[twright0]

An interesting anecdote, another good example of a reasonably modern example of paper ballots enabling election stealing: https://en.wikipedia.org/wiki/Box_13_scandal

Caro covers this pretty extensively in his LBJ biography series, but it's reasonably clear from the evidence that LBJ won his senate seat by some pretty crude paper voting record manipulation after the fact - changing a '7' to a '9' by writing over the number with a pen - almost certainly with LBJ's knowledge. Given that his senate seat eventually put him in the presidency, it's probably the most consequential voter fraud ever committed in American history (that we know about, I suppose).

[rmunn]

From the second paragraph of the Wikipedia article: "Six days after polls had closed, 202 additional votes were added to the totals for Precinct 13 of Jim Wells County, 200 for Johnson and two for Stevenson."

Those numbers alone should make anyone suspicious. If you have an urn containing about 20,000 balls in two colors, red and green (this election happened in 1948 and the 1950 census listed that county's population as 27,991; let's assume that roughly 20,000 people would have been old enough to vote in 1948) and you randomly draw out 202 balls (about 1% of the total number in the urn), you would expect the number of balls you draw out to be roughly proportional to the red-blue mix in the urn. (1% of the total is big enough to expect a roughly-unbiased sample). So if you draw out 99% red balls and 1% green balls, then either you have a very very skewed proportion of colors in the urn, or else someone is cheating. Given the TINY margin of victory in that race (87 votes out of nearly a million, 988,295 to be precise), it's very very unlikely that precinct 13 happened to be skewed 99% towards LBJ when the state as a whole was so closely balanced.

[twright0]

I really encourage interested folks to read the biography (though it's an undertaking).

According to Caro, part of the background is that the relevant southern Texas precincts were well understood to have vote counts up for purchase; over the course of election counting, both sides would have their controlled districts release counts based on what the other side was reporting to stay in the race. These counts would vary in legitimacy and how skewed they were based on the precinct and need of the candidate that had swayed the boss to their side. But tactics like having armed guards supervise the casting of votes to ensure the favored candidate got a large majority, or simply distributing vote receipts to people who never voted at all and recording votes on their behalf, or making numbers up entirely, were quite common. Typically, though, Caro argues that because both sides did this, and they did it incrementally, it usually wasn't enough to sway an election one way or another, but rather was just part of the cost of doing business. He even says that LBJ lost his Senate election earlier that decade because he got cocky and told the bosses of the districts he had bought to just release all their numbers right away, letting his opposition then juice their numbers just enough to win.

It's really the timing, more than the margin, that makes it clear what happened (and the crudeness of the forgery); after every other precinct reported and finalized, they corrected their number by barely more than needed to win. The 100 to 1 vote margin was actually not that far off from the vote margin that the precinct reported in the first place (... which, of course, really tells you that the whole thing was made up from whole cloth).