Only tenuously related, I suppose, but... Consider a race between N candidate c_N down to c_1, in which Each candidate c_i receives a total of i votes. Each candidate c_i receives his first vote before c_(i-1) and thereafter remains >= 1 vote ahead of c_(i-1). For N = 4, there are 12 voting sequences meeting these criteria: 4444333221 4444332321 4443433221 4443432321 4443343221 4443342321 4443324321 4434433221 4434432321 4434343221 4434342321 4434324321 Now suppose the race follows these rules Each candidate c_i receives a total of i votes. Each candidate c_i receives his first vote before c_(i-1) and thereafter remains 0 or 1 vote ahead of c_(i-1). For N = 4, there are again 12 voting sequences meeting these criteria: 4342341234 4342314234 4342312434 4342134234 4342132434 4324341234 4324314234 4324312434 4324134234 4324132434 4321434234 4321432434 It turns out that for both types of races, the number of voting sequences for N candidates is A003121(N). Colin Mallows termed the first type of voting sequence a "strict ballot" sequence, because each candidate stays strictly ahead of the next. He called the latter type "interleaved ballot" sequences because the votes of any two adjacent candidates are interleaved. I might call the latter type "close ballot" sequences, since each candidate stays close in votes to the next candidate. Dr. Mallows long ago submitted A003121 to count the first type of voting sequences, and later submitted a second, different sequence counting the second type of voting sequences. Yet later I stumbled on the latter sequence and in trying to extend it, found that some of the submitted values were incorrect, and when corrected seemed to agree with A003121. I related this to Dr. Mallows, who had the latter sequence removed from the OEIS, and soon sent me a very clever proof that both types voting sequences have the same counts (which I remember but have never published, being put off by LaTeX).