How Legislatures Elected Senators, 1789 to 1866: Joint Ballots, Instructions, Contests and the Act of 1866

Lesson 01 ended with the text. This lesson is about what the text did not say. Article I, Section 3 told the states that their legislatures would choose senators. Article I, Section 4 let each legislature set the time and manner, subject to a federal override that Congress declined to use until 1866. So for the first 77 years of the republic there were as many methods of electing a senator as there were states, and the only referee was the Senate itself, which under Article I, Section 5 was "the Judge of the Elections, Returns and Qualifications of its own Members." The history of that period is the history of the gaps in the text getting filled, one dispute at a time, until Congress wrote a single rule.

Two methods and a doctrine

A state legislature has two houses. The Constitution said the legislature would choose. It did not say whether the two houses would vote as two bodies or as one. George H. Haynes, whose 1906 study The Election of Senators remains the fullest account, summarized the practice: "In the early years the choice of senators was generally, though not universally, made by concurrent vote of the two houses of the legislature in separate session. At a later period, about one-half of the States came to require the election to be made by a vote in joint convention."

The concurrent vote treated the election like a bill. The senate voted; the house voted; each had to produce a majority for the same man or there was no election. The joint ballot, or joint convention, put every member of both houses in one room, usually the larger house's chamber, and let a majority of the whole body decide. In a state with a 30-member senate and a 100-member house, the difference was enormous: under the concurrent method 16 senators could block any candidate; under the joint ballot the house could nearly elect on its own.

Haynes reports that "the weight of constitutional authority seems to have been against" the joint ballot, "on the ground that when the Constitution prescribed an election by the legislature, the intent was that that body should perform that function legislatively, as in the passing of any ordinary act, i.e., by concurrent vote." He cites Story's Commentaries, sections 705 to 708, and Chancellor Kent, who in turn cited the Federal Farmer's twelfth letter, as authority for the concurrent reading. Read Story yourself before repeating that. In the 1833 text as the Founders' Constitution prints it, the passage is section 703, and it is more cautious than Haynes's citation suggests: "The constitution has not provided for the manner, in which the choice shall be made by the state legislatures, whether by a joint, or by a concurrent vote," and the choice "is made by a concurrent vote" only "generally, but not universally." Story also defines the two terms the opposite way from later usage, calling the single-assembly vote "concurrent" and the separate vote "joint." The firm constitutional claim belongs to Kent and to Haynes's reading. He adds a point that goes back to lesson 01: Gerry and others had favored legislative election partly because they believed "the aristocratic upper house would hold in restraint the 'turbulency of democracy' in the lower branch," a confidence that made sense only if each house voted separately. Put the two houses in one room and the lower house's numbers swamp the upper house's judgment. If the election was a legislative act, one more question followed: could the governor veto it? Haynes: "universal practice has been against recognizing any such executive participation in the choice of senators," and Story said the same, that executive participation had been "silently and universally settled against."

	Concurrent vote	Joint ballot
Who votes	Each house separately, in its own chamber	All members of both houses together as one body
What it takes to win	A majority in the senate and a majority in the house, for the same person	A majority of the members voting in the joint assembly (New Jersey's 1865 plurality rule was the exception that caused the 1866 Act)
Where it was used	Most states in the early republic; New York and New England kept it longest	About half the states by the mid-nineteenth century, especially in the West and South
Constitutional view	Favored by Kent, and by Haynes citing Story, as the natural meaning of "chosen by the Legislature"; Story's own text says the Constitution "has not provided for the manner"	Defended as "the uniform practical construction of the Federal Constitution for the last half-century" in the Cameron case (1857)
Failure mode	Deadlock when the two houses were held by different parties (Maine, Tennessee 1841-43, Indiana 1855-57)	Deadlock when no faction had a majority of the whole; minority blocking by refusing to attend
After 1866	Required as the first vote, on the second Tuesday after organization	Required as the fallback the next day and every day thereafter until election

The method also varied by ballot type. Some states voted by secret written ballot; others, especially in the West, viva voce, each member's vote called and recorded. Maryland had a geographic rule requiring one senator from the Eastern Shore, repealed in 1867 when the dominant party wanted to elect a man from the wrong side of the Chesapeake, promptly re-enacted, later ignored, and gone only in 1896 (Haynes, ch. II). Other states had their own quirks about the day of the election and the form of the certificate. None of it was federal law.

