Almost every argument conducted about the Civil Rights Act of 1964 vote breakdown is conducted with half the table. One side reports that a larger share of Republicans than Democrats voted for the bill in both chambers, which is true. The other side reports that the bill was signed by a Democratic president and managed on the Senate floor by a Democratic whip against opposition led by Democratic senators, which is also true. Both sides are reading the party columns, and the party columns are the less informative cut of this data.

The regional columns are where the explanation sits. On these four roll calls, whether a member came from the eleven states of the former Confederacy predicted their vote far better than which party they belonged to. Outside the South, support was overwhelming in both parties and the partisan gap was small. Inside the South, opposition was nearly unanimous across party lines. A statistical model of these votes that uses only party performs poorly; a model that uses only region performs well; and a model using both attributes most of the explanatory power to region. That is the region-over-party finding, and it is the single fact that makes the four roll calls intelligible.

This article reports every roll call with its date and chamber, breaks each one by party and by region, names the members whose votes are most often asked about along with the reasons they stated, and stops where the evidence stops. It does not use these numbers to argue about any later party, because the numbers do not support that argument in either direction, and the claims built on them are examined on their own terms in the myths about the 1964 act, which owns those questions.

The four roll calls

H.R. 7152 was introduced in the 88th Congress and cleared four significant recorded votes. Naming them precisely matters, because writers frequently quote one chamber’s numbers as though they were the whole record, and because two of the four are votes on procedure rather than on the bill.

The House passed the bill on February 10, 1964, by 290 to 130. This was passage of the House version, following the discharge petition pressure that dislodged the bill from the Rules Committee and the committee work in Judiciary that produced the text.

The Senate invoked cloture on June 10, 1964, by 71 to 29. This is the procedural vote and, in substantive terms, the decisive one. Under Rule XXII as it then stood, cloture required the votes of two-thirds of senators present and voting, which on a full Senate meant 67. The motion carried with four votes to spare, ending a filibuster that had occupied the chamber for fifty-seven working days and marking the first time the Senate had ever cut off debate on a civil rights measure.

The Senate passed the bill on June 19, 1964, by 73 to 27, nine days after cloture. The gap between the cloture tally and the passage tally is small and is itself informative, as discussed below.

The House agreed to the Senate’s amended text on July 2, 1964, by 289 to 126, sending the bill to the President, who signed it that day. There was no conference committee, which is a procedural fact with consequences: the Senate’s substitute became the enacted text without further negotiation, and the House’s choice was to accept it or lose the bill. The procedural sequence that produced that posture is the subject of how the act passed Congress, which owns the passage narrative.

Two of these four are commonly omitted from popular accounts. The cloture vote is omitted because it is procedural, even though it is the vote that determined the outcome. The July 2 House vote is omitted because it looks like a formality, even though it is the vote on the text that actually became law.

Why the party columns mislead

The party breakdown of these votes is not in dispute and is worth stating in full before explaining why it explains so little.

On House passage in February, Democrats split 152 in favor to 96 against, and Republicans split 138 in favor to 34 against. In percentage terms, roughly 61 percent of Democrats and roughly 80 percent of Republicans voted yes.

On Senate cloture in June, Democrats split 44 to 23 and Republicans split 27 to 6. On Senate passage, Democrats split 46 to 21 and Republicans 27 to 6. In percentage terms the Republican share supporting the bill was again higher than the Democratic share.

On the final House vote in July, Democrats split 153 to 91 and Republicans 136 to 35.

Those numbers support the narrow claim that a higher proportion of Republicans than Democrats voted for this bill in both chambers on every roll call. They do not support the broader inferences usually attached to them, for a reason that is arithmetic rather than interpretive: the Democratic caucus in 1964 contained essentially the entire congressional delegation of the states where opposition was concentrated, and the Republican caucus contained almost none of it. There were ten Republican House members and one Republican senator from the eleven states of the former Confederacy in that Congress. The Democratic percentage is therefore a weighted average of a northern Democratic bloc that supported the bill at rates comparable to northern Republicans and a southern Democratic bloc that opposed it almost unanimously, while the Republican percentage is almost entirely the first kind of member.

Comparing the two party percentages without adjusting for that composition is comparing a mixed group to an unmixed one. It is the same error as comparing average temperatures between two countries with different latitudes and concluding something about their climates policies. The regional cut removes the problem, and once it is applied, the partisan gap outside the South narrows to a few percentage points.

The Civil Rights Act of 1964 vote breakdown: four roll calls, two ways

The table below is this article’s findable artifact. It gives each recorded vote by chamber and date, with the party totals and the regional totals set side by side so both cuts of the same data are visible at once. South means the eleven states of the former Confederacy. Every account that reports only the left half of this table is reporting half the evidence.

Roll call	Date	Result	Democrats	Republicans	Southern members	Members outside the South
House passage of H.R. 7152	February 10, 1964	290 to 130	152 yes, 96 no	138 yes, 34 no	7 yes, 97 no (7 southern Democrats yes, 87 no; no southern Republicans yes, 10 no)	283 yes, 33 no (145 Democrats yes, 9 no; 138 Republicans yes, 24 no)
Senate cloture motion	June 10, 1964	71 to 29	44 yes, 23 no	27 yes, 6 no	1 yes, 21 no	70 yes, 8 no
Senate passage	June 19, 1964	73 to 27	46 yes, 21 no	27 yes, 6 no	1 yes, 21 no (1 southern Democrat yes, 20 no; no southern Republicans yes, 1 no)	72 yes, 6 no (45 Democrats yes, 1 no; 27 Republicans yes, 5 no)
House agreement to Senate text	July 2, 1964	289 to 126	153 yes, 91 no	136 yes, 35 no	Southern opposition again nearly unanimous, with the same small group of supporters	Support again overwhelming in both parties

Read the party columns alone and the story is a modest partisan gap favoring Republicans. Read the regional columns alone and the story is a near-total sectional split. The second reading explains the first; the first cannot explain the second.

The arithmetic that connects them is worth doing once explicitly, using the House vote of February 10. Of the 130 members voting no, 97 came from the eleven southern states, which is roughly three-quarters of the entire opposition drawn from a region holding roughly a quarter of the seats. Of the 96 Democrats voting no, 87 were southern, meaning that northern Democrats voted no nine times in a caucus of 154 voting members from outside the South. Northern Democrats supported the bill at roughly 94 percent and northern Republicans at roughly 85 percent. The partisan gap outside the South runs in the opposite direction to the gap in the national totals, and it is small either way.

The House vote of February 10, 1964

The House vote is the least discussed of the four and contains the clearest demonstration of the regional pattern, because the chamber is large enough that the sample is not dominated by a handful of individuals.

The bill reached the floor after a discharge petition campaign put sufficient pressure on the Rules Committee, chaired by Howard W. Smith of Virginia, to release it. Smith had used the committee’s scheduling power to hold civil rights measures indefinitely in previous Congresses, and the discharge mechanism, rarely successful, was credible enough here to change his calculation. The floor management was bipartisan by necessity: Emanuel Celler of New York chaired Judiciary and managed the bill, and William McCulloch of Ohio, the ranking Republican on that committee, delivered the Republican votes and extracted commitments about what would and would not be traded away in the Senate. Charles Halleck of Indiana, the Republican leader, made the decision to put the party’s institutional weight behind the measure.

The regional result is stark. Members from the eleven southern states voted 7 in favor and 97 against. The seven supporters were Democrats, and they came overwhelmingly from the Texas, Tennessee and Florida delegations, the parts of the South with the largest urban populations, the most economically diversified constituencies and, in several cases, districts where Black registration was already meaningful. All ten southern Republicans voted against.

Outside the South, the picture reverses completely. Democrats from the other thirty-nine states voted 145 to 9. Republicans from those states voted 138 to 24. Both parties delivered large majorities, the Democratic majority was proportionally larger, and the twenty-four Republican no votes came disproportionately from members representing rural districts in the Midwest and Mountain West who framed their objections in terms of federal power over private property.

What does the House vote show that the national party totals hide?

That opposition was sectional rather than partisan. Southern members voted 7 to 97 against the bill regardless of party, while members from the other thirty-nine states voted 283 to 33 in favor. Northern Democrats supported it at a slightly higher rate than northern Republicans, reversing the national gap.

One further feature of the February vote is worth recording because it is so frequently misdescribed. During floor consideration, Howard Smith offered the amendment adding sex to the categories protected by the employment title, and it was adopted. Whether that amendment was intended to defeat the bill, to advance an aim Smith held for other reasons, or both, is contested, and the claim that it was purely a wrecking maneuver is examined in the myths article rather than here. What belongs in a vote analysis is narrower: the amendment was adopted, the bill passed with it, and the sex provision has been operative law ever since.

The cloture vote of June 10, 1964

The cloture vote is the one that decided the outcome, and it is the one most often omitted from popular accounts because it is procedural. A reader who understands only one of the four roll calls should understand this one.

