance or the rule of the Nation by New York, the Midwestern industrial states, Texas, and California, a referendum could call for (say) 75% assent nationwide and 51% in each of three fourths of the states. Such a double or conjunctive requirement would not replicate the federalist type of minority amendment. Rather, it would give the numerous but thinly populated states a veto, but neither a superminority veto nor the power to impose an amendment on the majority.

A referendum thus represents one way, but not the only way, to restore democracy to the amending process. A supermajority of 75% of the national population could almost always have its way under such a democratic scheme. Under the present federalist and republican scheme, this supermajority could only have its way if it is both (1) widely distributed among the states and (2) fairly represented in the state legislatures. The second condition, changes unpredictably, and we tolerate a great deal of imperfect representation in the name of republicanism. The first condition, however, does not exist and is not likely to exist in the near future. Moreover, the dilution of popular sovereignty that results is not justified by eighteenth century federalist principles or by twentieth century notions of democracy.


The addition of states of unequal size to the Union, and the migration of the population into urban areas, has in effect amended article V of the Constitution. As a result of these demographic changes, the amending process under article V now permits a minority of the population (about 40%) to adopt amendments, and a tiny minority (about 4%) to veto amendments. Thus, a sufficient number of sparsely populated states can override the intense preference of up to 96% of American citizens. The principle that makes this possible is not protection of minorities but federalism. It is not the federalism of local control and experimentation, but the federalism of state sovereignty and state ingratiation needed to get the original Constitution adopted. Yet by treating all states equally in the amending process, article V now treats their citizens unequally. It allows the numerous but thinly populated states to outvote large supermajorities of the population. This result could not have occurred in the generation of the framers. Such a result became possible only as the number of states increased and as the population became concentrated in a few large states.

Delegates to the constitutional convention almost certainly did not foresee or intend the antidemocratic effects of the present amendment process, although Alexander Hamilton envisioned the general possibility. The delegates did not raise the issue at the convention, even though the result primarily injures the larger states, in which most of the framers resided. The present amendment process subverts the ideals of popular sovereignty, supermajority rule, and even majority rule. The framers provided for the addition of new states, but not for the day when such additions, and state rule itself at the level of federal amendments, would injure the rights and even compromise the sovereignty of the people.

Even if the federalist type of minority amendment was foreseen by the framers and justified by the strong desire to knit together a federation, or to adopt the Constitution, those particular federalist designs are achieved today. The states have formed a federation that is in no danger of dissolving if article V abandons its federalist principle for a democratic one. To maintain the federalism that holds the Nation together, then, we need not maintain the federalist principle of article V that penalizes migration into population centers, lessens the voice of urban and industrial American communities, blacks, hispanics, labor unionists, women heads of household, and the poor, and turns over the supreme power in the American legal system to a minority of the national population.

We can distinguish the fear of this danger of minority amendment from the fear that our representatives will betray us once elected. That republican possibility is separable from the federalist possibility; the former has existed since the Nation was founded as a republic, and was clearly intended by the framers as part of the risk of founding a republic. The federalist danger exists even under the quixotic conditions of perfect representation. The framers did not intend it, and it was not provided by the Constitution until innocent additions of new states and internal migrations exposed an unforeseen weakness in the amending process.

The minority favored by the anomaly in article V is not "discrete and insular" in our society and not in need of protection, for it already holds power in the lightly populated states. In fact, the minority protected by this anomaly already dominates the legislatures of the lightly populated states. The amending power —the supreme power in our legal system— should not be left in the hands of an absolute and shrinking minority, especially one that is neither a fair cross-section of the American public nor elected to represent that public in its own way.

The potential for abuse, either by permitting basic change by a minority or by permitting veto by a scintilla of the population, does not present an intermittent, remote, or new danger. It first became possible thirty years after the Constitution was adopted, and has continued for the past 130 years. Over this long history, the minority that can adopt amendments and the sliver that can veto amendments have both shrunk; simultaneously, of course, the supermajority that can be thwarted has grown. Finally, this danger is far from speculative. Here we have a rare case in which an argument against antidemocratic abuse can be documented statistically. I invite the reader to attend to the figures in the appendix. They tell a story that should arouse the concern of all Americans, and stimulate reflection on the merits of alternative remedies.


