The Senate Cast One Bipartisan Vote and Two Party-Line Votes in Forty-Eight Hours. The Data Says Where Crossing the Aisle Survives.

Two votes, one subject, forty-eight hours

Volodymyr Zelensky sat in the front row of the Senate gallery on Tuesday, July 28, 2026, and watched 86 senators vote to take up a sanctions bill named for a man who had died sixteen days earlier (Kurdistan24, July 28, 2026; CBS News, July 14, 2026). The bill was Lindsey Graham's, and the vote to advance it, 86 to 12, was one of the most lopsided the Senate has produced this year on a contested piece of foreign policy (Axios, July 29, 2026).

Two days later the same chamber looked nothing like that. On Thursday it voted 49 to 50 against a resolution to limit the president's war in Iran, and 47 to 52 against overturning a federal child care rule (Roll Call, July 30, 2026; U.S. Senate floor proceedings, July 30, 2026). One week, three roll calls, and a 39 vote gap between the winning side of the first and the losing side of the last. The question worth asking is not why the Senate is divided. It is why, on one subject and not the others, it was not.

The number that made the sanctions vote unusual

Start with what 86 votes actually is. In a chamber where most contested measures now finish within a few votes of a party line, 86 to 12 is not a normal count. It is the kind of number the Senate reserves for a narrow category of questions, and history says which one. When the Senate voted to admit Finland and Sweden to NATO in 2022, the tally was 95 to 1 (U.S. Senate, Treaty Document 117-3, 2022). A foreign adversary is close to the last subject that still moves senators of both parties into the same column in large numbers.

The sanctions vote fit that mold. The twelve who dissented were not defending Russia. They were ten Democrats, independent Bernie Sanders, and Republican Rand Paul, and their stated objection was to the bill's authority to impose tariffs of up to 100 percent on the biggest buyers of Russian oil, not to the sanctions themselves (The Hill, July 28, 2026; TechTimes, July 29, 2026). Even the opposition agreed on the target and split only on the tool.

Where crossing over lives, measured

The Iran vote is where the chamber's ordinary shape returned, and it is worth reading name by name. Three Republicans crossed to support the resolution: Susan Collins, Lisa Murkowski, and Rand Paul. One Democrat, John Fetterman, crossed the other way (Roll Call, July 30, 2026). Two of those names are not accidents. Collins and Murkowski are, by the numbers, the two most bipartisan members of the Senate.

US Political Rank's ranking of the most and least bipartisan senators is built on a plain statistic from GovTrack: the share of a member's cosponsorships that go to bills introduced by the other party, across the full 118th Congress. Collins leads the chamber at 66.7 percent. Murkowski is second at 61.2 percent. The chamber median is 26.4 percent (GovTrack, 2024 Report Cards). Both women represent states their party cannot take for granted, and both cross the aisle at more than double the typical rate. When a war powers resolution drew exactly the Republicans who cross over most, the roster was doing what the data predicts.

The ranking carries a second finding that cuts against the easy story. The widest ideological gap in the chamber is not between the parties but inside one of them. On the Lugar Center's Bipartisan Index, Texas Republican John Cornyn ranked fifth in the Senate while his same state colleague Ted Cruz ranked 89th, an 84 place split between two men who share a party and a state (Lugar Center and McCourt School, 2024). Neither party owns crossing over, and neither owns the refusal of it.

The senator who fits no count

Rand Paul is the clearest illustration that a party label is a poor predictor. He was one of the twelve who voted against advancing the Russia sanctions bill on Tuesday, and one of the three Republicans who voted to constrain the president on Iran on Thursday (TechTimes, July 29, 2026; Roll Call, July 30, 2026). On both days he broke from most of his party, in opposite procedural directions, for a single reason he has held for years: skepticism of concentrated executive power, whether the instrument is a tariff or a war. A ranking of bipartisanship counts cosponsorships, not motives, and Paul is a reminder that the same vote can come from very different places.

The honest limit

The data has edges, and they should be named. A cloture vote to take up a bill is not final passage; the 86 who agreed to debate the sanctions may yet divide over amendments before the measure clears (Axios, July 29, 2026). Cosponsorship, the basis of the bipartisanship ranking, is a cheaper act than a hard floor vote, and a high score can reflect symbolic joining as much as real compromise, a limit the ranking states plainly in its own methodology. And one week of roll calls is a small sample. Three votes do not make a trend.

What they make is a snapshot, and the snapshot is consistent with a decade of the underlying numbers rather than a break from them. The same names sit atop the Lugar Center's Bipartisan Index year after year, and Collins has finished first nine times in eleven years (Lugar Center and McCourt School, 2024). Stability like that is not noise. It is behavior.

What to watch

So here is what to watch as the Senate moves from taking up the Graham bill to actually passing it. If the final vote holds near 86, the pattern is confirmed: a foreign adversary is still the one subject that can assemble a supermajority. If it erodes toward a party line as the tariff fight sharpens, that tells you the consensus was on Russia and never on the tool. Either way, keep your eye on the handful of senators who show up in the crossover column again and again. Collins and Murkowski have led the bipartisanship ranking for years, and they were exactly where the data said they would be this week (GovTrack, 2024 Report Cards). In a Senate that agrees on less each year, the members who still cross over are not a mystery. They are a measurable, and mostly predictable, few.