Under Rule XXII as it stood in 1964, ending debate required an affirmative vote of two-thirds of senators present and voting. With all one hundred senators voting, the threshold was 67. The Senate had never once reached it on a civil rights measure. Every previous attempt, across decades, had failed, and the record of those failures was the reason the bill’s managers treated the vote count rather than the merits as the operative problem. The mechanics of the rule, how the threshold changed in 1975, and what a cloture motion does and does not accomplish are covered in the filibuster and cloture explained, which owns the procedural questions for this series.

The motion carried 71 to 29. Democrats voted 44 to 23 and Republicans voted 27 to 6. Members from the eleven southern states voted 1 in favor and 21 against. Members from everywhere else voted 70 to 8.

Two arithmetic points follow directly and are the whole reason the coalition had the shape it did. First, the Democratic caucus could not deliver 67 votes by itself. With roughly two dozen of its own members committed against, the majority party’s supporters numbered in the mid-forties, which is well short. Second, Republican support was therefore not a courtesy but a mathematical requirement, and the number required was large enough that it could not be assembled from the party’s most liberal members alone.

The person who delivered them was Everett Dirksen of Illinois, the Republican leader, and the price was a substitute text. The Dirksen substitute narrowed several enforcement provisions, most consequentially by limiting federal intervention in employment and public accommodations cases in states with their own enforcement agencies and by adjusting the pattern-or-practice authority given to the Attorney General. What the substitute traded away, and why the bill’s managers accepted it, is analyzed in the passage history article. For vote arithmetic the relevant fact is that the substitute converted a bloc of Republican senators from uncertain to committed, and that bloc is the difference between 71 and a number below 67.

The opposition was organized rather than spontaneous. Richard Russell of Georgia led the southern caucus of senators who had coordinated against civil rights legislation for years, and their strategy was explicitly procedural: hold the floor, force the majority to prove it had 67 votes, and rely on the historical record that it never had. Russell’s own assessment afterward, that his group had been beaten by the arithmetic rather than out-argued, is a fair summary of what the roll call shows.

Why is the cloture vote more important than the passage vote?

Because the passage vote was never in doubt once debate ended. A simple majority for the bill existed throughout. The obstacle was Rule XXII’s two-thirds requirement to end debate, which no civil rights measure had ever cleared, so the cloture tally is the vote that determined whether a final vote would occur at all.

The scene of the vote has entered the general narrative in ways that are accurate but incomplete. Clair Engle of California, gravely ill and unable to speak, was brought into the chamber and signaled his vote by pointing to his eye. Robert Byrd of West Virginia concluded a speech of more than fourteen hours that morning. Both details are true and both are frequently used to substitute for the arithmetic, which is the more useful thing to carry away.

A closer look at the twenty-nine who voted against cloture

The cloture opposition is small enough to describe precisely, and describing it removes most of the ambiguity that surrounds the more familiar passage vote.

Twenty-nine senators voted against the motion on June 10. Twenty-one came from the eleven southern states: twenty Democrats and one Republican, John Tower of Texas. Eight came from elsewhere: three Democrats and five Republicans.

The composition of that group of eight is the analytically interesting part, because it is the entire non-southern opposition to ending debate on the most heavily debated bill of the era. Five of the eight were the Republican senators whose stated constitutional objections are described above, and their positions were confined to the public accommodations and employment titles rather than extending to the measure as a whole. The three Democrats among them came from outside the eleven-state bloc, and their stated grounds combined objections to the enforcement provisions with, in at least one case, an institutional argument about the use of cloture itself.

Set beside the numbers the southern caucus had assembled in previous decades, this group is the story of the vote. The bloc had never needed to persuade a majority of the Senate. It had needed to hold roughly twenty of its own and recruit fourteen more, and in June 1964 the recruitment produced eight. That is a shortfall of six votes against a threshold that had never been reached, and it is the entire margin between a statute and another failed attempt.

One further arithmetic point closes the account of the cloture margin. Seventy-one votes against a threshold of sixty-seven is a cushion of four, which sounds comfortable and was not. Two of those four came from senators whose commitments were secured in the final days of the substitute negotiation, and the count maintained by the bill’s managers had shown the motion at or below the threshold for most of the preceding month. A margin measured after the fact tells a reader how the vote ended; it says nothing about how close the outcome stood while it was still in doubt, which in this instance was close enough that the managers scheduled the vote only once the count held above sixty-seven for several consecutive days.

Two further points keep this from being read as a simple story of resolve. A senator voting against cloture is not necessarily voting against the bill, as the two-vote difference between the cloture and passage tallies shows. And the eight non-southern opponents were not part of the organized floor campaign; they were members who declined to end debate for reasons they stated, on a bill most of them had not fought.

The Senate passage vote of June 19, 1964

Nine days after cloture, the Senate passed the bill 73 to 27. Democrats split 46 to 21, Republicans 27 to 6.

The regional cut is close to absolute. Of the twenty-two senators from the eleven southern states, one voted yes: Ralph Yarborough of Texas. Twenty southern Democrats and the region’s single Republican senator, John Tower of Texas, voted no. Outside the South, senators voted 72 to 6.

The difference between the cloture tally of 71 and the passage tally of 73 is small and is worth explaining, because it is sometimes read as evidence of a shift. It is not. A senator may vote against cloture on institutional grounds, believing debate should not be cut off, and then vote for the bill once debate has in fact ended. That combination is a recognized position rather than an inconsistency, and it accounts for the two-vote movement. The reverse combination, voting for cloture and against the bill, is rarer and generally reflects a judgment that the chamber should reach a decision even by a member who intends to vote no.

The single northern Democratic vote against passage came from Robert Byrd of West Virginia, and five northern Republicans joined the opposition. Naming them is the subject of the next section, because the reasons they stated are the part of this record most often flattened into a single explanation.

The House agreement vote of July 2, 1964

The Senate had amended the House bill substantially through the Dirksen substitute, which meant the House had to act again. The ordinary route would have been a conference committee to reconcile the two versions. The bill’s managers avoided one, because a conference would have required the Senate to appoint conferees and to vote again on the conference report, exposing the bill to a second filibuster after the first had been broken at maximum political cost.

The House therefore voted on July 2 to agree to the Senate text as it stood, 289 to 126, with Democrats splitting 153 to 91 and Republicans 136 to 35. The regional pattern held: southern opposition remained close to unanimous, and support outside the region remained overwhelming in both parties.

Two observations belong with this vote. The first is that it is the vote on the text that became law. The February vote was on a different document, and a reader comparing the act’s provisions to the February debate is comparing to a superseded version. The second is that the choice presented to House members was structurally constrained: accept the Senate’s narrower enforcement provisions or lose the bill entirely. Several members who had preferred the House text voted yes on that basis and said so, which is a normal feature of endgame legislating and is worth remembering when a final tally is read as a measure of enthusiasm.

President Johnson signed the bill the same day.

The named opponents and the reasons they stated

Reporting a no vote without the reason the member gave is incomplete, and inferring motives beyond the stated reasons is not something the roll call supports. What follows is what members said, attributed to them, with no assessment of sincerity.

The six Republican senators voting against passage were Barry Goldwater of Arizona, John Tower of Texas, Bourke Hickenlooper of Iowa, Edwin Mechem of New Mexico, Milward Simpson of Wyoming, and Norris Cotton of New Hampshire.

Goldwater’s is the vote most asked about, and his stated position was specific. He had supported the Civil Rights Acts of 1957 and 1960, had supported civil rights measures in Arizona, and stated that he agreed with the aims of most of the 1964 bill. His objection was to two titles: the public accommodations title and the employment title, which he argued exceeded federal constitutional authority over private conduct and would require an enforcement apparatus intruding into private business decisions. He said he had sought legal advice on the constitutional question before deciding. His floor statement made the point that he was voting against a bill whose objectives he shared because he believed two of its titles could not be squared with the Constitution as he read it. The Supreme Court rejected that constitutional reading within six months in Heart of Atlanta Motel v. United States, 379 U.S. 241 (1964), and Katzenbach v. McClung, 379 U.S. 294 (1964), a sequence traced in the cases that construed the act.

Whether that stated reasoning was the operative one, and what the vote meant for the presidential campaign Goldwater was then conducting, are questions this article does not answer, because a roll call cannot answer them. What the record supports is the content of the argument he made and the fact that the constitutional objection he raised was tested in court and rejected.

Among Democratic opponents, Sam Ervin of North Carolina made the most sustained constitutional case, arguing across many hours of floor time that the commerce power could not reach the conduct the bill regulated and that the enforcement provisions displaced state authority in areas the Constitution reserved. Richard Russell of Georgia organized the opposition as a coordinated procedural campaign and argued principally in terms of federalism and property rights rather than defending segregation as such in floor debate, a rhetorical choice that historians of the period have examined at length. J. William Fulbright of Arkansas and Albert Gore of Tennessee also voted no, and their positions are frequently cited in discussions of southern members whose records on other questions diverged from the regional pattern.

The five northern Republican no votes carried different arguments. Hickenlooper’s objections centered on the reach of federal enforcement authority and the treatment of jury trials in contempt proceedings, an issue that had been contested in the 1957 act as well. Cotton, Simpson and Mechem framed their positions around federal power over private property and small business. None of the five was among the bill’s floor opponents in the organized sense; they voted no on a bill they had not fought.