1. The amending process is laid out in article V, as follows:

The Congress, whenever two-thirds of both Houses shall deem it necessary, shall propose Amendments to this Constitution, or, on the Application of the Legislatures of two-thirds of the several States, shall call a Convention for proposing Amendments, which, in either Case, shall be valid to all Intents and Purposes, as part of this Constitution, when ratified by the Legislatures of three-fourths of the several States, or by Conventions in three-fourths thereof . . . .

U.S. Const. art. V. [Resume]

2. A "supermajority" is any fraction significantly greater than half, such as two-thirds or three-fourths. A simple majority, by contrast, is any fraction greater than half, no matter how small. [Resume]

3. Supreme Court has struck down attempts to make state ratification depend upon a state referendum. National prohibition Cases, 253 U.S. 350 (1920); Hawke v. Smith, 253 U.S. 221 (1920); see also Opinion of the Justices to the House of Representatives, 262 Mass. 603, 160 N.E. 439 (1928); State ex rel. Tate v. Sevier, 333 Mo. 662, 62 S.W.2d 895, cert. denied, 290 U.S. 679 (1933). [Resume]

4. All statistical and demographic claims in this Article summarize data presented in the appendix. Unless otherwise noted, references to census data in the present tense are to 1980 figures. [Resume]

5. Tinkering with the Constitution, 24 Yale L.J. 573, 587 (1915). [Resume]

6. The Proposed Amendment of Article V:  A Threatened Disaster, 72 Yale L.J. 957,959 (1963); see also W. Livingston, Federalism and Constitutional Change, 234,242-44, 246-47, 310-14 (1956). For other references, see id. at 242 n.6, 243 nn.1-4. Another commentator emphasized the extraordinarily small group sufficient to veto constitutional amendments, rather than the minority sufficient to adopt them. Carman, Why and How the Present Method of Amending the Federal Constitution Should Be Changed, 17 OR. L. REV. 102, 102-03 (1938) (noting incorrectly that states comprising 100,000 citizens could veto an amendment favored by states comprising 30,000,000). [Resume]

7. For example, with 50 states, 13 states (1/4 x 50 = 12.5, rounded up to 13) are necessary to veto a proposed amendment. If there were 52 states, 15 states would be necessary (1/4 x 52 = 14, plus 1 = 15). [Resume]

8. The least populous three-fourths of the states (38) and the most populous one-fourth (13) do not add up to 50 because each number is a fraction rounded upwards. [Resume]

9. In mathematical terms, which suffice here, a federation is a set of sets of people, while a democracy is a set of people. In a federation, votes at the "federal level" are taken by the sets of people, not by the people themselves. In a democracy, votes are taken by the people directly. In a federation, the sets of people may differ in size; it is inessential for voting that one unit averages many voices into one and another averages very few into one. In a democracy, the people may cluster or be clustered into any number of groups; it is inessential for voting that the people form this or that subset for this or that purpose. A republic stands midway between a federation and a democracy. In a republic, the people are counted only in clusters (or are counted only through representatives of clusters) when the clusters are generally of equal size. Hence, the House of Representatives forms a republican chamber while the Senate forms a federalist chamber.

Note that I will argue only against federalism in legal change at the national level, including constitutional amendment, not federalism (as local autonomy and experimentation) in criminal law, education, or other areas. [Resume]

10. The Federalist No. 22, at 146 (A. Hamilton) (C. Rossiter ed. 1961). Hamilton added the following footnote to the quoted statement:  "New Hampshire, Rhode Island, New Jersey, Delaware, Georgia, South Carolina, and Maryland are a majority of the whole number of the States, but they do not contain one third of the people." Id. Hamilton states that the least populous two-thirds of the states also comprised a minority of the national population, and that the addition of new states might aggravate the problem. Id. at 147. [Resume]

