FOR THE past 40 years, the story of Box 13 in tiny Alice, Texas, has stood as one of the great mysteries and richest tales in American politics. It is the story of how Lyndon Johnson stole the 1948 election and then how he sealed his theft with a brilliant manipulation of the judicial system and the state Democratic political organization.

This ground has been plowed before, but never as thoroughly as by Robert A. Caro in the second volume in his massive biography of Johnson. Caro has found no dramatic, new smoking gun in his investigation of Johnson's 87-vote victory, but he has added some important facts to the historical record. More impressively, he has masterfully spliced together events that were playing out in courtrooms, hotel ballrooms and political backrooms to create a gripping narrative that has the intensity and drama of an international thriller.

Unfortunately that isn't enough for Caro. He concludes that Johnson didn't just steal the election but pulled off a heist of historic proportions, far greater than anyone has recognized until now. Here he's stretching things. Caro's distaste for Johnson, his infatuation with Johnson's opponent and his personal revulsion for the political bossism and corruption in south Texas color his interpretation of the facts and ultimately distort his rendering of history.

First the story. With his political life on the line, Johnson runs 70,000 votes behind Coke Stevenson, a popular former Texas governor, in the 1948 Democratic primary, but forces Stevenson into a runoff. On the morning after the runoff, the Texas election bureau shows Stevenson leading Johnson by 854 votes.

There follow several days of recounts, corrections and phony vote-shifting, but by mid-day on the Friday after the runoff, Stevenson still leads by more than 150 votes. Then miraculously comes the report that in Box 13 in Alice, in the heart of the machine of boss George Parr, an additional 200 Johnson votes have been discovered. He is the winner by 87 votes out of nearly 1 million cast.

Stevenson, understanding the ways of the Rio Grande Valley in south Texas, goes to investigate, taking with him Frank Hamer, a legendary Texas Ranger who led the posse that captured and killed Bonnie and Clyde. In Alice they discover signs of fraud: The last 202 names on the rolls in Box 13 were written in a different color ink; the new names were listed in alphabetical order; the handwriting was identical; some of the new voters claim they never voted.

But Johnson, having lost the 1941 Senate race by being outstolen, is determined to preserve his victory. His lawyers find a friendly judge in Austin to issue an injunction preventing the Jim Wells County (Alice) Democratic organization from scratching out those 202 new votes and putting Stevenson back on top. The legal ploy designed as a means of delay gives Johnson the time he needs to win certification as the winner by the state Democratic executive committee (on a one-vote margin).

Stevenson counters by finding a friendly federal judge, who issues an injunction preventing Johnson's name from going on the ballot and orders an investigation into vote fraud in south Texas.

Time -- and the explosive investigations -- become Johnson's new obstacles. The candidate calls in his old friend Abe Fortas, who sizes up the situation and lays out a bold and risky legal strategy, designed to move the case through the appeals court and quickly to the Supreme Court.

As Johnson's lawyers scramble through their appeals, investigators in south Texas are closing in on the truth of what happened. Finally Supreme Court Justice Hugo Black rules that the federal government has no right to interfere in a state election, stopping the vote-fraud investigation dead and assuring Johnson's election.

Like others who have dug into the record, Caro is unable to answer a central question: Did Johnson himself go to south Texas and cut a deal with George Parr that resulted in the extra votes in Box 13? He concludes it is possible but not provable.

Failing that, he sets out to build his case that Johnson won the '48 race with "tens of thousands" of bought or stolen votes. Johnson, he writes, "received a 35,000-vote plurality from areas {south Texas} in which the voting pattern was dramatically out of keeping with the normal patterns in a democracy."

But not out of keeping with the normal patterns of south Texas, where bloc voting and boss-control were endemic and well known.

The flaw in Caro's biography is his insistence on interpreting Johnson's actions in the worst possible light and Johnson's opponents in the best possible light. When Johnson goes judge-shopping, it's devious; when Stevenson does it, it's legitimate. When Johnson gets bloc votes, he's bought or stolen them; when Stevenson get them, it's because the bosses have no choice.

Caro makes much of Johnson's margins in south Texas (in Duval County, his margin was 4,622-40), but he either overlooks or discounts the similar voting patterns in earlier elections, including ones involving Stevenson. In 1942, Stevenson carried Duval with 2,836 votes. His five opponents got 77. In another celebrated contest, one candidate won the primary in a south Texas county by 3,000-5, then got into an argument with the county boss and lost the runoff by the exact same margin.

Coke Stevenson, who as governor never challenged this corruption in south Texas, must have factored these realities into his own plan for winning in 1948.

Caro's failure to come to terms with political realities of south Texas four decades ago doesn't absolve Johnson of his theft. But an otherwise stunning job of research, reporting and storytelling is flawed by the author's willingness to exaggerate the sins of his subject. Dan Balz, a staff reporter for the National section of The Washington Post, is the paper's former Southwest correspondent based in Texas.