Caucuses, campaigns and the legislative election in practice

What actually decided a Senate election in the first half of the nineteenth century was usually the majority party's caucus. The Senate Historical Office's overview puts the shift in the 1830s and 1840s: party caucuses in the legislatures chose candidates, and the majority party's choice was then elected in form. By 1858 the Illinois contest between Abraham Lincoln and Stephen Douglas showed a further stage: the two Senate candidates campaigned across the state, and candidates for the legislature publicly declared which man they would vote for. Bryce, in 1893, described the mature form of this: the state party convention would vote on the senatorial nominee, "This vote binds the party within and without the State legislature," and at the legislative election "candidates for seats in that legislature are frequently expected to declare for which aspirant to the senatorship they will, if elected, give their votes" (The American Commonwealth, vol. 1, ch. X, with the Lincoln-Douglas campaign cited in a footnote as an early example). His conclusion: "the election of senators has in substance almost ceased to be indirect."

That is the system as it had evolved by the 1880s. In the earlier decades it was looser. Legislatures sometimes elected men who had not sought the office; sometimes a dozen ballots were needed as factions shifted; sometimes a state went without a senator because its legislature could not or would not act. New York did not elect its first senators in time for the First Congress, which met on March 4, 1789: the Senate Historical Office attributes the delay to "a deadlocked state legislature," and Rufus King and Philip Schuyler did not take their oaths until July 25 and July 27, 1789 (New York timeline). The Senate itself could not muster a quorum until April 6 (Pennsylvania timeline). Hamilton's reassurance in Federalist 59 that a few states "forbearing the appointment of senators" could not "impair the activity of the body" was tested in the Senate's first month.

Schiller and Stewart's Electing the Senate (Princeton, 2015), built on a dataset of state legislators' roll call votes for senator from 1871 to 1913, finds that parties controlled the party label of the winner far more reliably than they controlled the field of candidates, and that money, personal ambition and intraparty fights shaped outcomes within the majority party. Their data begin after this lesson's period, but the pattern they describe, party control of the result with disorder inside the party, is already visible in the Indiana and Pennsylvania cases below.

Instruction: the legislature tells its senator how to vote

Under the Articles of Confederation the states could recall their delegates. The Constitution dropped recall, as lesson 01 explained. What survived was the older practice of instruction: a legislature passing a resolution directing its senators to vote a certain way. Rufus King had predicted at the Massachusetts convention in January 1788 that senators "will be bold men indeed to act contrary to" public instructions (Founders' Constitution, doc. 10). William H. Riker's 1955 article in the American Political Science Review notes that seventeen state constitutions, mostly in New England and the Midwest, eventually guaranteed a right to instruct, and lists fifteen forced resignations under instruction from Adams in 1808 to Haywood of North Carolina in 1846, plus six he calls "almost forced" (Sprague 1835 through Everett 1854). Two are well documented in Senate and presidential records.

John Quincy Adams, 1808

Adams had been elected to the Senate from Massachusetts as a Federalist in 1803. He would not oppose the Jefferson administration's Embargo, which Massachusetts shipping interests hated. The legislature responded in two moves. It instructed him to work for repeal of the Embargo Act, and it held the election for his seat months before it was due, in the spring of 1808 rather than the following winter, choosing James Lloyd Jr. The New Nation Votes election record gives the General Court vote as Lloyd 248, Adams 213, Laban Wheaton 1, and notes that the election was "held several months earlier than usual" and that Adams resigned in response. The Senate Historical Office's Massachusetts timeline dates the resignation June 8, 1808 and says Adams "was forced from office for refusing to obey the legislature's instructions to push for repeal of the Embargo Act." A special election five days later filled the rest of his term. Adams did not obey the instruction. He quit.