The single northern Democratic no vote, Robert Byrd of West Virginia, came from a senator who had participated in the filibuster and whose later record on civil rights measures differed substantially from his position in 1964. Recording both facts is the neutral treatment: the 1964 vote is what it is, and a member’s later record is a separate fact that neither excuses nor aggravates it.

On the other side, the small group of southern members voting yes deserves the same attention. Ralph Yarborough of Texas cast the only southern yes vote in the Senate and faced sustained political consequences for it in his state. In the House, the handful of southern supporters came from urban and economically diversified districts, principally in Texas, Tennessee and Florida, and several of them stated their positions in terms of their districts’ commercial interests as much as in terms of the bill’s aims.

The delegations, described

Aggregate regional figures conceal how the delegations were actually composed, and a short description makes the composition effect concrete rather than abstract.

In 1964 the eleven former Confederate states sent roughly one hundred and six members to the House, of whom the overwhelming majority were Democrats; the Republican contingent from those states numbered around ten and was concentrated in a few districts, several of them won only in the preceding two elections. In the Senate the same eleven states sent twenty-two members, of whom twenty-one were Democrats and one, John Tower of Texas, was a Republican elected in a special election three years earlier.

That distribution is the whole of the composition effect. A party whose caucus includes essentially an entire region will show that region’s preferences in its national totals. A party with ten House members and one senator from the same region will not. When the region divides sharply from the rest of the country on a question, the two parties’ national percentages will diverge for reasons that have nothing to do with the parties’ positions outside the region.

Outside the eleven states the two delegations were both large and both geographically diverse. Democrats outside the region held roughly one hundred and fifty-four voting members in the February House vote and Republicans roughly one hundred and sixty-two, which is close to parity, and their support rates were 94 percent and 85 percent respectively. Those are the numbers a reader should compare if the question is about the parties rather than about the region, and they show a real but modest gap in the opposite direction from the aggregate figures.

The Senate outside the region divided forty-six Democrats and thirty-two Republicans on final passage, with support rates of roughly 98 percent and 84 percent. The Republican shortfall here is entirely accounted for by the five members whose stated constitutional objections are described above, which is a small enough group that naming them individually is more informative than any percentage.

Two implications follow. The first is that the aggregate party gap of roughly nineteen points in the February House vote and thirteen points on Senate final passage is generated almost entirely by ninety-four southern Democrats and ten southern Republicans, and would disappear or reverse if the two parties had held similar regional delegations. The second is that neither party outside the region was unanimous, and the members who broke from their delegations in either direction are individually identifiable, which is the level at which claims about motives can be examined against evidence.

The coalition that assembled

A roll call records an outcome; it does not record how the outcome was built. Four groups had to hold simultaneously, and the failure of any one of them would have left the cloture motion short.

Northern and western Democrats supplied the largest bloc and the least uncertainty. They voted for the bill at rates above ninety percent in the House and lost only one member in the Senate. This bloc was not the constraint.

Northern and western Republicans supplied the votes that made 67 arithmetically reachable, and they were the constraint. Their support was conditional on a text they could defend to constituents skeptical of federal enforcement authority, which is what the Dirksen substitute provided. Thomas Kuchel of California served as the Republican floor manager alongside Hubert Humphrey, and the pairing was deliberate: a bill managed by one party in a chamber requiring two-thirds would have signaled that the votes were not there.

The Senate leadership of both parties committed institutionally. Mike Mansfield, the majority leader, made the procedural decision to bypass the Judiciary Committee, chaired by James Eastland of Mississippi, by holding the bill at the desk, which denied the opposition its most reliable choke point. Dirksen made the corresponding decision on the Republican side and absorbed the cost of delivering members who had reservations.

The executive branch applied sustained pressure. President Johnson’s involvement is well documented and is treated in the passage history article, and the relevant point for vote analysis is that the pressure was directed at specific undecided senators rather than at the public, because the constraint was a vote count in a chamber of one hundred.

Outside government, the Leadership Conference on Civil Rights coordinated organizational pressure, and religious denominations mobilized in states with few Black constituents, which mattered specifically because the votes still needed were from senators whose states had small Black populations. The targeting of that effort at Mountain West and Plains senators is one of the better documented instances of an advocacy campaign designed around a specific arithmetic problem rather than around general persuasion.

The region-over-party finding stated formally

The claim this article advances can be stated in a form specific enough to be checked against the table.

Take the House vote of February 10, 1964. Knowing only a member’s party, a predictor would guess yes for Republicans and yes for Democrats, since both delivered majorities, and would misclassify 130 of the 420 members voting. Knowing only whether a member came from the eleven southern states, a predictor would guess no for southern members and yes for everyone else, and would misclassify 40, being the 7 southern yes votes and the 33 non-southern no votes. Region is roughly three times more accurate than party on this vote, using the same amount of information.

The same exercise on Senate passage gives a starker result. Region misclassifies 7 of the 100 senators voting. Party misclassifies 27, since party alone predicts yes for both caucuses.

That is what region-over-party means operationally, and it is why any account reporting only the party columns is reporting the weaker predictor. The finding does not depend on a preferred definition of the South, either. Adding the border states of Kentucky, Oklahoma, West Virginia, Maryland, Delaware and Missouri to the southern group weakens the pattern somewhat, because those delegations divided, and the regional predictor still outperforms party by a wide margin.

The finding also has a straightforward historical explanation that no one disputes. The one-party South of the period meant that essentially every member from those states was a Democrat, elected in an electorate from which most Black citizens had been excluded, and their positions on this question were established long before 1964 and were coordinated. The party affiliation of a southern member in that period carries almost no information about their views, because there was no alternative affiliation available to them. Reading a national party percentage that includes this bloc as though it measured party positions is therefore a compositional error, not a matter of interpretation.

Among members from the eleven states, Democratic support was 7 of 94, and Republican support was 0 of 10. The gap between the parties within that region is small in percentage terms and rests on a Republican delegation too small to support any inference.

Now compare across regions within each party. Democratic support falls from roughly 94 percent outside the region to roughly 7 percent inside it, a gap of about eighty-seven points. Republican support falls from roughly 85 percent to zero.

The comparison is the finding. Within-region party differences are around ten points or less. Across-region same-party differences are eighty points or more. When a variable produces an eighty-point spread and another produces a ten-point spread on the same data, the first is the explanation and the second is a detail, and reporting only the second is what the party columns do.

The Senate votes produce the same structure with smaller numbers. Outside the region, Democratic support on final passage was 45 of 46 and Republican support was 27 of 32. Inside the region, Democratic support was 1 of 21 and Republican support was 0 of 1. Same pattern, same conclusion.

This is what statisticians call a composition effect, and it is not a subtle one. It is worth naming plainly because the arithmetic is accessible to any reader and because both sides of the public argument about these votes have an incentive to quote the aggregate that favors them. The cross-tabulation removes the incentive by removing the ambiguity.

The border states, and why the definition of the South matters

Every regional analysis depends on where the line is drawn, and a reader is entitled to know how sensitive the finding is to that choice.

This article uses the eleven states of the former Confederacy, which is the standard convention in congressional voting analysis for this period and the one that produces the cleanest match to the political structure that organized the opposition. Six states are commonly treated as border states and excluded from that grouping: Kentucky, Oklahoma, West Virginia, Maryland, Delaware and Missouri.

Those delegations did not behave as a bloc. They divided on the bill, in both chambers and in both parties, in a pattern closer to the rest of the country than to the eleven states, though with a larger share of no votes than the northern average. That mixed behavior is itself informative, because it indicates that the near-unanimity of the eleven-state bloc was a property of that specific political system rather than a general southern or border-state disposition.

Moving the six border states into the southern group weakens the regional finding without reversing it. The southern support rate rises, because these delegations supplied yes votes, and the non-southern support rate rises as well, because the no votes move out of it. The gap between the two regions narrows and remains very large, and the regional predictor continues to outperform the party predictor by a wide margin on every roll call.

The practical rule for a reader evaluating someone else’s numbers is to ask which definition produced them before comparing to another source. Two writers using the eleven-state and the seventeen-state conventions will report different regional percentages from identical roll calls, and neither is wrong. A writer who does not state the convention has produced a figure that cannot be checked, which is a common enough problem in this material that it belongs on the list of recurring errors above.

What the votes did and did not predict

The most contested question attached to these roll calls is what they show about the later trajectory of the two parties. The evidence supports some claims here and not others, and separating them is the whole task.

What the record supports is that the coalitions on this bill were not the party coalitions of the following decades. Both parties contained substantial pro-civil-rights majorities outside the South in 1964. Both parties contained members who opposed the bill. The bill could not have passed without votes from both caucuses, and its managers on the floor came from both.

What the record also supports is that the regional bloc that opposed the bill was, at the time, entirely within one party, and that the same bloc’s congressional delegations changed party affiliation over the following decades through a process that took roughly a generation and proceeded at different rates in different states and at different levels of office. That the change happened is not disputed. The mechanism, the timing, and the relative weight of racial politics, economic development, in-migration, religious realignment and generational replacement in producing it are the subject of a genuine and continuing scholarly disagreement, and this article does not resolve it.

