The Amendments Project (TAP) is a searchable archive of the full text of nearly every amendment to the U.S. Constitution proposed in Congress between 1789 and 2022 (more than 11,000 proposals); records of petitions introduced in Congress between 1789 and 1949 that propose, support, or oppose constitutional amendments (more than 9,000 petitions); and thousands of proposed amendments that never made it to Congress. The congressional proposals come from congressional records. The petitions are extracted from the Congressional Petitions Database.1 Other records come from everything from party platforms to online petitions. We have tagged these records with a set of topics developed by the Comparative Constitutions Project for its sister site, Constitute. The project received funding from the National Endowment for the Humanities, the Harvard Data Science Initiative, and Harvard's Inequality in America Initiative.

Only twenty-seven amendments to the U.S. Constitution have ever been ratified; the Amendments Project is, then, an archive of failures. Yet the failures matter, both to the historical record, and to ongoing constitutional reform.

The U.S. Constitution can be altered by formal amendment, as outlined in Article V of the Constitution, and also by changing meaning and shifting interpretation, especially that provided by decisions of the U.S. Supreme Court. Proposed amendments are very unlikely to meet with success but there are other reasons than ratification to pursue this path. Members of Congress very often introduce proposed amendments for political purposes: sending a message to their constituents, say, or altering the direction of a political debate, or influencing pending legislation, or advancing a campaign being fought outside the halls of Congress. In some cases, “failed” amendments have been more consequential, in American political discourse and even in the nation's constitutional order, than ratified ones.

The Amend data includes not only amendments formally introduced in Congress but also all those that were instead advanced in newspapers and pamphlets, in party platforms, at political conventions, in presidential papers and, in the twenty-first century, on the Internet, in formats that include social media campaigns and online petitions through venues like change.org. TAP aims to be as comprehensive as possible and has the particular objective of “constitutionalizing” proposals made by people who were disenfranchised—the enslaved, many immigrants, other people of color and, until 1920, women. Those ideas include, for instance, amendments discussed at “Colored Conventions,” available through the digital Colored Conventions Project; proposals made during women's rights conventions; proposals made in foreign-language American newspapers such as The Chinese-American; and collections of Tribal Constitutions. Some of this research has already been added to the data; other research is not yet complete. The project is ongoing.


Only twenty-seven amendments to the U.S. Constitution have been ratified over the course of American history but more than eleven thousand have been introduced in Congress. The records in our data are principally based on the following works:

Although we used these sources to compile our initial list of amendments, the Congressional amendments in the Amend dataset move considerably beyond them. None of these earlier collections includes the full text of the amendments; none have tagged them by topic; none have cross-referenced their primary and co-sponsors to biographic data. And the Ames data, in particular, was compiled using a different methodology than the other sources, which we have made corrections to.2

Efforts to keep track of these proposed congressional amendments began more than a century ago. In 1891, the historian Herman Ames completed a doctoral dissertation at Harvard that he published five years later as The Proposed Amendments to the Constitution of the United States during the First Century of Its History. It included, as a nearly hundred-page appendix, “A calendar of amendments proposed to the constitution, from the date of its ratification to March 1889.” Unfortunately, Ames's calendar does not include the actual text of the proposed amendments, a practice that has been followed by each of his successors. Ames's work was continued by Charles Tansill, professor of history at American University. In 1926, working with the Library of Congress, Tansill produced a compendium of amendments introduced in Congress from 1889 to 1926. The Library of Congress published a continuation in 1947, with amendments from 1926 to 1947; and another in 1953, and another in 1963, in a collection that regularized all entries from 1926 to 1963These print compilations were later abandoned as the records of Congress were made available in new formats, beginning with microfilm and microfiche. In 2003, the political scientist John Vile compiled all of these earlier collections into a single, four-volume printed collection, Proposed Amendments to the U.S. Constitution, 1787-2001; a supplement was added that updates the collection to 2010. Within the Amendments Project, this body of research is known as “the Ames data.” In 2016, for purposes of an exhibit called Amending America, staff at the National Archives and Record Administration entered the Ames data into a spreadsheet, producing a table (“the NARA data”) with basic information on some 12,000 amendments proposed between 1787 and 2014. Like the Ames data, the NARA data does not include the text of proposed amendments. It also reproduces a number of errors in the Ames data, including considerable duplication of records.