John Tyler, 1836

The Senate had censured President Jackson in 1834. Jackson's supporters campaigned to expunge the censure from the Senate Journal, and Virginia's legislature instructed its senators to vote for it. Tyler, whom a coalition of Clay and Adams supporters and Democrats opposed to John Randolph had elected to the Senate for the term beginning March 4, 1827 (Richardson's sketch dates the legislature's choice to December 1826), would not. The biographical sketch in Richardson's Messages and Papers of the Presidents states it plainly: the legislature "instructed the Senators from that State to vote for expunging the resolutions of censure upon President Jackson"; Tyler "refused to obey the instructions, resigned his seat, and returned home February 29, 1836."

These two cases show what instruction could and could not do. It could not compel a vote. It could, when backed by a legislature willing to elect someone else, make a senator's position untenable, and a senator who took the doctrine seriously resigned rather than either obey against his conscience or defy a body he believed had the right to instruct him. Most senators did neither. Bryce's dry footnote in 1893 is the realistic summary: "A State legislature sometimes passes resolutions instructing its senators to vote in a particular way, but the senators are of course in no way bound to regard such instructions" (The American Commonwealth, vol. 1, ch. X, note). Riker's reading is that the practice faded as party conventions took over the job of pledging legislators to Senate candidates in advance, so that the discipline moved from the legislature's resolution to the party's nomination.

The Senate as judge of its own elections

Article I, Section 5 made each house "the Judge of the Elections, Returns and Qualifications of its own Members." For the Senate, that meant judging whether a state legislature had validly elected the man whose credentials lay on the table. The Senate Historical Office's index of contested Senate elections lists every case; the practice that developed was to seat a claimant whose credentials were regular on their face and then refer the challenge to committee, usually the Judiciary Committee before 1871 and the Committee on Privileges and Elections after. Four cases from before 1866 show the range.

Gallatin, 1794: the qualification clause

Albert Gallatin, born in Geneva, had been elected by the Pennsylvania legislature. He presented his credentials and was seated on December 2, 1793, and the same day the Senate received a petition from Conrad Laub and others arguing that he had not been a citizen for the nine years Article I, Section 3 requires (Pennsylvania timeline). The petitioners' counsel argued that Gallatin had never met Virginia's or Pennsylvania's own citizenship requirements of oaths, property and residence; Gallatin answered that he had lived in America thirteen years, that the Revolution had made inhabitants mutually citizens, and that doubtful laws should be read in the defendant's favor. On February 28, 1794 the Senate resolved that his election was void because he had not been a citizen for the required period. The vote was 14 yeas to 12 nays, recorded in the Annals of Congress and reprinted in the Founders' Constitution. It was the Senate's first contested-election case and the first use of its power to unseat a member a legislature had chosen.

Indiana, 1857 to 1859: who may hold a joint convention

Indiana's legislature had failed to fill one Senate seat since 1855, when the Democrats, holding a slim majority in the state senate, refused to meet the larger house, controlled by the new anti-slavery opposition, in a joint session they would have lost (overview). By 1857 the roles had reversed: the Republicans in the state senate were the ones refusing a joint session, hoping to stall until they controlled both houses (case page). In February 1857, with Jesse D. Bright's term also expiring, a minority of the state senate went to the house chamber and, with a majority but not a quorum of house members, balloted for both seats, electing Bright to a full term and Graham N. Fitch to the vacancy. The Senate Historical Office's case page says the Democrats "simply ignored the state law that required a joint resolution" to hold an election. The Senate seated Fitch on February 9, 1857, referred his credentials the next day, seated Bright on March 4, and in December 1857 referred both to the Judiciary Committee. The committee, five Democrats and two Republicans, upheld both men on May 24, 1858, and on June 12 the Senate rejected a motion to deny them their seats, 23 to 30. When Republicans won the Indiana legislature later that year, they treated the seats as vacant and elected Henry S. Lane and William M. McCarty, who presented credentials January 24, 1859. The Judiciary Committee ruled their elections invalid because no vacancies existed, and on February 14, 1859 the Senate voted 30 to 15 that its June 1858 judgment was final. Haynes draws the lesson: the case was "the clearest of proof that the absence of uniform regulation of senatorial elections served as a constant temptation to sharp practice for partisan advantage." The committee also laid down a rule that mattered later: a state legislature "possessed no authority to revise the decision of the Senate" on its members' qualifications.