What the record does not support is either of the two claims most often built on these numbers in popular argument. It does not support the claim that the roll call demonstrates the enduring civil rights record of one party against the other, because the roll call’s own structure shows the division running through both parties along regional lines rather than between them. And it does not support the claim that the numbers are meaningless because the parties later exchanged positions, because the votes are a factual record of who supported the bill in 1964 and that record does not become false when later coalitions change. Both claims are examined directly in the myths article.

The disciplined position is that a roll call is evidence about the votes it records. Using it as evidence about political coalitions forty years later requires an argument about intervening causes, and that argument has to be made and defended on its own evidence rather than smuggled in through a tally.

Why the southern bloc had been unbeatable before 1964

Understanding what changed requires understanding the mechanism that had worked, and it was arithmetic rather than persuasion.

A blocking coalition under a two-thirds cloture rule needs thirty-four votes when all senators are present. The southern bloc supplied roughly twenty. It therefore needed to attract about fourteen additional senators to any given cloture vote, and it had been able to do so consistently for two decades from three sources: senators with constitutional objections to particular federal enforcement mechanisms, senators from western and mountain states who valued the filibuster as an institutional protection for small-state interests and voted against cloture on principle regardless of the underlying bill, and senators who preferred a narrowed bill and used the threat of a failed cloture vote as leverage.

Each of those three sources shrank in 1964. The constitutional objectors remained but were fewer, and the Dirksen substitute addressed several of their concerns directly by deferring to state agencies in the first instance. The institutional filibuster defenders were reduced by a sustained argument, made publicly by Dirksen among others, that the rule’s protective function did not extend to indefinite obstruction of a measure with clear majority support in both chambers. And the leverage-seekers were bought out by the substitute before the cloture vote rather than after it, which converted potential no votes into yes votes at a price paid in statutory text.

The result was that the bloc held its twenty-one votes and could attract only eight more, which is twenty-nine, five short of a blocking third. Nothing about the bloc’s discipline or determination changed. The additional votes it had always relied on stopped being available, and the mechanism failed the first time that happened.

This is the most transferable lesson in the article for readers interested in legislative procedure generally. A supermajority requirement does not give a minority a veto. It gives a minority a veto conditional on its ability to recruit, and the recruiting pool is where the real contest happens. Anyone modeling a filibuster fight by counting the committed opposition is counting the wrong number.

The precedent votes: 1957 and 1960

The 1964 roll calls look different when set against the two civil rights measures that preceded them, because those votes are the record the southern bloc was relying on and the record the bill’s managers were trying to break.

The Civil Rights Act of 1957 passed the Senate by 72 to 18 and the House by a comparable margin, and it reached those numbers by being narrowed until the opposition could accept it. Its enforcement provisions were weakened during floor consideration, most consequentially by an amendment guaranteeing jury trials in certain criminal contempt proceedings, which in the jurisdictions at issue meant that enforcement actions would be tried before juries drawn from electorates that largely excluded the people the statute protected. Strom Thurmond of South Carolina held the floor alone for more than twenty-four hours against it, which remains the longest individual speaking filibuster in the Senate’s records, and the bill passed anyway because it had been reduced to something the bloc could survive.

The Civil Rights Act of 1960 followed a similar pattern: extended obstruction, a negotiated narrowing, and passage of a measure that added incremental voting protections without the enforcement machinery its supporters wanted.

The lesson the southern caucus drew from both was that the Senate would always trade text for time, and the lesson the 1964 managers drew was the same fact read in reverse. A bill that could be narrowed to the point of acceptance would be, and the only way to keep the enforcement provisions in the enrolled text was to break the filibuster rather than to bargain it down to nothing. That is why the operation described above was organized around a cloture count from the beginning rather than around the merits, and it is why the Dirksen substitute is best understood as a deliberate, bounded concession made in exchange for the threshold rather than as the same open-ended erosion that had produced the 1957 and 1960 statutes.

The comparison also explains a feature of the 1964 tallies that otherwise looks strange. Large margins on final passage are compatible with a bill that nearly died, because in a chamber governed by a supermajority threshold the contest happens before the merits vote and the merits vote records a settled question. Anyone reading 73 to 27 as evidence of easy passage has read the wrong number.

Why did the 1964 bill not follow the 1957 and 1960 pattern?

Because its managers refused the trade those bills had accepted. In 1957 and 1960 the Senate ended obstruction by weakening enforcement until opponents allowed a vote. In 1964 the leadership organized instead around assembling sixty-seven votes for cloture, and paid a bounded price in the Dirksen substitute rather than an open-ended one.

The discharge petition and the arithmetic in the House

The Senate filibuster receives nearly all the attention, and the bill’s first near-death was in the House, where the mechanism was a committee chairmanship rather than a debate rule.

Howard W. Smith of Virginia chaired the Rules Committee, which controls the terms on which legislation reaches the House floor. A chairman who declines to schedule a bill can hold it indefinitely, and Smith had done so with civil rights measures before. The countermeasure available under House rules is a discharge petition, which removes a bill from a committee once a majority of the membership, 218 members, sign it.

Discharge petitions rarely succeed, because signing one is a public act against the committee system that members of both parties have reasons to protect, and because leadership generally discourages them. What made this one credible was the combination of public attention following the assassination of President Kennedy, the willingness of the Republican leadership to allow its members to sign, and the fact that the signature count is visible, which converts a petition into a running vote count that a chairman can read as easily as anyone else.

The petition did not reach 218. It did not have to. Smith agreed to hold hearings and to report the bill once the signature count made clear that the votes to bypass him existed, which is the ordinary way such petitions work: the threat is exercised through the count rather than through the completed procedure. The committee reported the bill, the House took it up in late January, and floor consideration produced the February 10 vote.

Two observations belong with this sequence. The first is that the House obstacle was structural in the same way the Senate obstacle was: a rule concentrating power in a member from a safe seat in the region most opposed to the bill, which is a feature of the seniority system rather than an accident. The second is that the House arithmetic and the Senate arithmetic differed in kind. The House needed a majority and a procedural path to a vote. The Senate needed a supermajority. A bill can be popular enough to win the first and still lose the second, and the record before 1964 is largely a record of exactly that.

The other recorded votes that shaped the outcome

Four roll calls decided the bill, and several dozen more shaped what the bill was by the time it was decided. A vote-analysis article that reports only the passage votes reports the conclusion without the argument.

In the House, floor consideration ran ten days in late January and early February 1964, and more than one hundred and twenty amendments were offered. Most came from opponents and most failed. Their purposes clustered into four categories: narrowing the definition of a covered establishment or employer, adding procedural requirements before enforcement could begin, transferring authority from federal agencies to the states or to juries, and inserting exemptions for particular industries or institutions. Each amendment that came to a recorded vote produced a smaller-scale version of the same regional pattern visible in the passage vote, which is why the passage roll call is a good summary of the chamber rather than a snapshot of one moment.

Two House amendments did carry and both matter. The addition of sex to the employment title’s protected characteristics was offered by Rules Committee Chairman Howard W. Smith and adopted. What Smith intended by offering it is one of the most argued questions in the act’s history and is handled in the myths article, which owns the contested-claims territory. The roll-call fact is narrow: the amendment was adopted, and the bill passed two days later by a margin that gives no indication the amendment cost it support.

In the Senate the picture is different because the filibuster meant that for most of the spring there was nothing to vote on. Amendments were offered in volume after cloture, when the rule limited each senator to one hour of further debate and the chamber worked through a backlog of pending proposals in a compressed period. The great majority failed. The one that mattered had already been agreed before the cloture vote: the substitute text negotiated with Everett Dirksen, which the Senate adopted as the vehicle for final passage.

There were also procedural votes that never became famous and were decisive. The motion to take up the bill directly rather than refer it to the Judiciary Committee was itself subject to extended debate, and losing it would have sent H.R. 7152 to a committee whose chairman had buried previous civil rights measures. The motion to proceed and the cloture motion are the two votes where the bill could have died on procedure while retaining majority support, which is the pattern this series returns to repeatedly: majorities lose to thresholds.

Absences, pairs and the members who did not vote

Roll-call totals do not always sum to the chamber’s membership, and readers comparing sources will encounter small discrepancies traceable to how absences are handled.

The House had four hundred and thirty-five seats. The February vote recorded 290 yes and 130 nay, which accounts for four hundred and twenty members, with the remainder absent, not voting, or vacant. The July vote recorded 289 and 126, a total of four hundred and fifteen. Neither figure indicates anything about the missing members’ positions, and treating an absence as a silent no is an error that appears in secondary writing regularly.

The Senate is cleaner. All one hundred senators were present for the cloture vote on June 10, which is unusual and reflects how the vote was understood at the time. Final passage on June 19 recorded one hundred votes as well.

Two conventions also affect published figures. Live pairs, in which an absent member’s position is recorded in agreement with a present member who withholds a vote, appear in the Congressional Record and not in the raw tally, so a source drawing on the Record may report a member’s position that the roll call itself does not contain. And announced positions, in which a member states how they would have voted, carry no procedural weight and are sometimes folded into secondary summaries.