11. See The Federalist No. 39, at 246 (J. Madison) C. Rossiter ed. 1961) (arguing that the federal amending process was federalist in "requiring more than a majority, and particularly in computing the proportion by States, not by citizens" (emphasis in original)). Madison added that the amending process has a "national" as opposed to a "federalist" character because it settled for a supermajority rather than unanimity of the states. Id. [Resume]

12. Selden Bacon provided the most spirited attack on the republican form of minority amendment. Bacon, How the Tenth Amendment Affected the Fifth Article of the Constitution, 16 Va. L. REV. 771 (1930). In 1930, Bacon calculated that the number of state legislators necessary to adopt an amendment was 1300. Id. at 771-72. Finis J. Garrett, writing one year before Bacon, could not get this number much below 4000. See Garrett, Amending the Federal Constitution, 7 Tenn. L. Rev. 286, 305-06 (1929). Speaking 10 years earlier, Senator Ashurst of Arizona found the number to be even higher:  "We set ourselves up as the leader among the nations in thought and as responsive to the people's will, and yet 4,500 men, if they saw fit, could Prussianize the Republic." 58 CONG. REC.5697 (1919), quoted in Martig, Amending the Constitution — Article Five:  The Keystone of the Arch, 35 MICH. L. REV. 1282 (1937). [Resume]

13. See infra Part V. [Resume]

14. This claim is based on 1790 census data, published four years after the Annapolis Convention met to amend the Articles of Confederation, three years after the Constitutional Convention was assembled in Philadelphia, and one year before the ratification of the new constitution became unanimous. Presumably these data were collected before 1790. We do not know the demographic profile in the nation prior to 1790 more precisely, nor, unfortunately, do we know exactly what the framers knew or assumed about it. But see supra Part II(A). [Resume]

15. These 12 states are listed in decreasing order by population, according to the 1980 census. See appendix, table 4. [Resume]

16. These 13 states are listed in decreasing order by population, according to the 1980 census. See appendix, table 4. [Resume]

17. Find the population of the most populous state (call it c for California). Set the voice of a citizen of that state at one unit c divided by c). To measure the voice of citizens of another state (whose population s), divide c by s.

Because we apportion senators by state and not by population, just as we apportion power in the federal amending process, this calculation also measures the relative voices of Americans in the Senate. Hence, it also measures the relative voices of Americans in those acts performed by the Senate without the concurrence of the House —in particular, the ratification of treaties and the confirmation of Supreme Court justices. U.S. Const. art. II, 2, cl. 2. [Resume]

18. An unabridged version of this table, with other data interposed for comparison, may be found in table 4 of the appendix. [Resume]

19. Courts have not interpreted the fourteenth amendment to require the one-person, one-vote principle for federal amendments because article V clearly establishes its own procedures in that regard. The (scanty) case law. however, is divided on the question of whether the fourteenth amendment requires that principle for state amending procedures. The New Jersey amending process was once challenged for replying in part upon the State Senate, whose members were not elected under the one-person, one-vote principle. The State's amending process survived the challenge despite the conceded departure from the principle. Jackman v. Bodine, 78 N.J. Super. 414, 188 A.2d 642 (Ch. Div. 1963). By contrast, the New Mexico Supreme Court struck down a section of the State's amending clause for violating the one-person, one-vote principle. State ex rel. Witt v. State Canvassing Bd., 78 N.M. 682, 437 P.2d 143 (1968). The offending section in effect treated the new Mexican counties as entities much like states in the federal Senate. See Comment, The New Mexico Constitutional Convention 1969, Constitutional Revision — Constitutional Amendment Process, 9 NAT. RESOURCES J. 422, 425-26 (1969). [Resume]

20. The term "Discrete and insular minorities" originated in Justice Stone's famous fourth footnote in United States v. Carolene Products Co., 304 U.S. 144, 153 n.4 (1938). [Resume]