Cameron, 1857: is a joint ballot constitutional?

Simon Cameron took the oath as Pennsylvania's first Republican senator on March 6, 1857. Two days later the Democratic Senate received "a petition from 69 members of the Pennsylvania legislature protesting Cameron's election," on grounds that included the lack of a concurrent majority in each house and, among other things, "corrupt and unlawful means" (Pennsylvania timeline; Haynes, ch. II and III). The Judiciary Committee found the bribery allegation "entirely too vague and indefinite" to investigate and the concurrent-majority objection "untenable under the statute of Pennsylvania, and 'the uniform practical construction of the Federal Constitution for the last half-century.'" That sentence settled, as a matter of Senate practice, that a joint ballot was a valid way for a legislature to choose. The constitutional theorists had lost to the fact that half the states had been doing it for decades. Haynes counts Cameron's as the first of ten bribery cases to reach the Senate, with the next not until 1872 (lesson 03).

Year	State, senator	Question	Senate's decision	Vote
1794	Pennsylvania, Albert Gallatin	Nine-year citizenship qualification	Election void, Feb. 28, 1794	14 to 12
1857-58	Indiana, Jesse D. Bright and Graham N. Fitch	Election by a joint session held without a state senate quorum or a joint resolution	Seats upheld, June 12, 1858	Motion to deny seats lost 23 to 30
1857	Pennsylvania, Simon Cameron	No concurrent majority in each house; vague bribery charge	Election upheld; joint ballot accepted as "uniform practical construction"	Committee report adopted
1859	Indiana, Henry S. Lane and William M. McCarty	Whether a legislature could re-elect to seats the Senate had already upheld	Elections invalid; no vacancy existed; Senate's judgment final, Feb. 14, 1859	30 to 15
1866	New Jersey, John P. Stockton	Election by plurality under a joint-meeting rule adopted 41 to 40	First upheld Mar. 23 with Stockton voting; reconsidered; seat denied Mar. 27, 1866	22 to 21, then 23 to 20

Deadlock before 1866: the concurrent vote fails

The concurrent method had one predictable weakness. When the two houses were held by different parties, neither would yield, and the seat stayed empty. William Pitt Fessenden of Maine told the Senate during the 1866 debate that the Maine Senate had elected him eighteen times in a single session while the lower house refused to concur, so the seat stayed vacant for that Congress. Tennessee had only one senator in the Twenty-seventh Congress (1841 to 1843) because its two houses refused to go into joint convention. California's legislature failed to elect three times in five years, in 1851, 1855 and 1856. Indiana's seat sat empty from 1855 until the irregular election of February 1857, and the Senate Historical Office notes that California and Missouri were also deadlocked in 1855 and each left a seat vacant until 1857 (Haynes, ch. II; overview). These were the cases that persuaded Daniel Clark of New Hampshire, who would report the 1866 bill, that the trouble came from legislatures not being compelled to meet jointly. Haynes thought Clark had misdiagnosed it, and the next lesson is the evidence.

Congress had noticed before 1866. Haynes records that the first bill to regulate Senate elections was introduced in 1814, that another was brought in less than three weeks after the Indiana election of 1857 "to prescribe the time and manner of electing senators in Congress, and the form of their credentials," referred to the Judiciary Committee "and heard from no more," and that a third was reported with a substitute in 1858 and never considered. Then came the war.