For an article that reports counts, the rule is to use the recorded roll call and to say so, which is what the table above does. For a researcher tracing an individual member, the Congressional Record entry for the date carries the pairs and announcements the roll call omits, and the technique for locating it belongs to how to find legislative history.

What members said, on both sides

Reporting stated positions rather than assigning motives is the standard this series applies, and the floor record for this bill is unusually full because the Senate debate ran for months.

Supporters argued on several distinct grounds, and the differences among them matter. Hubert Humphrey, managing the bill, argued primarily from the record of failure under existing law, contending that the state and private mechanisms available had been given a decade after Brown and had produced minimal change. Everett Dirksen argued from a different premise, framing his support around the proposition that the measure had become inevitable and that the responsible course was to shape it rather than to obstruct it, a position he stated publicly and which drew criticism from members of his own conference. Republican supporters including William McCulloch, the ranking member on House Judiciary whose insistence on maintaining the bill’s core was a condition of Republican support throughout, argued from party history on civil rights measures reaching back to Reconstruction. Northern Democratic supporters argued from constituency and from the party’s national platform commitments.

Opponents also argued on distinct grounds, and collapsing them into one position misdescribes the record. The southern bloc’s stated case rested principally on federalism and on the constitutional limits of federal authority over private conduct and over state institutions, and secondarily on the argument that the public accommodations title took property rights without compensation. Richard Russell stated repeatedly that the bloc’s objection was to federal compulsion rather than to the objective, a characterization his opponents disputed at the time and which a reader is entitled to weigh against the bloc’s record on earlier measures. Non-southern opponents, principally the five Republican senators discussed above, stated constitutional objections confined to two titles while supporting the remainder.

Reporting these positions is not endorsing them, and a reader should notice that the stated grounds and the regional pattern of the votes are in tension: a constitutional objection to federal authority over private conduct is not a regionally distributed belief, yet the votes were regionally distributed to an overwhelming degree. Stating that tension is fair. Resolving it by asserting what members privately believed is not something a roll call can support, and this article declines to do it.

The substitute that bought the votes

The cloture coalition was assembled by trading statutory text for votes, and identifying what was traded is part of a complete vote analysis, because the price appears in the law rather than in the tally.

The negotiation ran through May 1964 among Dirksen, Humphrey, Kuchel, Mansfield and the Justice Department, and it produced a substitute text that the Senate adopted as the vehicle for final passage. Its changes clustered in three areas.

In the public accommodations title, the substitute gave state and local agencies a first opportunity to resolve complaints where a state or locality had its own public accommodations law and an agency to enforce it, deferring federal court proceedings for a period. It also narrowed the Attorney General’s authority to initiate suits, confining it more closely to cases showing a pattern or practice of resistance rather than to individual complaints.

In the employment title, the substitute made a parallel change, deferring in the first instance to state fair employment practice agencies where they existed, and it constrained the new enforcement commission’s role. The commission emerged with investigative and conciliation authority and without power to sue, a design whose consequences dominate the statute’s first decade and are traced in the Title VII charge process from filing to court.

Across both titles, the substitute added language limiting the reach of the prohibitions in ways the sponsors accepted as the cost of proceeding, including the provision stating that nothing in the employment title requires preferential treatment to correct an imbalance in a workforce.

The trade is visible in the arithmetic. Twenty-seven Republican senators voted for cloture. Supporters had estimated they needed about twenty-five. The substitute converted a coalition that was short of the threshold into one that exceeded it by four, and the price was an enforcement design that Congress spent the next twenty-seven years amending. That sequence, from procedural threshold to weakened enforcement to legislative correction, is set out in the amendments that rebuilt the statute.

Two observations follow for a reader interested in votes rather than in doctrine. The first is that a roll call showing a comfortable margin can conceal a bill that was substantially rewritten to obtain that margin, so margin is a poor measure of how contested a measure was. The second is that the members whose votes were purchased are identifiable, since they are the Republicans who voted for cloture after having stated objections to the original text, and the price they extracted is readable in the enrolled statute.

The vote against the constituencies

A roll-call analysis that never mentions constituencies is incomplete, and one that explains every vote by constituency is deterministic. The defensible middle is to describe the constituency pattern and to note where votes departed from it.

The eleven states in question had, in 1964, Black populations ranging from roughly a fifth to more than a third of total population, and Black registration rates that were low and in some states very low, a condition that the Voting Rights Act of 1965 was written to address and whose measured change is traced in the voting cluster. The practical consequence for this vote is that members from those states faced electorates in which a large share of the population most directly affected by the bill could not vote, while the portion that could vote was, in many districts, organized in opposition. That is a structural fact about the constituencies rather than a claim about any member’s beliefs, and it is documented in registration data from the period.

Outside the region, the constituency picture was different in kind. Northern and western districts contained Black populations that were smaller in share and, critically, enfranchised, along with organized labor, religious and civic organizations that had campaigned for the bill. Members from those districts faced pressure in favor and comparatively little organized pressure against.

Two implications, and both cut against simple readings. A yes vote from a northern member carried little electoral risk, and treating it as equivalent in cost to a yes vote from a Texas or Tennessee district would be a mistake. And a no vote from a member representing a district where a large share of affected residents could not vote is a vote cast in a constituency structure that the same Congress moved to change the following year, which is a fact worth stating and which does not by itself excuse or condemn anyone.

The seven southern Democrats who voted yes in February represented districts that differed from the regional pattern, and the fact that they existed at all establishes that the regional vote was overwhelming rather than compelled.

What the tallies meant for the members who cast them

A roll call is a public act with consequences, and the consequences fell very unevenly across the members who voted.

Ralph Yarborough of Texas cast the only southern vote for the bill in the Senate and faced sustained opposition in his state afterward, in a political environment where the vote was used against him directly. The small group of southern House members who voted yes represented districts that could absorb the position, principally urban and economically diversified constituencies in Texas, Tennessee and Florida, and even there the vote was consequential.

For members outside the region the calculation ran the other way. A yes vote carried little electoral risk in most northern and western districts, where organized pressure ran in favor of the bill and against it only sporadically. Treating a northern yes vote as equivalent in cost to a southern one misdescribes what each member was doing, and any account that assigns equal credit to both is measuring the vote rather than the choice.

The five non-southern Republican senators who voted no had a different problem. Their stated objections were confined to two titles, and they were voting against a measure their party’s leadership had negotiated and delivered, which is an uncomfortable position in either direction and is one reason their statements are unusually detailed in the record.

For the southern bloc, the vote was the least costly of the four positions available in their constituencies and the most consistent with two decades of coordinated action. Recording that is not a claim about any individual member’s beliefs. It is a statement about the structure of the choice, and it is the same structural observation that applies to every member in the chamber, which is why an honest vote analysis describes incentives for everyone or for no one.

The vote-counting operation behind the tallies

A roll call is the output of a counting operation that runs for months before it, and the 1964 operation is documented well enough to describe.

On the Senate side the count was maintained jointly by Humphrey’s staff for the Democrats and Kuchel’s for the Republicans, with the bill’s supporters tracking each senator’s position on cloture separately from their position on passage, because the two were not the same question and several members were reliable on one and uncertain on the other. That distinction is the reason the operation focused on cloture from the beginning: passage was never in doubt and cloture was, so the entire effort was organized around a threshold rather than around a majority.

The count was organized by state delegation and by the specific objections each uncertain member had raised, which is what made the substitute negotiation possible. A leadership that knows precisely which four senators are held back by state-agency deferral and which three by the Attorney General’s authority can draft an amendment that addresses those objections and no others, which is cheaper in statutory terms than a general weakening. The Dirksen substitute reads like what it was: a set of targeted concessions rather than a broad retreat.

On the House side the counting operation had run since the previous autumn and was tied to the discharge petition, since a signature on the petition is a public commitment that functions as a vote count. Ranking Republican William McCulloch’s role is central here and is frequently omitted from popular accounts. His condition for Republican support was that the bill’s core not be traded away in the Senate without House Republicans being consulted, and that commitment is one reason the July vote on accepting the Senate text held together as well as it did. A negotiated Senate text that House Republicans regarded as a betrayal would have shown up as defections in the final tally, and it did not: Republican support in July was 136, against 138 in February.

The general point is that the stability of the four tallies is evidence of a functioning count rather than of members having fixed views independent of the process. Positions were identified early, objections were addressed specifically, and the recorded votes came in close to where the counters expected. A vote analysis that treats roll calls as spontaneous expressions of member preference misses the operation that produced them.

The comparison that sharpens the pattern: the 1965 vote

The Voting Rights Act of 1965 came to the floor thirteen months later, addressed a related problem, and produced different numbers. Setting the two side by side isolates what was distinctive about 1964.

The Senate passed the voting rights bill 77 to 19 and the House passed it 333 to 85, both by wider margins than the corresponding 1964 votes. Cloture was invoked again, and the threshold was met more comfortably. The southern bloc opposed again, though not quite as uniformly, and support outside the region was again overwhelming in both parties. The composition and the defections on that bill are analyzed in the coalition behind the Voting Rights Act, which owns the 1965 vote.

