What to make of the results of the first two of this spring's special House elections? Start off by putting them in perspective. They pose a challenge to both political parties, but especially to Republicans, which have been used to an unusually stable partisan alignment, an alignment that has become scrambled by Donald Trump.

Those of us who can remember the 1964-84 years have seen much greater partisan churning. Almost half the congressional districts that voted for Richard Nixon in 1972 elected Democratic congressmen. Some 191 districts split tickets. In 2012 that number was down to 26, the lowest since 1920.

The number rose in 2016 to 35, with another dozen or so on the cusp. That reflects Trump's distinctive appeal. Exit polling reported he increased the Republican margin among non-college-graduate whites from 25 points to 39 while reducing it among white college graduates from 16 points to 4. Other demographic groups voted pretty much the same.

Which leads us to the special elections. The first, on April 11, was in Kansas 4, a district heavily white non-college, with two-thirds of its voters in Sedgwick County (Wichita) and the remainder in rural counties. Republican Ron Estes won by 53 to 46 percent—well below Donald Trump's 59 to 32 percent district margin.

Democrat James Thompson carried Sedgwick County, apparently because of switches by college-graduate voters. But Estes carried a solid 62 percent in the rural counties, well ahead of the 2014 percentages there for Republican Gov. Sam Brownback and Sen. Pat Roberts.

Given the dynamics of special elections—you can cast a protest vote, and for a locally attuned candidate, without turning the whole government over to the opposition—this looks something like a traditional, pre-Trump margin in what has been a safe Republican seat for 20 years.

The turnout was heavier and the race more contested Tuesday in Georgia 6 in the northern Atlanta suburbs, a district with one of the highest college graduate percentages in the nation. Mitt Romney carried it by 23 points, Trump by 1.5. Despite its Republican leanings, it has heavily Democratic black, Hispanic and Jewish blocs.

National Democrats rallied to 30-year-old filmmaker and former House staffer Jon Ossoff, who raised a phenomenal $8 million. When the first early voting returns came in, Ossoff had 71 percent of the vote, while Republicans were split among 11 candidates. But as election day returns poured in, that was reduced to 48 percent. Ossoff faces a June 20 runoff against Republican Karen Handel, former secretary of state and Fulton County Commissioner.

Altogether 51 percent of voters chose Republicans, 49 percent Democrats. Ossoff got 1.3 points more than Clinton. The 11 Republicans got 1.4 points more than Trump. Obviously, either candidate could win in June.

There's a clear contrast with Kansas's 4th district, whose results suggest that traditional Republican margins in other non-college, non-metropolitan areas are greatly threatened. Georgia 6 suggests that in places heavy with college graduates, the 2016 Trump numbers are the new norm—at least in races without incumbents who have established themselves in sync with the district.

A glance at the list of the 23 Republican-held districts carried by Hillary Clinton shows that a half dozen are heavily Hispanic, with well-known incumbents. But most are heavily affluent and college-educated. Five such districts in southern California and one in northern Virginia have increasing immigrant populations; three in Texas, like Georgia 6, have affluent traditionally Republican voters repelled enough by Trump to vote for Clinton.

There would be many more such heavily college-graduate districts vulnerable to a Democratic takeover, but for the fact that Democrats have long since taken them over, starting in the 1990s.

The good news for pro-Trump Republicans is that most of his November 2016 voters have stuck with him. His current 42 percent job approval is only 4 points below the percentage of the national vote he won five months ago.

The bad news for pro-Trump Republicans is that there is zero evidence that he is making inroads among the slightly larger number of those who voted against him. Georgia 6 suggests that the highly educated among them are heavily motivated to get out and vote Democratic. Republican incumbents who considered their districts safe may not have worked them hard enough to survive a spirited challenge.

Trump threaded the needle by winning over enough non-college voters to win 100 Obama electoral votes. Republicans may need to thread a different needle to hold the House.