The Stockton case and the Act of 1866

New Jersey, 1865

In 1865 the New Jersey legislature met in joint session to fill the Senate seat for the term beginning March 4, 1865. Before voting, the joint meeting adopted, 41 to 40, a rule that a plurality of those present would elect, rather than a majority of all members. John P. Stockton, a Democrat, then won with a plurality. Opponents argued that a rule change of that kind needed a majority vote in each house separately, which it had not received in the assembly. Stockton's credentials were presented on December 4, 1865, along with a memorial from part of the legislature disputing the election. The Senate seated him and on January 8, 1866 referred the credentials to the Judiciary Committee. On January 30 the committee, chaired by Lyman Trumbull of Illinois, reported the election valid: New Jersey's statute "contained no stipulations regarding how the joint meeting would be organized and the election conducted," and such rules were customarily adopted at the start of each session (Senate Historical Office, Stockton case).

The two votes

Daniel Clark opened the challenge on the floor on March 22, 1866, arguing that the Constitution was silent on plurality elections and that the custom of the states required a majority. On March 23 the Senate voted 22 to 21 to affirm the committee's report and Stockton's seat. The margin was Stockton himself. Lot M. Morrill of Maine had been paired with Stockton's absent Democratic colleague William Wright; Morrill broke the pair and voted against the resolution; Stockton then demanded to be recorded and cast the deciding vote in his own favor. On March 26 Charles Sumner objected to a senator voting on his own case. After debate the Senate chose to reconsider rather than simply strike Stockton's vote, and on March 27, 1866, with Stockton not voting, it denied him the seat, 23 to 20. Stockton's own defense had been candid about the state of the law: "Senators are not always elected in New Jersey, or in any other State, I presume, in precisely the same way."

The politics were not hidden. Radical Republicans including Sumner and Fessenden wanted the two-thirds majority needed to override President Johnson's vetoes, and unseating a Democrat helped. On April 6, 1866 the Senate overrode the Civil Rights Act veto 33 to 15. Stockton's seat stayed vacant until September 1866, when the legislature elected a Republican to finish the term. James G. Blaine later called the 1866 election law "the direct fruit" of the Stockton controversy and read it as a sign of the postwar shift toward national power: before the war "every power was withheld from the national government which could by any possibility be exercised by the state government," and now "everything which may be done by either nation or state may be better and more securely done by the nation" (Blaine, Twenty Years of Congress, vol. 2, p. 160, quoted in Haynes, ch. II).

Drafting and debate

After Stockton was unseated the Senate instructed the Judiciary Committee to look into "a uniform and effective mode of securing the election of senators by the legislatures." Clark reported the bill on July 9, 1866. The Senate debated it for a single day. Haynes, working from the Congressional Globe, identified three points of real disagreement.

The first was open voting. The bill required each legislator to vote viva voce. Fessenden objected that the open vote "was liable to put men under restraints from party discipline which would lead them to act against their conscientious convictions." Willard Saulsbury of Delaware made two arguments in a row that cut against each other: that constituents would know how their legislator voted anyway, so the secret ballot did no harm, and that the open vote would expose a legislator "to the view of those to whom he might be under, it may be, pecuniary obligations." Haynes called the second remark "surprisingly prophetic of Delaware senatorial elections of the present day." On the other side, Clark and Sumner, who favored the secret ballot in ordinary elections, argued that a legislator voting for senator "acts in a representative capacity; his constituents may even have given him specific instructions, and it is, therefore, their right to know for whom he votes." The open vote stayed.

The second was concurrent versus joint. Kent's view that the Constitution required a concurrent vote was before the Senate, but as Haynes says, "practical considerations had great weight." To reduce failures to elect like Indiana's in 1857, there had to be a joint vote; as a concession "to the practice in New York and in New England," the bill required the two houses to vote separately first and go to joint assembly only if they disagreed. John Sherman of Ohio objected to the preliminary separate vote as useless and worse than useless, since it would disclose each member's preference and the division between the houses, showing "at the very outset how easy it might be for a small minority to prevent, if it could not control, the election." The provision stayed. Haynes: "experience has proved that the objections were well grounded."

The third was how far the election could stop the state's business. As reported, the bill said the joint assembly "shall continue to vote for senator, without interruption by other business, until a senator is elected." Sherman protested that a small faction could use that to hold all state legislation hostage. Clark answered: "I do not believe it would occur once in a hundred years that any third party would stand out in the way the Senator from Ohio suggests." Senator Johnson would have gone further, saying electing senators was a higher duty than ordinary legislation and that "depriving the State of power to make its own laws was not a disproportionate penalty." The clause was softened to one vote a day. Haynes's verdict on the exchange: experience "has left no doubt as to whether the Senator from New Hampshire or the Senator from Ohio had the clearer comprehension of political tendencies."