Three differences explain most of the gap. The 1964 act regulated private conduct, including the practices of businesses and employers, which is where the constitutional objections and the property-rights arguments were concentrated. The 1965 act regulated state and local election administration, which sits squarely within the Fifteenth Amendment’s enforcement power and generated far weaker constitutional objections. The events at Selma in March 1965 changed the political environment in a way the 1964 debate had no equivalent for. And the 1964 act had already been passed, survived, and produced compliance rather than the disorder its opponents had predicted, which removed one of the strongest arguments available in the earlier debate.

The comparison matters for how a reader interprets the 1964 numbers. A bill regulating private commerce drew constitutional objections from members who had supported prior civil rights legislation and would support the next one. A bill regulating state election practices did not. Treating the 1964 no votes as a single undifferentiated bloc obscures a real distinction between members organizing to defeat civil rights legislation as a category and members objecting to two specific titles on grounds they had stated before and would state again.

Two contrasts that show what the 1964 pattern depends on

Setting the 1964 tallies beside other major enactments clarifies what was and was not unusual about them.

Against the Social Security Amendments of 1965 establishing Medicare, the structure is entirely different: opposition was partisan rather than regional, distributed across the country, and organized around the role of the federal government in health financing rather than around region. That contrast is the cleanest available demonstration that the 1964 pattern is a property of the subject matter and the period rather than a general feature of mid-century congressional voting.

Against the Fair Housing Act of 1968, the pattern is closer to 1964 but with a weaker coalition, a narrower margin and a cloture fight that came far nearer to failing. Comparing the two shows what happens when the same coalition is asked to legislate on a subject where the costs fall on constituents outside the South as well as inside it, and the answer is visible in the vote totals.

The general observation is that a roll call is a measure of a bill as much as of a chamber. Where the burden of a statute falls on an identifiable region, the vote will be regional. Where it falls on an economic sector or on a general theory of government, the vote will be partisan or ideological. Reading a vote pattern backward to a claim about the parties requires knowing which kind of bill produced it, and the 1964 act produced the first kind.

This vote in the context of the 88th Congress

Placing the roll calls in their session makes one further feature visible.

The 88th Congress convened in January 1963 and H.R. 7152 was introduced in June of that year. The bill was pending in the House Judiciary Committee when President Kennedy was assassinated in November 1963, and it moved to the Rules Committee in the weeks afterward. President Johnson made its passage his stated legislative priority in his first address to a joint session. The February House vote came less than three months after the assassination.

Whether that sequence changed votes is a question the roll calls cannot answer, and confident claims in both directions are common. What the record does show is that the House totals in February and July are nearly identical, which means that whatever moved between the two dates did not move recorded positions. It also shows that the bill’s core coalition, northern members of both parties, had voted for civil rights measures repeatedly across the preceding decade, so their February votes were not conversions.

The 88th Congress also passed the Economic Opportunity Act of 1964, and the following Congress passed the Voting Rights Act of 1965, Medicare, and the Elementary and Secondary Education Act. The 1964 civil rights vote sits at the beginning of a legislative period of unusual productivity, and the coalitional questions it raises recur in each of those measures with different regional and partisan structures. Readers tracing that arc across statutes will find the pattern developed in the era and thematic guides for each cluster in this series.

The later civil rights roll calls, and what they show about 1964

Following the same coalition forward through the next three decades of civil rights votes puts the 1964 tallies in a sequence rather than leaving them as an isolated event.

The Fair Housing Act of 1968 is the closest test, because it asked the same chamber to legislate on a subject where the costs fell on constituents outside the South as well as inside it. Cloture failed on the first attempts and succeeded only narrowly on a later one, and the final margins were thinner than in 1964 or 1965. The coalition that had held comfortably for public accommodations and voting held barely for housing, which is the clearest available evidence that the 1964 coalition was specific to the question rather than a general alignment.

The Equal Employment Opportunity Act of 1972, which gave the enforcement commission authority to sue and extended the employment title’s coverage, passed with substantial bipartisan support and without a comparable procedural fight. By that point the constitutional objections that had organized the 1964 opposition had been resolved against them by the Supreme Court, and the argument had moved to enforcement design rather than to federal authority.

The Civil Rights Act of 1990 and the Civil Rights Act of 1991 show the pattern in its later form. The 1990 bill passed both chambers, was vetoed on the argument that its burden-of-proof provisions would push employers toward hiring by the numbers, and the Senate override attempt fell one vote short of the two-thirds required. The renegotiated 1991 bill then passed by wide margins in both chambers and was signed. That sequence is analyzed in the amendments that rebuilt this statute, and the override mechanics are treated in the veto and the override explained.

The trajectory across those votes is worth stating carefully, because it is easy to overread. The regional structure that defined 1964 weakened as the southern delegations changed and as the questions moved from access to remedies. The partisan structure strengthened, in the sense that later civil rights votes divided more along party lines and less along regional ones. Whether that represents a transfer of the same underlying disagreement into a new container, or a genuinely different disagreement about remedies that happens to divide the parties, is contested among people who study it, and this article does not adjudicate it. What the roll calls establish is the shape of each vote, not the continuity between them.

Why a supermajority coalition mattered beyond the tally

One consequence of the 1964 margins is worth recording because it operated outside Congress.

A statute enacted by narrow, single-party majorities carries a different institutional standing from one enacted by supermajorities in both chambers with support from both parties, and the difference shows up in how the statute is treated afterward. The 1964 act was upheld unanimously in result on its first constitutional test within six months, it was amended rather than repealed in every subsequent Congress that revisited it, and no serious repeal effort has been mounted against its core titles. Those facts are not caused by the roll call in any direct sense. They are consistent with a measure whose enactment coalition was broad enough that unwinding it would have required assembling an equally broad coalition in the opposite direction.

The contrast within the act’s own history is instructive. The provisions that have proved most durable are the ones that passed in 1964 with the full coalition behind them. The provisions added later by narrower margins, including several of the enforcement changes made in 1972 and 1991, have been contested more persistently in litigation and in subsequent legislative proposals.

A reader should not convert that observation into a general rule that broad coalitions produce durable statutes, because the causation runs in several directions at once and the sample here is one law. It is offered as a specific observation about this statute: the size and composition of the enacting coalition is part of the explanation for why the act has been amended six times and repealed zero times.

What the roll calls show about the Senate as an institution

Beyond the partisan and regional cuts, these votes are among the best available evidence about how the Senate functioned in the period, and three institutional features stand out.

The first is the power of a committee chairmanship in a seniority system. James Eastland of Mississippi chaired the Judiciary Committee, which had jurisdiction over civil rights legislation and had buried such measures reliably for years. The bill reached the floor only because Mansfield used a procedural device to keep it off the committee’s calendar. Howard Smith of Virginia held equivalent power in the House through the Rules Committee, and the discharge petition was the mechanism that circumvented it. Both facts illustrate that the seniority system of the period concentrated gatekeeping authority in members from safe one-party districts and states, which in this era meant disproportionately in the South. The committee system’s gatekeeping function is treated in its own right in committees, markups and gatekeeping.

The second is that the filibuster in 1964 functioned as it was theoretically supposed to and rarely does: as sustained debate that actually occupied the floor, conducted by a determined minority, broken only when the majority assembled a supermajority. Fifty-seven working days of floor time is an enormous expenditure in a chamber with other business, and the opposition’s theory was that the cost would prove unbearable. The theory had been correct for decades.

The third is that a two-thirds requirement in a chamber where one party holds well under two-thirds makes bipartisan negotiation structurally mandatory rather than merely desirable. The Dirksen substitute is what that requirement produced, and reading it as a concession made from weakness misunderstands the arithmetic. It was the price of the only path available.

Why this coalition was difficult to reassemble

The 1964 coalition held again in 1965 and in 1968, and then the pattern that had produced it changed. Naming the changes precisely is more useful than characterizing anyone’s motives.

The southern delegations that supplied nearly all the opposition were, in 1964, uniformly of one party. Over the following decades those seats changed hands in a process that ran through presidential voting first, then Senate and House seats, then state legislatures, taking roughly a generation and completing at different times in different states. Whatever caused it, the effect on vote arithmetic was that the regional bloc and the party blocs stopped being cross-cutting and started to align, which removes the structural condition that made the 1964 coalition possible.

The second change was in the character of the legislation. The 1964 and 1965 acts addressed practices that were legally mandated and openly defended, which made the target unambiguous. Later civil rights measures addressed questions where the disagreement was about remedies, standards of proof and the allocation of burdens, which are questions on which the two parties diverged in a way they had not diverged over whether hotels must rent rooms. The veto of the 1990 civil rights bill and the negotiation that produced the 1991 act are the clearest illustration, and that sequence is traced in the amendments that rebuilt this statute.

The third is that Rule XXII changed in 1975, lowering the cloture threshold to three-fifths of senators duly chosen and sworn. That made cloture easier to obtain in arithmetic terms and, over time, made filibusters cheaper to mount, since the modern practice does not require holding the floor. A reader comparing 1964’s supermajority coalition to later ones is comparing across a rule change that altered what a coalition had to accomplish.