21. This figure uses 1920 census data; the figure is 62.52% using 1910 census data. See appendix, table 2. [Resume]

22. This figure also uses 1920 census data; the figure is 80.78% using 1910 census data. see appendix, table 2. [Resume]

23. This figure uses 1960 census data; the figure is 70.13% using 1970 census data. Forty-seven states, comprising 96.10% of the national population (using 1960 data, or 96.15% using 1970 data), eventually ratified the twenty-fifth amendment. See appendix, table 2. [Resume]

24. The six proposals approved by Congress but defeated by the states were as follows:  (1) a scheme to apportion the members of the House (submitted with the Bill of Rights in 1789); (2) a rule to bar congressional salary increases from taking effect until after the next election of representatives (submitted with the Bill of Rights in 1789); (3) a rule to strip U.S. citizenship from any citizen accepting a title, present, or office from a foreign power without the consent of Congress (1810); (4) the so-called Corwin amendment, which would have barred future amendments authorizing Congress to abolish the "domestic institutions" of any state, including slavery (1861); (5) the so-called Child Labor Amendment, which would have given Congress authority to regulate the labor of those under 18 years of age (1924); and (6) the Equal Rights Amendment, which would have prohibited discrimination by the state the federal governments on account of sex (1972). [Resume]

25. When amendment proposals contain no expiration date, the states may consider them for a "reasonable" time. Congress determines how long a period is reasonable. Dillon v. Gloss, 256 U.S. 368 (1921). [Resume]

26. The two defeated proposals submitted with the original Bill of Rights are known to have had more than enough rejecters; the other four are known to have had too few ratifiers. On the first two, see B. Schwartz, The Great Rights of Mankind:  A History of the American Bill of Rights 187, 190-91 (1977). On the ERA, see infra text accompanying notes 27-29. [Resume]

27. In 1972, Congress passed the ERA and sent it to the states. The ERA expired unratified in 1982 after a three-year extension of the original seven-year period. [Resume]

28. See supra note 12 and accompanying text. [Resume]

29. If we count rescissions as ineffective, then the states that ratified the ERA comprised 70.13% or 72.05% of the national population, depending upon whether we use 1980 or 1970 census data. [Resume]

30. See Bacon, supra note 12, at 771-72; Garrett, supra note 12, at 304-06. [Resume]

31. See infra Part IV. [Resume]

32. W. Livingston, supra note 6, at 235 n.4. [Resume]

33. For the three mathematical tests used to indicate discriminatory impact, see the introduction to table 3 in the appendix. The table itself gives the data on which the tests are based. [Resume]

34. Alf Ross has argued that no constitutional amending clause can be used to amend itself. Ross, On Self-Reference and a Puzzle in Constitutional Law, 78 Mind 1 (1969). Ross finds a logical impossibility in self-amendment, but makes no case at all that logical impossibilities must also be legal impossibilities. The amending clause in the federal Constitution has never been used to amend itself (depending on what one counts as amendment), but the amending clauses in 48 state constitutions have done so without challenge. These data, legal arguments against Ross, and variations on the theme of self-amendment are collected in P. Suber, The Paradox of Self-Amendment:  A Study of Logic, Law Omnipotence, and Change (forthcoming). [The book was published three years after this article appeared, from Peter Lang Publishers, 1990. It is now available on the web.][Resume]

35. The proponents of at least one proposed amendment to article V tried to bootstrap the amendment into effect by asking that it be adopted in the way it proposed for future amendments. In 1869, Senator Garrett Davis of Kentucky proposed to allow ratification by popular referendum and expressly included language allowing his very proposal to be ratified by referendum. Four years later Representatives Charles Howell Porter of Virginia proposed to permit ratification by referendum, but eliminated Davis' paradox and submitted his proposal for adoption under existing methods. H. Ames, The Proposed Amendments to the Constitution of the United States During the First Century of its History, in 2 Annual Reports of the American Historical Association for the Year 1896 293, 294 (1897). [Resume]

36. See L. Orfield, The Amending of the Federal Constitution 217 (1942). [Resume]

Ribbon] Peter Suber, Department of Philosophy, Earlham College, Richmond, Indiana, 47374, U.S.A. Copyright © 1987, Peter Suber.