Saulsbury opposed the whole bill as "a deplorable interference" by the federal government in state affairs and said that if legislatures had failed to elect "a little oftener, perhaps it would have been for the public good, but certainly the legislation of the country has not suffered owing to this fact." At the end he moved to postpone indefinitely, comparing the bill to cucumbers best dressed "with vinegar, pepper, salt and mustard, and then throw them to the hogs." The motion failed. The Senate passed the bill 25 to 11. The House passed it under the previous question without debate, over motions from Iowa and Kentucky members to table and adjourn. It was approved July 25, 1866 (Haynes, ch. II, citing the Senate Journal of July 11 and House Journal of July 23, 1866, and the Congressional Globe, 39th Congress, 1st Session).

What the Act of July 25, 1866 required

The statute is chapter 245 of the first session of the 39th Congress, 14 Stat. 243, "An Act to regulate the Times and Manner of holding Elections for Senators in Congress." It is three sections long and the Statutes at Large page is one page. Haynes reprints it in full (ch. II, pp. 34-35).

Section	Requirement	Exact language
1. When	The legislature chosen next before a senator's term expires elects his successor on the second Tuesday after it meets and organizes	"shall, on the second Tuesday after the meeting and organization thereof, proceed to elect a senator in Congress"
1. First vote	Each house votes separately, aloud, and records on its journal the person who has a majority of votes cast in that house, or the fact that no one did	"Each house shall openly, by a viva voce vote of each member present, name one person for senator"
1. Joint assembly	At noon the next day the two houses meet together and read the journals; if the same person has a majority in each house he is declared elected	"At 12 o'clock, meridian, of the day following ... the members of the two houses shall convene in joint assembly"
1. Joint vote	If not, the joint assembly votes viva voce; a majority of all votes, with a majority of all members elected to both houses present and voting, elects	"a majority of all the members elected to both houses being present and voting, shall be declared duly elected"
1. Daily ballot	If no majority on the first day, the joint assembly meets at noon every succeeding day of the session and takes at least one vote until a senator is elected	"take at least one vote until a senator shall be elected"
2. Vacancies	A vacancy existing when the legislature meets is filled the same way on the second Tuesday after organization; one arising during the session, on the second Tuesday after the legislature is organized and has notice	"on the second Tuesday after the legislature shall have been organized and shall have notice of such vacancy"
3. Certificate	The governor certifies the election under the state seal to the President of the Senate, countersigned by the state's secretary of state	"to certify his election, under the seal of the State, to the president of the Senate of the United States"

Read against the two methods of the first section of this lesson, the Act is a hybrid. It starts with a concurrent vote, out of respect for the old doctrine and the Northeastern practice, and falls back to a joint ballot by majority, which is what the Western states already did and what made a Stockton-style plurality impossible. It bans the secret ballot. It fixes the date so that no legislature could claim, as Indiana's had, that no joint session was lawfully called. And it keeps the joint assembly voting, at least once a day, for as long as the session lasts. What it does not do is require anyone to compromise, or set any consequence for a session that ends without an election.

What the Act did not solve

The Senate Historical Office's assessment is that the 1866 law "brought consistency to the election of senators but did not put an end to electoral deadlocks and resulting vacancies," and that "in some cases, state legislatures adjourned without electing a senator" (overview). Haynes was blunter after forty years of evidence: "Much that was expected from this law it has failed to accomplish. It cannot be said that it has had any considerable effect in discouraging deadlocks, or in preventing vacancies in the Senate. Indeed, whether because of the law or in spite of the law, both of these evils have been on the increase since its passage." The record of deadlocks between 1891 and 1905 that he compiled, 46 contests in 20 states and 14 sessions that ended with no election, is the subject of the next lesson.