None of this is a claim that the 1964 coalition was superior to what followed, and none of it is a prediction. It is an account of why the specific arithmetic that produced 71 votes for cloture on June 10, 1964 depended on conditions that did not persist.

How the numbers looked to contemporaries

Reading the tallies as a person in 1964 would have read them changes their meaning in one respect worth noting.

The cloture vote was understood at the time as the significant event, and contemporary coverage treated June 10 rather than June 19 or July 2 as the day the bill was decided. That understanding was correct and has faded, which is why popular accounts often quote the final passage figure as though it were the achievement. The achievement was seventy-one.

The regional composition was also entirely unremarkable to contemporaries, who did not need a cross-tabulation to know that the opposition was southern. The party-level framing that dominates modern discussion of these votes is a later development, produced by the changed party coalitions that make the 1964 figures look surprising. A reader in 1964 encountering the claim that Republicans supported the bill at a higher rate than Democrats would have regarded it as an arithmetic consequence of the southern delegation rather than as a discovery.

That difference in reading is itself the strongest argument for printing both cuts. A statistic that was uninteresting when it was generated and became contested decades later has not changed. The context that made it obvious has.

How to read a roll call as a research document

Roll calls are among the most abused primary sources in American political writing, and the abuse is possible because they look self-explanatory. Six habits make them reliable.

Identify which vote you are looking at. A bill of this size generates a passage vote in each chamber, procedural votes, amendment votes, and a vote on agreeing to the other chamber’s text. They produce different numbers and support different claims. A source citing the Civil Rights Act vote without saying which of the four it means has already lost precision.

Read the question presented. A vote on cloture is a vote to end debate, not a vote on the merits, and members have voted for cloture on bills they then voted against and against cloture on bills they supported. In 1964 the two Senate votes tracked each other closely, with only a two-vote difference in the totals, so the distinction does not change the analysis here. That is a fact about this bill rather than a general rule.

Check the denominator. Members voting, members present, and members of the chamber are three different denominators producing three different percentages from the same tally.

Cross-tabulate before comparing. Any comparison between two groups that differ systematically in a third characteristic is measuring that third characteristic. In 1964 the parties differed systematically in regional composition, which is why the party comparison is unreliable and the within-region comparison is not.

Distinguish the recorded vote from the surrounding record. Pairs, announced positions and explanatory floor statements sit in the Congressional Record and not in the tally, and merging them without saying so produces figures that cannot be reconciled with the official count.

Resist the inference from vote to belief. A roll call records what a member did on a date under a set of institutional and constituency pressures. It is strong evidence of a position and weak evidence of a private conviction, and the difference matters most in exactly the cases where writers most want to assert the second.

Applied to the four votes in this article, those habits produce a modest and defensible set of claims: the bill had majority support throughout in both chambers, the opposition was regionally concentrated and crossed party lines within the region, the decisive contest was a procedural threshold in the Senate, and the party gap in the aggregate figures is largely a composition effect. Everything beyond that requires evidence a roll call does not contain.

The misstatements that recur

Six errors account for most of the inaccurate claims made about these roll calls, and each is correctable from the table above.

Quoting one chamber’s numbers as the whole story. Four roll calls exist and a citation that names none of them cannot be checked.

Reporting party percentages without the regional composition that produces them. This is the single most common error, and it is committed in both directions.

Treating the aggregate party gap as a measure of party positions. Outside the region the gap runs the other way, and inside the region there is no meaningful Republican delegation to compare.

Misstating the cloture threshold. It was two-thirds of senators present and voting, which meant sixty-seven with all one hundred present in 1964, not sixty, which is the modern three-fifths figure adopted in 1975.

Describing southern opposition as unanimous. A small group of House Democrats from the region voted yes on both House roll calls, and one senator from the region voted yes on both Senate roll calls.

Using the votes to argue about present-day parties. The roll calls are conclusive about 1964 and silent about subsequent coalitions, and the argument that they settle later questions is the one thing all four tallies cannot do.

Why these roll calls are searched more than any others

A short observation about why this article exists in the form it does.

Very few congressional votes are the subject of sustained public argument decades later. This one is, for three reasons that are worth naming. The subject matter remains contested in ways that other mid-century legislation does not. The party columns produce a headline figure that is both accurate and counterintuitive against modern party coalitions, which makes it attractive to quote. And the underlying data are public, so anyone can check any claim in minutes, which means the argument recurs rather than resolving.

The third reason is the one that makes a reference article useful. Where evidence is contested but unavailable, an article can only report the dispute. Where evidence is contested and fully available, an article can end the dispute for any reader willing to look at it. Four roll calls, two cuts each, all counts sourced to the recorded votes. That is the entire contribution, and it is enough.

The limits of what a roll call can establish

A closing statement of limits belongs in an article built entirely on one kind of evidence.

A roll call establishes that a named member cast a recorded vote on a stated question on a given date. It is conclusive on that and on the aggregates computed from it.

It is strong evidence of a member’s public position, since a recorded vote is the most consequential public statement a legislator makes.

It is weak evidence of private conviction, because members vote under constituency pressure, party pressure, procedural constraints and negotiated commitments, and the vote records the resultant rather than any of the components.

It is not evidence at all about causation, whether the question is what caused a member to vote as they did, what the vote caused in a subsequent election, or what it caused in the following decades of party coalitions. Those questions require different evidence and are answered, when they are answered at all, with more uncertainty than the roll calls carry.

Holding those four levels distinct is what allows an article on the most politically contested vote of the twentieth century to be useful to readers who disagree with each other. The numbers are the numbers, they are available to anyone, and the analysis above adds only the regional cut and the arithmetic that follows from it.

A note on the figures in this article

Every count in this article comes from the recorded roll calls of the 88th Congress: the House votes of February 10 and July 2, 1964, and the Senate votes of June 10 and June 19, 1964. The aggregate totals and the party splits are the figures carried in the official records of both chambers and in the standard congressional reference compilations of the period.

The regional splits are computed by assigning each voting member to the eleven states of the former Confederacy or to the rest of the country and re-summing the same roll calls, which is the standard convention described above. A reader who prefers a border-state definition can recompute from the same records and will find the direction of every result unchanged.

Where a figure could not be stated with confidence, this article describes the fact qualitatively rather than supplying a number, which is why the small groups of southern members voting yes in the House are characterized by state rather than named individually, and why the three non-southern Democratic senators voting against cloture are counted rather than listed. Precision about what is known and what is not is worth more in an article of this kind than completeness, because a single wrong name in a piece built on roll-call accuracy would undermine everything else in it.

What the roll calls support

The four votes support a small number of claims firmly, and stating them plainly is the point of the exercise.

The bill passed both chambers with substantial bipartisan majorities on every recorded vote. A higher proportion of Republicans than Democrats voted for it in both chambers on every roll call. Opposition was concentrated in the eleven states of the former Confederacy, where members voted against the bill at rates above ninety percent regardless of party. Outside that region, both parties supported the bill overwhelmingly, with northern Democrats supporting it at a marginally higher rate than northern Republicans. The Senate could not have reached the cloture threshold without Republican votes, and those votes were secured through a substitute text that narrowed the enforcement provisions. And the decisive vote was procedural, taken on June 10, 1964, and it was the first successful cloture vote on a civil rights measure in the Senate’s history.

Everything else commonly asserted about these numbers requires evidence the roll calls do not contain. That is not a limitation of the record; it is the difference between a vote count and a historical argument, and a reader who keeps the two apart will be right more often than one who does not.

For orientation across the whole statute, the complete guide to the act is the hub for this cluster, and readers who want to keep the four tallies, the regional cuts and the named votes in one place while working through the claims built on them can keep their statute notes, vote tables and citations together free on VaultBook.

What a reader should carry from this

Four things, and they fit in a paragraph each.

The numbers are settled and public. Four roll calls, two chambers, five months, all counts available in the official records. Nothing in this article rests on a disputed figure, and any reader can check every claim in it.

The regional cut explains the partisan cut, and not the reverse. Outside the eleven states both parties supported the bill overwhelmingly, with northern Democrats slightly ahead. Inside them, opposition was nearly unanimous across party lines. The national party gap is the arithmetic shadow of that distribution.

The decisive vote was procedural. Cloture on June 10, 1964, at 71 to 29 against a threshold of 67, was the first time the Senate had ever ended debate on a civil rights measure, and it was reached by trading statutory text for Republican votes.

And the roll calls answer the questions they answer. They establish who voted how in 1964 with complete authority, and they establish nothing about later party coalitions without a separate argument built on separate evidence. Keeping those two things apart is the whole discipline this article recommends, and it is what allows a set of numbers that both sides of a public argument want to own to be reported in a way that neither side can fairly dispute.

Frequently Asked Questions

Q: How did Democrats and Republicans vote on the Civil Rights Act of 1964?

Both parties delivered majorities on every roll call. On House passage on February 10, 1964, Democrats voted 152 to 96 in favor and Republicans 138 to 34. On Senate cloture on June 10, Democrats voted 44 to 23 and Republicans 27 to 6. On Senate passage on June 19, Democrats voted 46 to 21 and Republicans 27 to 6. On the final House vote on July 2, Democrats voted 153 to 91 and Republicans 136 to 35. The proportional gap favoring Republicans reflects composition rather than party position: nearly every member from the eleven southern states was a Democrat, and that bloc opposed the bill almost unanimously, which pulls the Democratic percentage down.