We began the Amend project in 2020. After we had already imported and corrected the Ames data, Howard University political scientist Robinson Woodward-Burns published Hidden Laws: How State Constitutions Stabilize American Politics (New Haven: Yale University Press, 2021). Woodward-Burns had separately corrected two variables in the NARA data: using Appendix C in John R. Vile's Encyclopedia of Constitutional Amendments, Proposed Amendments, and Amending Issues, 1789-2010, 3rd ed. (Santa Barbara, Calif.: ABC-CLIO, 2010), along with other sources, he corrected for missing and incorrect entries in the “year” variable and he eliminated typos in the “description” field. He also used the corrected descriptions to create binary variables coding amendments by topic. Woodward-Burns generously shared his data with us. Although we opted to separately digitize Ames's summaries and in other ways proceeded following different methods, we would not have uncovered some of the methodological differences between our sources without Woodward-Burns's prior efforts.


The right to petition is guaranteed in the First Amendment, to all persons. It can be used by anyone, including by people ineligible to vote and even ineligible to become citizens. The Congressional Petitions Database (CPD), aspires to remedy the problem by “tracking virtually every petition introduced to Congress from 1789 to 1949.” The dataset contains the records of more than 500,000 petitions, documents that promised to represent the “voice of the voteless.” Early analysis demonstrates that, “in several key moments, the unenfranchised—women, African Americans, Native Americans, the foreign born, and even children—leveraged the petition process, often with great success”.3 We searched the CPD for petitions relating to proposed constitutional amendments and identified more than nine thousand petitions. Very often, these petitions were endorsements of existing proposals. Sometimes, they were proposals for new amendments.

Due to the variable text quality within the CPD, we identified amendment-related petitions using various natural language processing techniques to regularize the text (including stopword, punctuation, and number removal as well as word stemming), followed by regular expression searches. At risk of oversimplification, we searched for petitions that either mentioned a constitutional amendment or included a call to change or amend the constitution. We also searched for named amendments such as the suffrage amendment or the anti-polygamy amendment and numbered amendments such as proposals for a sixteenth amendment when only fifteen amendments had been ratified. Our search results included several false positives, which we removed manually. We intend to post a technical paper with more details in the near future.


The online petitions in our data were sourced in 2023 from three active online petitioning websites (change.org, moveon.org, and thepetitionsite.com) as well as from archived data for the “We the People” petition platform for the Obama and Trump administrations.4 To merit inclusion, these petitions had to meet a minimum signature count of 100 signatures by March 2023 for the active online petitioning websites and, due to data limitations, 150 signatures for the archived White House petitioning data.

Incorporating online petitions into the Amend database required choosing which petitions to count. Should any petition that suggests any kind of change inconsistent with the current constitutional order be included among the data? We decided not, for two reasons: First, not everything posted on the internet represents a serious proposal or one with significant public support. The right to grill, for instance, does not clear our bar. Second, the focus of the Amend Project is specifically the Article V amendment process—as opposed to other means of achieving social and constitutional change, such as ordinary legislation.

To address the first issue, we set a minimum signature count: Petitions with fewer than one hundred signatures do not appear among the Amend data. (The signature count minimum threshold for petitions from the Obama and Trump administration websites was 150 due to data limitations.) Addressing the second issue proved more difficult. What counts as a call for formal constitutional change? If a petitioner calls upon Congress to take action that Congress is not constitutionally capable of taking, is that petitioner implicitly calling for constitutional change? We decided not: Such a petition is not, on its own, evidence of a formal effort to amend the Constitution; it may simply reflect disregard for or ignorance of what the Constitution says.