Haynes listed smaller failures too. The certificate provision in section 3 was so loosely followed that in 1883 a senator proposing an amendment claimed that of the senators chosen at the last election, "one-half did not have certificates which would stand the" test of the statute. The requirement that some legislatures elect fifteen or eighteen months before the vacancy, because of the timing of their sessions, produced early elections of men whose party had since lost the state. And the procedure itself stayed obscure to the public and the press. In January 1905 the Springfield Republican reported that Nelson W. Aldrich had been "nominated by both houses" of the Rhode Island General Assembly, when under the Act the separate majority votes had elected him outright and the joint assembly next day was a formality; the same week newspapers announced that Missouri's Thomas K. Niedringhaus had been "elected" because he had a majority of the combined votes of the two houses, when the Act requires "a majority of all the votes in each house" on that first day and he had not got it in both. Missouri went sixty days without a senator-elect (Haynes, ch. II and III).

Sherman's two objections in 1866 are the key to why. The separate first vote showed every faction exactly how many votes stood between it and a block. And the daily vote, with no bar on other business, let a minority of a joint assembly refuse to vote, break the quorum, or vote for a hopeless candidate for weeks while the legislature carried on. The Act made failure to elect orderly and visible. It did not make it costly to the people causing it.

Questions and answers

Which method did most states use before 1866, and did it matter?

Haynes found that the early practice was generally the concurrent vote, with each house voting separately and both having to agree, and that about half the states later moved to joint conventions. It mattered because the two methods fail differently. Concurrent voting let a small upper house block indefinitely, which is how Maine, Tennessee and Indiana lost representation. Joint balloting let the larger house dominate and, in New Jersey's case, let a one-vote majority of the joint meeting change the rules to elect by plurality.

Were instructions from a legislature legally binding on a senator?

No. Nothing in the Constitution gives them force, and Bryce's 1893 footnote states the settled understanding that senators "are of course in no way bound to regard such instructions." They had political force. Adams and Tyler resigned rather than defy or obey, and Riker lists thirteen others between 1808 and 1846. A senator who ignored instructions and intended to seek re-election from the same legislature took a risk; one who did not care about re-election took none.

Why did John Quincy Adams resign instead of just voting as he pleased for the rest of his term?

Because the Massachusetts legislature had already elected his successor, James Lloyd Jr., months early, by 248 to 213 in the General Court, and had instructed him to work for repeal of the Embargo. Staying meant serving out a term as a repudiated senator under instructions he would not follow. The Senate Historical Office's timeline says he "was forced from office." He resigned June 8, 1808, and a special election five days later filled the remainder of the term.

How could the Senate unseat Gallatin when Pennsylvania's legislature had plainly elected him?

Article I, Section 5 makes the Senate the judge of the "Qualifications of its own Members," and Article I, Section 3 requires nine years of citizenship. The legislature's choice could not override a constitutional qualification. The Senate decided, 14 to 12 on February 28, 1794, that Gallatin had not been a citizen long enough, and the seat was void. The same power let the Senate later rule on whether a legislature had followed its own procedures, as in Indiana and New Jersey.

Did the Stockton case really turn on one senator voting for himself?

Yes. The March 23, 1866 vote to confirm his seat was 22 to 21, and Stockton's own vote made the majority after Lot Morrill broke a pair and voted against him. Sumner objected on March 26 to a senator voting in his own case. The Senate reconsidered, and on March 27, with Stockton not voting, denied him the seat 23 to 20. The Senate Historical Office page and Haynes agree on every number.

What exactly did the 1866 Act change about the ballot?

It made every vote for senator viva voce, meaning each legislator's choice was called aloud and entered on the journal. States that had used the secret ballot had to stop. Fessenden warned this would expose legislators to party pressure; Clark and Sumner argued constituents had a right to know how their representative voted for senator. Haynes noted that Wisconsin went further in 1899 and required viva voce voting in the party caucus too.

Did the Act set a majority requirement?

Two of them. On the first day each house separately had to give "a majority of the whole number of votes cast in each house" to the same person. In the joint assembly on later days, election required "a majority of all the votes of the said joint assembly, a majority of all the members elected to both houses being present and voting." A plurality never sufficed, which was the direct answer to New Jersey's 1865 rule.