Q: Did more Republicans than Democrats support the Civil Rights Act of 1964?

A higher proportion of Republicans voted yes in both chambers on every roll call, and a larger absolute number of Democrats voted yes in the House because the Democratic caucus was larger. Both statements are accurate and neither is the interesting one. The regional breakdown explains the proportional gap: outside the eleven southern states, northern Democrats supported the bill at roughly 94 percent and northern Republicans at roughly 85 percent on House passage, so the gap outside the South actually runs the other way. The national Democratic percentage is a weighted average that includes the southern bloc, while the Republican percentage includes almost none of it, since only ten southern House members and one southern senator were Republicans.

Q: What was the final Senate tally on the Civil Rights Act of 1964?

The Senate passed H.R. 7152 on June 19, 1964, by 73 to 27, nine days after invoking cloture. Democrats split 46 in favor and 21 against, Republicans 27 in favor and 6 against. Of the twenty-two senators from the eleven states of the former Confederacy, exactly one voted yes: Ralph Yarborough of Texas. Senators from the other thirty-nine states voted 72 to 6. The final House action came on July 2, when the House agreed to the Senate’s amended text 289 to 126, and the President signed the bill that day. The passage tally is often quoted alone, though the cloture vote nine days earlier was the decisive one.

Q: Which senators voted against the Civil Rights Act of 1964?

Twenty-seven senators voted against passage on June 19, 1964. Twenty-one were from the eleven southern states: twenty Democrats, including Richard Russell of Georgia, who organized the opposition, and Sam Ervin of North Carolina, who made the leading constitutional argument, along with the region’s single Republican, John Tower of Texas. J. William Fulbright of Arkansas and Albert Gore of Tennessee were among the southern Democratic opponents. Outside the South, one Democrat voted no, Robert Byrd of West Virginia, and five Republicans did: Barry Goldwater of Arizona, Bourke Hickenlooper of Iowa, Edwin Mechem of New Mexico, Milward Simpson of Wyoming, and Norris Cotton of New Hampshire.

Q: Why did Barry Goldwater vote against the Civil Rights Act of 1964?

He stated a constitutional objection to two of the bill’s titles rather than an objection to its aims. Goldwater had voted for the Civil Rights Acts of 1957 and 1960 and had supported civil rights measures in Arizona, and he said he agreed with most of the 1964 bill. His stated position was that the public accommodations title and the employment title exceeded federal authority over private conduct and would require an intrusive enforcement apparatus, and he said he had sought legal advice on the question before deciding. The Supreme Court rejected that constitutional reading within six months, upholding the public accommodations title under the Commerce Clause in December 1964.

Q: How did southern Democrats vote on the Civil Rights Act of 1964?

Almost uniformly against. On House passage, southern Democrats voted 7 in favor and 87 against, with the small group of supporters drawn overwhelmingly from urban and economically diversified districts in the Texas, Tennessee and Florida delegations. In the Senate, southern Democrats voted 1 in favor and 20 against on passage, the single supporter being Ralph Yarborough of Texas. This bloc supplied the overwhelming majority of the total opposition in both chambers, and its members had coordinated against civil rights legislation for years under Richard Russell of Georgia. The same regional pattern held on the cloture motion and on the final House vote in July.

Q: How many House members voted for the Civil Rights Act of 1964?

Two hundred ninety House members voted for the bill on February 10, 1964, against 130 opposed. On July 2, when the House voted to accept the Senate’s amended text, 289 voted yes and 126 voted no. The two votes were on different documents: the February vote was on the House bill, and the July vote was on the Senate substitute that narrowed several enforcement provisions and became the enacted law. A reader comparing the statute’s provisions to the February debate is comparing to a superseded version, which is a common source of confusion in accounts that report only one House tally.

Q: Does the Civil Rights Act of 1964 roll call show the parties switched sides?

The roll call shows that in 1964 the division ran through both parties along regional lines rather than between them, with large majorities in both caucuses outside the South and near-unanimous opposition within it. It does not by itself establish anything about later party coalitions. The southern delegations that supplied nearly all the opposition did change party affiliation over the following decades, through a process that took roughly a generation and proceeded at different rates by state and by office. That the change occurred is not disputed; its causes and timing are the subject of genuine scholarly disagreement, and a tally from 1964 is not evidence for any particular account of it.

Q: What was the cloture vote on the Civil Rights Act of 1964?

The Senate invoked cloture on June 10, 1964, by 71 to 29, ending a filibuster that had occupied the chamber for fifty-seven working days. Under Rule XXII as it then stood, cloture required two-thirds of senators present and voting, which with all one hundred voting meant 67, so the motion carried with four votes to spare. Democrats voted 44 to 23 and Republicans 27 to 6. It was the first time the Senate had ever invoked cloture on a civil rights measure. Because a simple majority for the bill existed throughout, this procedural vote, not the passage vote nine days later, is the one that determined the outcome.

Q: How many senators from the South voted for the bill?

One. Ralph Yarborough of Texas was the only senator from the eleven states of the former Confederacy to vote for cloture and for passage. The region’s other twenty-one senators voted against both, comprising twenty Democrats and one Republican, John Tower of Texas. Yarborough faced sustained political consequences in his state for the vote. In the House the southern group was slightly larger but still very small: seven of the ninety-four southern members voted yes on February 10, all of them Democrats, with the ten southern Republicans voting no. That is the concentration that makes region a far better predictor of these votes than party.

Q: Why was there no conference committee on the Civil Rights Act of 1964?

Because a conference would have required the Senate to vote again, exposing the bill to a second filibuster after the first had been broken at maximum cost. The Senate had substantially amended the House bill through the Dirksen substitute, and the ordinary route would have been a conference to reconcile the versions, followed by votes on the conference report in both chambers. The managers avoided that by having the House simply agree to the Senate text on July 2. The consequence is that the Senate’s narrower enforcement provisions became law without further negotiation, and House members who preferred their own version faced a choice between accepting them and losing the bill.

Q: How did northern Republicans vote compared with northern Democrats?

Both groups supported the bill heavily, with Democrats slightly ahead. On House passage on February 10, 1964, Democrats from outside the eleven southern states voted 145 to 9, roughly 94 percent in favor, and Republicans from those states voted 138 to 24, roughly 85 percent in favor. In the Senate, non-southern Democrats voted 45 to 1 on passage and non-southern Republicans 27 to 5. The gap outside the South therefore runs opposite to the gap in the national totals, which is the clearest single demonstration that the national party percentages are driven by where each party’s southern members sat rather than by the parties’ positions on the bill.

Q: Which Republicans voted against the bill in the Senate?

Six: Barry Goldwater of Arizona, John Tower of Texas, Bourke Hickenlooper of Iowa, Edwin Mechem of New Mexico, Milward Simpson of Wyoming, and Norris Cotton of New Hampshire. Tower was the only one from a southern state. The other five stated objections centered on federal authority over private property and business decisions, with Hickenlooper also raising the treatment of jury trials in contempt proceedings, an issue that had been contested during the 1957 act as well. None of the five was part of the organized floor opposition; they voted against a bill they had not fought, which distinguishes their position from that of the southern caucus.

Q: Why did the House vote on the bill twice?

Because the Senate changed the text. The House passed its own version of H.R. 7152 on February 10, 1964. The Senate then adopted the Dirksen substitute, which narrowed several enforcement provisions in order to secure the Republican votes needed to reach the cloture threshold, and passed the amended bill on June 19. Under ordinary procedure the two versions would have gone to a conference committee, but the managers avoided that to prevent a second Senate filibuster. The House therefore voted again on July 2 to agree to the Senate’s text as it stood, 289 to 126. That second vote is the one on the language that became law.

Q: How did the 1964 vote compare with the Voting Rights Act vote in 1965?

The 1965 margins were wider. The Senate passed the voting rights bill 77 to 19 and the House passed it 333 to 85, with cloture again invoked and the threshold met more comfortably. Three factors explain most of the difference. The 1964 act regulated private conduct by businesses and employers, which drew constitutional objections that a bill regulating state election administration under the Fifteenth Amendment did not. The events at Selma in March 1965 changed the political environment. And the 1964 act had already been implemented without the disorder its opponents predicted, removing one of the strongest arguments available a year earlier.

Q: Who managed the Civil Rights Act of 1964 on the Senate floor?

Hubert Humphrey of Minnesota, then the Democratic whip, was the principal floor manager, paired with Thomas Kuchel of California as the Republican floor manager. The pairing was deliberate, because a bill requiring two-thirds in a chamber where one party held well under that share could not be managed by a single caucus. Mike Mansfield, the majority leader, made the procedural decision to keep the bill off the Judiciary Committee calendar, and Everett Dirksen, the Republican leader, delivered the Republican votes through the substitute text that carries his name. Richard Russell of Georgia directed the opposition as a coordinated procedural campaign.