Ultimately, we decided to include petitions in the Amend database if they had a hundred or more signatures and satisfied any of the following criteria:

Even with these general criteria in place, we had to make a number of case-by-case decisions in ambiguous instances. We hope, however, that these guidelines have maximized the interest and usefulness of the petitions included in the Amend database.


Between 2020 and 2023, TAP researchers trawled through dozens of major and minor collections and databases of digitized historical records, especially periodicals, in search of proposed amendments to the U.S. Constitution. Here we tended to err on the side of including proposals rather than excluding them. For an amendment proposal to enter the realm of public discussion in a newspaper we took as, by definition, a higher level of seriousness and public engagement than, say, a Tweet with 4 likes. We also targeted specific periodicals. For instance, a search of conservative periodicals in the Hathi Trust collection revealed that the National Review is not included there; we therefore conducted a targeted search of the digitized archive of the National Review.

Some of the collections we searched included:

We are also grateful for the collections of failed proposals compiled by John R. Vile: Encyclopedia of Constitutional Amendments, Proposed Amendments, and Amending Issues, 1789-2010, 3rd ed. (Santa Barbara, Calif.: ABC-CLIO, 2010); Rewriting the United States Constitution: An Examination of Proposals from Reconstruction to the Present (New York: Praeger, 1991); and Re-Framers: 170 Eccentric, Visionary, and Patriotic Proposals to Rewrite the U.S. Constitution (Santa Barbara, California: ABC-CLIO, 2014).

This work continues and we would be happy to hear from researchers who have located sources of amendment discourse.


Each record for which we have sufficient data is tagged by at least one and usually more than one topic. To allow for comparison between the Amend data and a sister collection of the world's written constitutions, we relied on the topic tags developed by the Comparative Constitutions Project (CCP) for its site, Constitute. Constitute offers a description of its topics, along with a topic browser, here. And we have provided a list of our tags, with descriptions, here. Please note that topics do not indicate polarity. For instance, “reproduc” concerns reproductive rights. It is used both for proposals that restrict reproductive rights and for those that guarantee or extend those rights.

We inherited 345 topics tags from the CCP and, in the process of assigning tags to U.S. amendment proposals, we added about thirty U.S.-specific topics to ensure a better fit. We also modified or expanded the scope of several CCP tags to more closely fit concepts from American constitutional history. For amendment proposals for which we do not have the full text, we applied tags based on the knowledge of the proposal's general aim as well as the text of similar proposals from the time period. We assigned tags, and later reviewed them for accuracy and consistency.

The topic tags vary in breadth and scope. A tag like “legpow” (“Powers of the national legislature”) is broad: it was applied to any proposal with a provision that either empowers or limits the legislature, regardless of the subject matter. Other tags are narrower, like “illsub” (“Drugs, alcohol, and illegal substances”) which was applied almost exclusively to Prohibition-related proposals. (And note that any proposed amendment that seeks to control drugs, alcohol, or illegal substances by way of Congressional action will have both of these tags.)

Interestingly, the frequency with which a particular constitutional provision is introduced is not necessarily an indication of its likelihood of adoption. For instance, as illustrated below, the most frequent topic in the congressional dataset is gender equality. The Equal Rights Amendment was first formally introduced in Congress in 1923 and passed in 1972. It remains the subject of political and legal efforts.

Top Topics in Congressional Proposals, 1789-2021

Top topics in congressional proposals, 1789-2021. The top 5 topics include 'Equality regardless of gender' with over 1200 proposed bills, followed by 'deputy executive', 'head of state selection', 'powers of the national legislature', and 'balanced budget'

Our quest to find and categorize every proposed amendment to the U.S. Constitution is ongoing and we hope to add more amendments from existing and new sources in the near future.