If the Act forced a vote every day, how could a legislature end a session without electing?

Because a vote is not an election. The joint assembly could ballot daily for weeks with no candidate reaching a majority, as members scattered their votes or refused to vote at all to break the quorum. When the session's legal term expired, the legislature went home and the seat stayed empty until the next session or, if the governor appointed someone, until the Senate decided whether to seat him. Lesson 03 has Delaware's 217 ballots over 114 days in 1895.

Hard questions

If Kent was right that the Constitution required a concurrent vote, wasn't every joint-ballot election from 1789 to 1866 unconstitutional, and the 1866 Act too?

That argument was made and lost, and Story, often cited beside Kent, did not actually make it: his section 703 says the Constitution "has not provided for the manner." In the Cameron case of 1857 the Senate Judiciary Committee called the no-concurrent-majority objection "untenable under the statute of Pennsylvania, and 'the uniform practical construction of the Federal Constitution for the last half-century.'" The Senate, as judge of its own elections, had accepted joint-ballot senators for decades. Congress in 1866 then exercised its Article I, Section 4 power over the "Manner" of elections to require a joint ballot as the fallback, and no court ever disturbed it. Kent's reading was coherent as text and defeated by practice, which is a common fate for constitutional theories about procedure.

Doesn't the instruction record show the states had real control over their senators, which the Seventeenth Amendment destroyed?

It shows they had a hold on senators who wanted re-election or who, like Tyler, believed in the doctrine. It does not show control. Bryce's footnote is the settled law: no binding force. Riker's account is that formal instruction faded by mid-century as parties took over, pledging legislative candidates to a Senate candidate before the legislature met, so that by the 1880s the discipline came from the party convention and not the legislature's resolution. The verifiable cases of senators resigning under instruction cluster between 1808 and 1846, with Riker's last near miss in 1854. Whatever the Seventeenth Amendment destroyed in 1913, formal instruction was mostly gone sixty years earlier.

The 1866 Act was pushed through by Radical Republicans right after they unseated a Democrat for partisan advantage. Isn't it fair to call it a partisan power grab dressed as reform?

The partisan context is real and the Senate Historical Office states it: unseating Stockton helped the Radicals toward a veto-proof majority, and the Civil Rights Act veto was overridden 33 to 15 ten days later. Blaine, a Republican, celebrated the Act as a sign of national power displacing state power. But the Act itself is neutral on its face and was supported by a bipartisan 25 to 11; its only sustained opponent was Saulsbury, a Delaware Democrat whose state's later record is the best argument for the law. The provisions, a fixed day, open voting, a majority requirement and a joint fallback, had been proposed in 1814, 1857 and 1858, before the war. The timing was partisan. The content was not new.

Haynes says deadlocks increased after 1866. Couldn't the law have caused that, by forcing daily votes that gave minorities a stage?

Haynes raised the possibility himself: "whether because of the law or in spite of the law." Sherman's 1866 objection supports the causal reading: the mandatory separate first vote told every faction exactly how many votes it had, and the daily joint vote with no bar on other business gave a blocking minority a costless way to hold out. On the other hand, the states that deadlocked worst in the 1890s, Delaware, Kentucky, Pennsylvania, Oregon, were deadlocking over factional fights and money that the law did not create, and pre-1866 legislatures managed multi-year vacancies with no federal procedure at all. The honest answer is that the Act neither caused nor cured deadlocks; it made them follow a script.

Why does any of this procedure matter for the federalism argument today?

Because the claim that legislative election gave states a voice in Washington depends on how the voice was produced. A senator chosen by a party caucus after a pledged legislative election, confirmed by a viva voce joint ballot that the 1866 Act dictated, was the product of state party politics, not of the state government deliberating as a government. Bryce said by 1893 that the election "has in substance almost ceased to be indirect." Whether that undermines the federalism case for repeal, or merely complicates it, is argued in lessons 08 and 09. The procedure is the evidence either way.